7/25/2019 Investigation of Fouling Mechanisms Governing
1/5
Investigation of Fouling
Mechanisms Governing
Permeate Flux in the Crossflow
Microfiltration of Beer
P. Blanpain and M. Lalande
Laboratoire de GEnie des Pro&d& et Technologie Alimentaires (LGPTA),
lnstitut National de la Recherche Agronomique (INRA), 369 rue Jules Guesde, BP 39,
F-59651 Villeneuve dAscq, France
Based on a paper presented at the 7th World Filtration Congress, in Budapest,
Hungary in May 1996
The local phenomenology associated with membrane fouling has been investigated experimentally
at laboratory scale for the crossflow microfi ltration (CMF) of beer. Two downstream memb rane
processes were involved: clarification and sterilisatlon of beer. Fouling mechanisms were
interpreted and compared for two types of beer (clarified beer and rough beer), filtered through a
track-etched 0.2 pm polycarbonate membrane. Flux decay was analysed by using the combination of
the constant pressure blocking filtration laws with the measurement of membrane resistances
arranged in series. It was found that for both types of beer the permea te flux was governed by two
identical fouling mechanisms: an internal fouling of pores at the initial stages of filtration that
conforms to the standard blocking model, fo llowed by an external surface fouling conforming to the
cake filtration model. It was shown that the predominant membrane resistance arises from the build-
up of a loosely bound and reversible fouling layer over the membrane surface (representing more
than 80% of the total filtration resistance). Macrosolutes and colloids are likely to be involved both
in the progressive pore plugging and in the external fouling layer (in combination with the yeast
cells cake in the case of rough beer), because of their high tendency to interact with porous
material. Thus the relevance of using the so-called classical filtration laws for the investigation of
fouling mechanisms in terms of total resistance of the membrane in beer CMF has been
demonstrated.
C
ossflow microfiltration (CMF) has been evaluated as
an alternative method for beer processing in order to
obtain colloidal and microbial stabilisation.[-41
Research work has focused mainl y on two essential points
which are inherent in the application of membrane processes
in industry, namely:
J fouling mechanisms responsible for flux decay with
filtration time; and
0 the ability of membranes to be cleaned and regenerated
between two filtration cycles,
At the present time industrial membrane development for
beer filtration is limited both by low permeate flux and by
essential quality component retention. These two phenomena
arise from severe membrane fouling. Fouling in beer CMF
impli es several differen t mechanisms, such as:
0 Gel layer formationt3z 51 and concentration polarisation.t61
0 Cake layer formation.[7S *I
0 Pore blocking and in-depth adsorption/deposition.tg3 lo1
Relatively little work has been published on the local
phenomenology of membrane-solute interactions involved in
beer CMF, which may be because of the complexity of beer,
which contains a large variety of molecular and colloidal
fractions. This paper emphasises the key membrane fouling
mechanisms which are relevant to the crossflow microfiltra-
tion of beer with organic membranes.
THEORY
Blocking filtration laws
Constant pressure blocking filtrat ion laws, revised by Her-
mia,t] have been widely used in CMF for the analysis of
memb rane resistance increase in the course of filtrati on. They
have the advantage of a non-ambiguous interpretation of often
complicated phenomena which limit the filtration rate of a
complex solution like beer (including colloidal and particulate
fractions with a large particle size distribution). Blocking
filtration laws were derived from the dead-end filtration mode,
but they can also be applied in the crossflow filtration mode by
considerin g a constant solute back transport rate from the
membrane into the bulk stream at a stable crossflow velocity.
For constant-pressure filtrat ion, th e blocking filtration laws
can be written in a gener al characteristic form as follows :[l
where t is the filtration time (s), V the filtrate volume (m3), K
the fouling coefficient depending on the initial flow rate QO and
solution properties, and n is a parameter depending on
.filtration law.
In the present paper the convenient linearised form of the
blocking filtration laws has been chosen to fit experimental
data (see Table 1).
Resistances in series analysis
The group of filtration laws presented above is based on
Darcys Law, and expressions are develo ped in order to
model the change in overall hydraulic resistance. At any
:.-::.
. . . l / I
. * . . .
