Professional Documents
Culture Documents
, C. Wisniewski, A. Grasmick
Laboratoire de Gnie Des Procds de Montpellier II/GPSA, CC 024Universit Montpellier II,
Place Eugne Bataillon, 34095 Montpellier Cedex 05, France
Received 24 September 2001; received in revised form 25 March 2002; accepted 26 March 2002
Abstract
The aim of this study was to quantify the specic effect of adsorption on membrane fouling during ltration of a mem-
brane bioreactor (MBR) mixed liquor suspension. Adsorption experiments were performed on well-dened protein solutions
(-lactoglobulin solutions) to provide reference results and compare them to those obtained during the ltration of MBR
suspensions (raw suspension and settled suspension). Two different methods were used to quantify the role of adsorption in
membrane fouling: a static methodinwhichmembranes were immersedinthe biological suspensionanda dynamic method
supposing that the resistance due to adsorption is an irreversible phenomenon that remains after ltration and back-washing. It
was shown for the two types of suspensions that (i) due to limited diffusion, the dynamic method appears to be more adapted
than the static method; (ii) adsorption is a rapid fouling phenomenon that induces irreversible resistance and that, in frontal
mode takes place at the beginning of the operation; (iii) the adsorption phenomenon shows specic hydraulic resistance of
the same order of magnitude as the clean membrane resistance; (iv) other phenomena, i.e. progressive pore clogging, can also
take place though subcritical hydrodynamic conditions.
2002 Elsevier Science B.V. All rights reserved.
Keywords: Adsorption; Membrane fouling; Protein solution; Bacterial suspension; Membrane bioreactor
1. Introduction
Subcritical ltration conditions can be dened
to avoid macroscopic deposits from building up on
the membrane surface of the membrane bioreactors
(MBR) used to treat wastewater [1]. These conditions
allow long ltration periods without having to use
chemical cleaning procedures. However, the fouling
of the membrane is not avoided and generally a rapid
membrane permeability decrease occurs at the very be-
ginning of the ltration operation. Then, the decrease
ad
adsorbed membrane porosity
clean membrane porosity
dynamic viscosity (Pa s)
membrane area (m
2
)
This is partly because it is difcult to characterise
the soluble compounds liable to adsorb on the mem-
brane material present in complex suspensions such
as activated sludges. Conversely, many studies have
been carried out on synthetic protein solutions. These
solutions cannot constitute a model to represent the
soluble fraction of a bacterial suspension which is
composed of a wide variety of microbial polymers of
different nature (polysaccharides, proteins, etc.) and
of various molecular weights. However, a compar-
ative study between a well-dened protein solution
and a bacterial suspension could allow to estimate the
adsorption capacity of the soluble macromolecules
of the bacterial suspension in comparison with the
adsorption capacity of a well-dened protein.
Thus, the aim of this study was to quantify the spe-
cic effect of adsorption on membrane fouling in an
MBR suspension. Adsorption experiments were per-
formed on well-dened protein solutions to provide
reference results that were then compared with those
obtained during the ltration of an MBR suspension
(raw suspension and settled suspension).
2. Materials and methods
2.1. Materials
2.1.1. Experimental design
The experimental set-up, shown in Fig. 1, is a
Sartorius ltration module used with plane organic
membrane, the characteristics of which are given
in Table 1. The module consisted of a pressurised
ltration cell with a working volume of 200 ml.
2.1.2. Protein solutions
The protein solutions were prepared by dissolving a
commercial powder (Protarmor 907, Armor protine)
constituted of over 95% -lactoglobulin (molecular
weight of about 18 000 Da) in a ltered distilled water.
The -lactoglobulin molecule is approximately spher-
ical, with a diameter of 3.6 nm [3]. Its isoelectric point
is 5. At 20
C, 1 bar) (l h
1
m
2
)
400
Membrane resistance R
m
(m
1
) 0.9 10
12
S. Ognier et al. / Journal of Membrane Science 209 (2002) 2737 29
Fig. 1. Schematic diagram of experimental set-up.
molecular cut off of the ultraltration membrane
(
3
Z (1)
According to the experimental results, the charac-
teristics of the membrane (Table 1) and k equal to 4.5
(Kozeny Karman law), the specic surface area, S
m
,
can be calculated by using Eq. (1), i.e. 9 10
7
m
1
.