. . . . 8 ,
. * a . .
. . I . I
. I ) . . .
: : : : .
. . . . I
. * . . .
. . . .
. . . * *
. . * . I
. . . . .
* . . .
. . . . .
. . . .
. . . . .
. . .
. < .
: . : + : , y
=
7
EY
-.
2
a
cn
E
;
rT
0
3
0
4
i
3
P
7
;o
8
a
9
co
n,
z
i
7/25/2019 Investigation of Fouling Mechanisms Governing
2/5
. . . . .
. . . .
. . . . l
. . . .
. . l . .
. . . . .
0 . . . .
. . . .
. . . I .
. . . .
. . * . .
. . . . .
. . . . *
. . . . .
. * . . .
. . . . .
. . . . .
. .
. .
. 2. :
2
t
2
$
z
s
g
S
t ?
x
i
od
5
S
cd
b
-2
l
W
8
T
zi
d
glocklngliltratlon
law type
TABLE 1. CHARAC TERISTICS OF CONSTANT PRESSURE BLOCKING FILTRATION LAWS.
Coaespondlng Physical descrlptlon
(param& defined
l lnearlsedform
In Eqn.1)
Complete blocking
model (CBM)
Intermediate blocking
model (IBM)
2
Q = Qo - Kbv
1
1/Q = U/Qo) + Kit
Each particle reaching the membrane
blocks a pore (pore sea ling)
Each particle reaches the m embrane at any location
on the membrane.The proba bility of a particle
blocking a pore isevaluated
Standard blocldng
model (SBM)
Cakefiltration
model (CFM)
1.5
tlv =
U/&o)
+ W&-V
0
t/V = (l/Q01 +
(K/2)V
Porevolume decreases proportionally to filtrate volume
by particle deposition on pore walls
Particlesdepositonto themembranesurface
andafiltercakeforms
instant in the course of the fi l tration process, the permeate flux
J can be described as fol lows:
J2LP
S @t
(2)
where Q is the flow rate, S the membrane surface area, P the
transmembrane pressure, p the dynamic viscosity of the
filtrate and
Rt
the total filtration resistan ce.
The total f i l tration resistance & can be expressed by
adding up several resistan ces in series:
IJ An external foul ing re@tance, &f, located at the
membrane surface and w,hich can result from the concen-
tration polarisation and/or the build-up of a cake layer.
0 A fouled membrane resistance, &, comprising the virgin
mem brane resistanc e plus the internal fouling resistance.
Thus & is represented by & = R,f + &.
MATERIALS AND METHODS
Filtration experim ents for beer processing have been con-
ducted on a dead-end Sartorius filtration cell with a capa city of
200 ml. The schema tic diagram of the experimental f i l tration
unit with di fferent measuring instruments and detai ls has
been presented elsewhere. [W Two types of beer were used:
[7 A clarified (kieselguhr processe d) lager beer containing
less than one yeast cel l /ml, with haze below 1 EBC
(European Brew ery Conve ntion turbidity unit131).
0 A rough beer containing from 5 x lo5 to 1 x lo7 yeas t
cel ls/ml, with haze greater than 10 EBC.
The latter solution was, in fact, a reconsti tuted rough beer,
composed of the clar if ied beer with which a known am ount of
yeast cel ls was mixed. Yeast ceW size was evaluated using a
laser di ffraction technique (Mastersizer S, Malvern Instru-
ments Ltd, UK); yeast cei ls w ere found to have a diameter
ranging from 2 to 10 pm, with a mean diameter of 5 pm. This is
an order of magnitude larger than mem brane pores, which
ensured total rejection of yeast cells by the membrane. It
should be noted that neither solution contains the chill haze
that has to be removed in membrane beer processing.
Howeve r, they do contain the molecular compounds respon-
sible for chill haze, i.e. proteins and polyphenols, together
with dextr ins and ,@-glucans, whose concentration does not
exceed a few m g/l .
A track-etched Nuclepore membrane was used. It was a
polycarbonate membrane of 0.2 pm mean pore diameter, 10
pm thickness and 16% surface porosity. Such a membrane,
with regular cylindrical pores of uniform size and length, was
appropriate to this study. The transmembrane pressure P was
set either at 10 kPa or at 100 kPa. The stirring speed was
constant, and provided by a magnetic sti rring bar at 850 rev/
min. The fi l tration temperature was maintained at OC, as in
the brewing separation process.