To quantify the additional resistance caused by
monolayer adsorption, R
mono
, the porosity,
ad
, and
the specic surface area, S
ad
, of the adsorbed mem-
brane have to be estimated.
36 S. Ognier et al. / Journal of Membrane Science 209 (2002) 2737
Two hypothesis were suggested:
(i) the specic surface area, S
ad
, of the adsorbed
membrane is almost the same as the specic
surface area of the clean membrane [4] and so,
R
mono
R
m
=
(1
ad
)
2
3
(1 )
2
3
ad
1 (2)
(ii)
ad
can be determined by the following equation:
ad
=
V
adsorbed proteins
Z
(3)
If we suppose that the spherical -lactoglobulin
molecules are closely packed on the available mem-
brane surface,
V
adsorbed proteins
=
4
6
d
lactoglobulin
S
m
Z(1 ) (4)
The value of
ad
thus obtained is 0.66, approxi-
mately 15% lower than the porosity of the clean mem-
brane ( = 0.72). According to Eq. (3), the R
mono
/R
m
ratio was 0.92.
It is noteworthy noticed that R
mono
/R
m
is close to the
experimental value R
l
/R
m
which is equal to 1. This
supports the hypothesis of a close packed monolayer
of -lactoglobulin adsorbed on the membrane surface,
which in turn suggests that the dynamic method is a
valuable method for measuring adsorption.
4.2. Role of adsorption in membrane fouling in the
case of a MBR
In subcritical conditions, the formation of a macro-
scopic deposit on the membrane should be avoided.
Therefore, the fouling increase is extremely slow and
the membrane resistance can be considered as constant
with time: a stable ltration regime is obtained. How-
ever, the membrane resistance differs from the clean
one, indicating that a fouling phenomenon takes place.
In subcritical conditions, it was reported that this con-
stant fouling resistance did not depend on hydrody-
namic conditions. Considering these results, one can
ask about the role of internal fouling. Let us consider
some of the results published in the literature:
Tardieu et al. [1] and Fan et al. [2] used activated
sludge ultraltration on tubular ceramic membranes
with a pore size of 0.02 m. They show that the ini-
tial fouling resistance values were below 30% that
of the clean membrane. Then, a very slow increase
in fouling resistance with time was observed in sub-
critical conditions.
Defrance and Jaffrin [10] measured, in short term
experiments with subcritical conditions, a foul-
ing resistance equal to 2.3 10
12
m
1
. This was
approximately 10 times higher than the clean mem-
brane resistance (ceramic membrane with 0.1 m
pores, R
m
= 0.24 10
12
m
1
).
Beaubien et al. [11] studied membrane fouling
in the low pressure zone of an anaerobic MBR
with ceramic membranes of 0.2 m pore diame-
ter. The membrane resistance was independent of
pressure and shear stress and the fouling resistance
values ranged from 1 10
12
to 5 10
12
m
1
when the suspended solids concentration var-
ied from 0.5 to 20 g l
1
, which corresponded to
220 times the resistance of the clean membrane
(0.25 10
12
m
1
).
These results show that in subcritical conditions,
the fouling resistance values range from 20 to 2000%
of the clean membrane resistance R
m
. When the pore
diameter is small (0.02 m), the subcritical fouling re-
sistance values are comparable to those obtained in
the present study, below 100% of the clean membrane
resistance. However, they increase dramatically when
pore size increases (microltration range) with foul-
ing resistance values from 400 to 2000% of the clean
membrane resistance.
If the fouling compounds liable to adsorb on the
membrane material adsorption were only small macro-
molecules, the adsorption phenomenon would induce
a more signicant fouling for small pore diameters.
Therefore, it means that the greater sensitivity of mi-
croltration membranes to fouling is caused by high
molecular weight compounds unable to penetrate in
the membrane matrix when the pore diameter is in
the ultraltration range. For large enough pore diam-
eter, these compounds can adsorb or be physically
blocked in the membrane pores. Thus, the choice of
pore size is a critical point as are the hydrodynamic
conditions. In the case of a MBR used for wastewa-
ter treatment, the results show that the ultraltration
range is more efcient than the microltration one to
prevent the irreversible fouling: in this case, the resis-
tance increase is only due to adsorption and remains
relatively low.