The experimental procedure for measuring the resistances
in series was the same as has been reported elsewhere,r2, I41
namely:
0 R, +
R,f
from the value of the quasi-steady-state flux J,,
at the end of the filtration, i.e. (I$,, + R,f) = P/(pJ,,).
0 & from the water f lux J,, after gentle washing of the
membrane surface with Mil li -Q water, i .e. & =
P/(p&,).
This remo ves the reversible external fouling without
disturbing the @ternal mem brane fouling.
RESULTS
Blocking filtration laws analysis
Experimental data have been tested with the l inearised forms
of blocking fi ltration laws presented in Table 1. The minimum
m
E
-z
;
/ (4
Filtration time t (min)
J
(b) Filtrate volume v (ml)
I
Flgwo 1. SBM (a) and CFM (b) plots for rowk boar ml-a Ex-
slzh r
P =
100 kcI; 250 rovldm; 0% PC 0.2 F
s
7/25/2019 Investigation of Fouling Mechanisms Governing
3/5
F
E
z 150
2
.;
3
El00
50
o Experimental
- Calculated
0 20 40 60 80
100 12t
(a>
Filtration time t (min)
250
r
q
Experimental
- Calculated
Fi l trate volume v (ml)
I I
-0 2. SBN (a) l nd CFY (5) plol8 tw ekrl f lod b8w m lwofl thtlon.
hpUh88W dNl888: P = 100 It- 050 rov/mln; 0C; PC 0.2 pm
Nmkpore m8msr8 lw.
coefficie nt of linear regression R2 has been sought either on
the whole duration of the run or part of it.
It has been found tha t for all runs, experime ntal values
support SBM and CFM in the same manner for both clari f ied
beer and rough beer. Typical results of the data analysis are
presented in Figures la and lb for rough beer and in Figures
2a and 2b for clarified beer. F irst filtration confo rms to SBM
(for t imes ranging from 1 to 15 min), then CF M appl ies up to
the end of the run. As can be seen from the Figures, the end of
the SBM period corresponds to the beginning of the CFM
application (transition point A on the Figures).
In contra st with clarified beer, the rough beer experiments
som etime s exhibited a deviation from the linear relation
between
t/V
and
V
toward s the end of the run (point B in
Figure 1 b). This deviation, which occurs in the quasi-steady-
state filtration phase, is thought to be induced by the scouring
effect of the crossflow velocity, which tends to l imit the cake
layer build-up. It is also noted tha t the applicability of the
intermediate blocking law (IBM) cannot be rejected for rough
beer in the initial stag es of filtration (O-5 min).
Resistances in series ana/ysis
Measured resistances in the series &,
Ref
for both clarified
beer (at 100 kPa) and rough beer (at 10 and 100 kPa) are
summarised in Table 2. Measurements have only been
carr ied out for some runs, in contrast with the predicted
values presented in Table 3 , which have been carried out
using all of the runs. Figures 3a and 3b show the experimental
and calculated proportion of mem brane resistan ces for four
filtration runs at P = 100 kPa for clarified and rough b eer,
respectively. Experimental values are derived from the
experimen tal procedure described above, while the calcu-
lated ones are derived from the analysis of recorded filtration
data using the blocking filtration laws .
TABLE 2. EXPERIMENTAL RESISTANCES IN SERIES Ii& ,
Ref
AND MEMBRANE PORE DlAMmER DECREASE Ad.
Tvpe ol beer Ransmembrane Experimental pore Experimental
Experlmental
pressure P,
diameter decrease proportIonof&,
kPa Ad MI
properllon of R+
% %
Clar i f iedbeer 100 0.064
2
98
0122
a 92
0.064 2.9
97s
Roughbeer IO
0317 10.8
89.2
0115
10 90
0306 9 91
0319 11.5 88.5
Roughbeer 100
0161 5
94.6
0143
2 97.6
OS69
5.7 93.9
0371
1
98.6
TABLE 2. PREDICTED RESISTANCES IN SERIES &, & AND MEMBRANE PORE DIAMETER DECREASE Ad FROM BLOCKING
FlhATlON LAWS.
Type ol beer Ransmembrsne Calculated pore
Calculated
Calcolated
pressureP,
diameter decrease
proportlon of &,
kPa
Ad WI
propertlonof I?+
%
%
Clarif ied beer
100
0.085
7
93s
OS 11.6 88.4
0.062
3.7
96.3
0.065
4
96
OS1 17.3
82.7
Roughbeer 10 0.043
24.7
75.3
0.055 31.4 68.6
0.031
20.2
79.8
0.061 36 64
Roughbeer
100 0.081
63
93.9
0.054 33
96.9
0.085 6.9 93s
0.073 4.9
953
0.062
3.7
96.3
. . * l .
. . . . . I
. . . . .
. I . . . c
3 m
$2 20
q
0
run 1 run2
run3
run4
q
% Rm (experimental) j % Rm (calculated)
(b)
q % Ref (experimental) q % Ref (calculated)
lgurr 3. Expuime ntal and calcuk tad proportlon of mem brans co-
slstan ces in srios l& Ref for okrifiad boor (a) and rough beor (b)
for four filtration runs. Exporlmo atal condltlons: P = 100 kPq 850
rev/mln; 0C; PC 0.2 pm Nucl iepore membrane.
It is clearly seen that, for the two types of beer, the larger
p$rt of the total filtration resistance is represented by a
reversible fouling resistance on the membrane surface (%R,f
> 88). The fouled microporous membrane R, is thus not
mainly responsible for the flux decline, and represents for all
runs a proportion less than 12%. It is noted from Table 2 that
for rough beer the proportio n of R, decreases perceptibly
with the transmembrane pressure
P.
Table 2 also includes the
reduction in the mean pore diameter (ad) derived from the
measured resistance I&. The mean pore diameter decrease
is calculated from
.,=,[,- (+y5]
where do = 0.2 pm, and RQ is the virgin membrane resistance
at the beginning of filtration. It is found that irreversible
membrane fouling leads to a constriction of the pore section
ranging from 10% to 73%. Moreover, at 100 kPa the pore
diameter decrease Ad for rough beer is significantly higher
than that for clarified beer.
Blocking filtration laws allowed us to determine the
predicted proportions of membrane fouling resistances from
the flux decline analysis, and to compare them with experi-
mental ones. The relative proportions of R, and &f are
derived from the value of the quasi-steady-state flux J,, at the
end of the filtration and from the calculated membrane
resistance Ii& at the end of the SBM period (corresponding
to point A in Figures 1 and 2). It is thus assumed in the
calculation that the totality of the external fouling layer
inducing CFM is reversible.
It appears (see Table 3) that the predicted external fouling
resistance I&f prevails for both types of beer (%& > 64 at
10 kPa and %&f
> 83 at 100 kPa). Experimental and
calculated values of Ad (ranging from 0.06 to 0.12 pm) are
close for clarified beer. Furthermore, the proportion of R, is
enhanced at low transmembrane pressure for rough beer, as
has been measured experimentally (Table 2).
Such results are in agreement with those obtained
experim entally from resistances in series analysis. However,
it is seen that for rough beer the calculated pore diameter
decrease Ad is systematically less than the experimental one
(about twice as much). Such a disagreement is believed to
result from several fouling phenomena which have not been
considered here in the SBM and CFM development:
0 A fraction of the external fouling layer R, f over the
memb rane surface is irreversible.
0 Pore sealing occurs at the membrane surface (responsible
for the IBM applicability).
0 Internal membrane fouling continues at the end of the SBM
period.
DISCUSSION
Based on the data presented in this paper, the following
conclusions may be drawn:
(a) For both clarifie d and rough beer, constant pressure
blocking filtration laws confirm the transition from an initial
internal blockage of the membrane (n = 3/2) to the formation
of an external cake layer (n = O), also c alled the secondary
dynamic membrane.
(b) The larger part of the total filtration resistance (on the
basis of the quasi-steady flux .I,, at the end of filtration) is
represented by a reversible fouli ng resistance over the
memb rane surface for both types of beer.