S. Ognier et al. / Journal of Membrane Science 209 (2002) 2737 37
5. Conclusion
Adsorption experiments on synthetic protein so-
lutions and MBR suspensions allowed us to charac-
terise and to quantify the inuence of adsorption on
membrane fouling. It was shown for the two types of
suspensions that:
1. due to limited diffusion, the dynamic method
(irreversible fouling remaining after ltration)
appears to be more adapted than the static method
(membrane immersion);
2. adsorption is a rapid fouling phenomenon that in-
duces irreversible resistance; adsorption takes place
at the beginning of the ltration process before the
deposition mechanism;
3. the adsorption phenomenon shows specic hy-
draulic resistance of the same order of magnitude
as the clean membrane resistance;
4. in the ultraltration range, the membrane resistance
increase obtained in a MBR in subcritical condi-
tions can be explained by an adsorption pheno-
menon. In the microltration range however, higher
size compounds seems to be involved in the fouling.
However, during long-term treatment operation in
a MBR, sudden malfunctions can be observed even
in subcritical conditions [12] and further research is
needed to elucidate the phenomena at the origin of
these sudden resistance increases.
References
[1] E. Tardieu, A. Grasmick, V. Jaugey, J. Manem, Hydrodynamic
control of bioparticle deposition in a MBR applied to
wastewater treatment, J. Membr. Sci. 147 (1998) 112.
[2] X.-j. Fan, V. Urbain, Y. Qian, J. Manem, Ultraltration of
activated sludge with ceramic membranes in a cross-ow
membrane bioreactor process, Water Sci. Technol. 41 (1011)
(2000) 243250.
[3] K.M. Persson, G. Capannelli, A. Bottino, G. Trgardh,
Porosity and protein adsorption of four polymeric
microltration membranes, J. Membr. Sci. 76 (1993) 6171.
[4] M. Martin-Cereceda, F. Jorand, A. Guinea, J.C. Block,
Characterization of extracellular polymeric substances in
rotating biological contactors and activated sludge ocs,
Environ. Technol. 22 (2001) 951959.
[5] E. Iritani, S. Tachi, T. Murase, Inuence of protein
adsorption on ow resistance of microltration memb-
rane, Colloids Surf. A: Physicochem. Eng. Aspects 89 (1994)
1522.
[6] D. Mckel, E. Staude, M.D. Guiver, Static protein
adsorption, ultraltration behavior and cleanability of hydro-
philized polysulfone membrane, J. Membr. Sci. 158 (1999)
6375.
[7] F. Martinez, A. Martin, P. Pradanos, J.I. Calvo, L. Palacio,
A. Hernandez, Protein adsorption and deposition onto
Microltration membranes: the role of solute-solid interaction,
J. Colloid Interface Sci. 221 (2000) 254261.
[8] J. Kim, C. Lee, I.-S. Chang, Effect on pump shear on the
performance of a crossow membrane bioreactor, Water Res.
35 (9) (2001) 21372144.
[9] J.H. Roorda, J.H.J.M. Van der Graaf, Understanding
membrane fouling in ultraltration of WWTP-efuent, Water
Sci. Technol. 41 (10/11) (2000) 345353.
[10] L. Defrance, M.Y. Jaffrin, Comparison between ltrations
at xed transmemebrane pressure and xed permeate ux:
application to a memebrane bioreactor used for wastewater
treatment, J. Membr. Sci. 152 (1999) 203210.
[11] A. Beaubien, M. B aty, F. Jeannot, E. Francoeur, J. Manem,
Design and operation of anaerobic membrane bioreactors:
development of a ltration testing strategy, J. Membr. Sci.
109 (1996) 173184.
[12] S. Ognier, C. Wisniewski, A. Grasmick, Biofouling in
membrane bioreactors: phenomenon analysis and modelling,
in: Proceedings of the Third International Conference on
Membrane Bioreactors, Craneld.