Such fouling phenomena (point a) have already been reported
for beer CMF at laboratory and pilot scale.fer Point (b) is
contrary to some results reported on rough beer CMF with
ceramic membranes,lO, K w for which in-depth pore
plugging was found to be the dominant factor. This may be
related to the specific nature of the track-etched polycarbo-
nate membrane used. Its low surface porosity as well as
smooth surface are favourable to the build-up of a polarised
layer or labile fouling layer on the membrane surface.
Blocking filtration laws have been analysed here in terms
of the total resistance of the membrane. This method allows a
better understanding of the actual phenomena involved in
membrane fouling, and the significance of both the internal
and the external fouling resistances.
The predicted relative proportions of internal and external
fouling resistances from model equations have been found to
be in a fairly good agreement with experimental ones derived
from water permeability measurements (Tables 2 and 3).
Nevertheless, a scattering effect arises from the definition of
the actual nature and reversibility of the surface fouling layer
responsible for CFM applicability. This is especially evident in
the presence of yeast cells, for which calculated and
measured Ad have been found to be quite different.
In particular, the existence of an irreversible fouling layer
at the membrane surface, induced by strong m embrane-
solute interactions, cannot be discarded in our case.t, I
Such a strongly bound surface fouling, composed of deposited
colloids and macrosolutes such as proteins and carbohy-
drates, is thought to lead to the extension of the SBM
application, a phenomenon which has been previously
suggested in the CMF of colloidal suspensions., *, 13 This
foul ing mechanism, which consists of a progressive constric-
tion of the efficient pore section at the surface of the
membrane, is not considered in the SBM development.
From these experiments, it has been shown that fouling
mechanisms have not been significantly modified in the
presence of yeast cells, which are known to have a strong
fouling effect in crossflow microfiltration. It is rather the
dissolved solutes, shared by the two types o f beer (i.e.
colloids and macromolecules) whose size is far lower than
that of membrane pores, which govern membrane fouling. It
results that, even though the first objective in CMF of beer is
the separation of suspended solids, the presence of colloids
makes the process more complex and more difficult to control
throughout one filtration cycle.
7/25/2019 Investigation of Fouling Mechanisms Governing
5/5
ACKNOWLEDGMENTS
The authors.wish to thank the Terken brewery in Roubaix,
France for providing beer, and the Nord-Pas de Calais region
for financial support.
NO ME NCL A TURE
6 = in i t ia l pore d iameter of membrane, m
.J = permeate f lux, m/s
J,. = quasi-steady-state f lux at end of f i l tration, m/s
.J, = water f lux after washing of memb rane surface, m/s
K
= foul ing coef f icient depending on f i l t ra t ion law (Eqn. 1)
n
= parameter depending on f i l t ra t ion law (Eqn. 1)
P = transmembrane pressure, Pa
Q = fi l trate f low rate, m3/s
R. = init ia l memb rane resistance, mm
R,f = external fouling resistance, m-
R,,, = fouled memb rane resistance, rn-
Rt = total f i l tration resistance, m-
S = memb rane surface area, m2
t
= fi l tration t ime, s
V = fi ltrate volum e, m3
Ad = pore diamete r decrease, m
P
= dynamic v iscosity of f i l trate, Pa s
REFERENCES
1 Wagner, D., Donhauser, S. and Walla, G.: Cross-flow Mikrofi ltration
von Hefe und Bier. Filtration ohne Kieselgureinsatz . 21st Europe an
Brewery Conven tion Congress, Madrid, Spain , 1987, pp. 631-638.
2 Ryder, D .S., Davis, C.R., Anderson, D., Glancy, F.M. and Power, J.N.:
Brewin g experience with cross-flow fi l tration, MB AA Technical
Quarterly, 1988, 25, pp. 67-79.
3 Reed, R.: Advances in f i l tration, The Brewer, Septem ber 1989, 9, pp.
965-970.
4 Walla, G. and Donhauser, S.: Filtration of beer and residual beer
with crossflow microfi ltration. Proceedings of 5th World Filtration
Congress, Nice, France, 1990, Vol. 1, pp. 64-69.
5 Reed , R.J.R., Evans, S .P., Taylor, D.G. and Anderson, H.J.: Sing le
stage downstream processing of beer using pulsed crossflow fi l tration.
22nd Europe an Brewery Con vention Congress, Zurich, Switzerland,
1989, pp. 413-424.
6 OReil ly , S.M.G., Lumm is, D.J., Scott, J . and Molzahn, S.W.: The
applica tion of ceramic f i l tration for the recovery of beer from tank
bottoms and in beer f i l tration. 21st Europe an Brewery Conven tion
Congress , Madr id, Spain, 1987, pp. 639-646.
7 Leno el, M.: Beer recovery from yeast s lurr ies and tank bottoms from
pilot to industr ial results . European Brewery Conven tion, Monog raph
XVI : Separat ions Processes ; EBC Sympos ium, Leuven, Belg ium, 1990,
pp. 128-139.
8 Lenoel , M . , B lanpain, P. and Tay lor , J . : Rat ional des ign of a c leaning
procedure for microfi ltration memb ranes, MB AA Technical Quarterly,
1994,31, pp. 134-137.
9 Burreil, K.J. and Reed, R.J.R.: Crossflow microfi ltration of beer:
Laboratory-scale studies on the effect of pore s ize, Filtration &
Separa tion, June 1994, 31(4), pp. 399-405.
10 Gan, Q., Field, R.W., Bird, M.R., Engla nd, R., Howe ll, J .A.,
McKechnie, M.T. and OShaughnessy, C.L.: Beer c larif ication by
cross-flow microfi ltration: Fouling mechanisms and flux enhan cemen t,
Trans. IChem E, Part A, 1997, 75, pp. 3-8.
11 Hermia, J.: Constant pressure blocking fi l tration laws: Applica tion to
power-law non-New tonian fluids, Trans. Ind. Chem . Eng., 1982, 60, pp.
183-187.
12 B lanpain, P. , Herm ia, J . and Lenoel , M . : Mechanisms governing
permeate f lux and protein rejection in the microfi ltration of beer with a
Cyclopore memb rane, J. Membra ne Sci., 1993, 84, pp. 37-Y.
13 Analy t ica EBC, European Brewery Convent ion (Brauerei und
Getrlnke-Rundschau, Z&rich, Switzerland, 1987).
14 V isvanathan, C. and Ben-Aim, R. : S tudies on col lo idal membrane
fouling mechanisms in crossflow microfi ltration, J. Memb rane Sci.,
1989, 45, pp. 3-15.
15 McKechnie , M.T., Burrell, K.J., Gil l, C., Kotz ian, R. and OSull iva n,
P. : Ceramic membrane f i l t ra t ion of beer . Proceedings of IChemE Food
Process Engineer ing Conference, September 1994, Bath, UK.
16 Gan, Q., Howe ll, J ., Field, R.W. and Eng land, R.: Foulin g and backflush
in beer c larif ication using ceramic m embrane s. Proceedings of IChem E
Food Process E ngineer ing Conference, September 1994, Bath, UK.
0 Please enter m y order for the Hydraulic Ha ndbook 9th Ediion
[P IH l l+Bl ] ISBN: 185617 250 3 Pr ice: US$200/P125/NLG318
Name: Job Tit le:
Address:
Zip/Postcode:
Tel:
E-mai l :
Nature of Business:
State:
Country:
Fax:
Return to:
Alistair Davies
Priority Orders Department
Elsevier Advanced Technology
PO Box 150, Kidlington
Oxford OX5 lAS U K
Tel: +44 (0)lSSS 843181
Fax : +44 (0)1865 843971
E-mail : a.daviesOelsevier.co.uk
HYAD97
a..*.
. . . . .
. . . . l
. l * . .
. . . . .
. . . . .
. . * l .
. . l . .
. . . . .
. . , . .
. . . . .
. . . . .
. . . . .
. . . . .
* . . * .
. . , . .
. . . . ~
. . . . .
. . . .
. . .
. . 1
Y - n : , : .
, ; r
2
z
2
$a
z
x
i
S
2
a
w
2
F
G
2
0
9
a
CD
i
(z ,