Wat. Res. Vol. 30, No. 10, pp.

2385-2392, 1996
Pergamon Copyright © 1996 Elsevier Science Ltd
Printed in Great Britain. All rights reserved
P l h S0043-1354(96)00153-4 0043-1354/96 $15.00 + 0.00

FILTRATION CHARACTERISTICS OF HOLLOW FIBER

MICROFILTRATION MEMBRANES USED IN
MEMBRANE BIOREACTOR FOR DOMESTIC
WASTEWATER TREATMENT

YASUTOSHI SHIMIZU*, YU-ICHI OKUNO, KATSUSHI URYU,

SADAMI OHTSUBO and ATSUO WATANABE
Biotechnology and Water Treatment Laboratory, R & D Division, TOTO Ltd, 2-1-1, Nakashima,
Kokurakita-ku, Kitakyushu 802, Japan

(First received June 1995; accepted in revised form April 1996)

Abstract---Cross-flow microfiitration using submerged membrane with air bubbling can provide a
continuous solid-liquid separation system by simple equipment, such as a low-rate suction pump, an air
blower and a vessel. In this system, the size of the separation module could be reduced by using hollow
fiber membranes packed in high density. In order to apply this beneficial process to a wastewater treatment
bioreactor, we constructed filtration models to design the membrane system. Higher transmembrane
pressure and lower fluidity of feed operations (e.g. transmembrane pressure of over 40 kPa and air-liquid
two phase flow velocity, u*, of under 0.5 m • s-') caused the rapid crowding of hollow-fiber membrane
elements aml reduced the effective membrane surface area. Under the conditions that do not cause the
rapid crowding, the steady-state filtration flux, J~, is expressed as J,, = VL = K " ~b" u *~°" MLSS -°-s,
where, VL, /C and ~ were lift velocity, filtration constant and the geometric hindrance coefficient of
membrane module, respectively. The J,, values of hollow fiber membranes corresponded to that of rigid
tubular ceramic membranes packed in low density. In this study, we defined the conditions that the hollow
fiber membrane module packed in high density could be applied for solid-liquid separation. The flexibility
and movability of hollow fiber membrane elements could not improve the flux. Copyright © 1996 Elsevier
Science Ltd

Key words---cross-flow filtration, microfiltration, activated sludge, membrane bioreactor, wastewater

treatment

NOMENCLATURE IZ~----bubbling strength, bubbling air flow rate per

unit projection membrane area for base
A / V = membrane packing density (m2- m -3) (m 3 . m-2. h-~)
Dp = membrane pore size (m) Vt = lift velocity (m" s -I)
dp= particle slize (m) 6~ -- thickness of particle-packed layer on membrane
J = filtration flux (m3, m -2' d -~ or m ' s -~) (m)
J~ = initial filtration flux (m 3. m -2. d -j or m ' s -~) r/f -- viscosity of filtrate (Pa. s)
J~=steady-state filtration flux (m3-m-2.d-~ or
m . s -I) t/,= viscosity of suspension (Pa. s)
J~m~ = measured steady-state filtration flux for hollow- ~b = geometric hindrance coefficient of membrane
module (--)
fiber membrane module (m3.m-2.d-~ or
m • s -~)
J~au~ffi steady-state filtration flux under uniform fluidity
of feed (ra3" m -2. d -~ or m • s -I) INTRODUCTION
K' = filtration constant, defined by equations (2) and
(6)(kg°'5'm -~5) Cross-flow microfiltration is one o f the m o s t
MLSS = dried sludge concentration (Mixed Liquor promising solid-liquid separation processes t h a t can
Suspended Solids) (kg. m -j)
AP = transmembrane pressure (Pa) provide high filtration rate. A suspension to be
R~ = hydraulic resistanceofparticle-packed layer on filtered is forced tangentially over a m e m b r a n e
membrane (cake resistance) (m-') surface to b a c k - t r a n s p o r t the filtered colloidal
R-ffimembraneresistance(m-~) particles from the surface. This back t r a n s p o r t
Rpffiplugging resistance (m -~) continuously provides a high steady-state flux.
T = temperature (°C)
u = feed velocity (m. s-') Therefore, the process has been applied in d o w n -
u* = air-liquid two-phase flow velocity (m. s -~) stream processing such as clarification o f b r o t h ,
cell harvesting a n d collection o f precipitates in
• Author to whom all correspondence should be addressed, bioindustry.

2385
2386 Y. Shimizu et al.

Also, in upstream processing such as solid-liquid In this method, a cross-flow stream over the
separation in membrane bioreactors, the cross-flow membrane surfaces is produced by air bubbling by
microfiltration has been applied to improve the diffusers situated below the membranes in a vessel.
productivity of fermentation by increasing cell The air bubbles flow upward with a fluid along the
density and preventing the accumulation of inhibi- membrane surfaces and induce a moderate shear
tory products in culture broth (Taniguchi et al., 1987; stress which generates the back-transport of filtered
Shimizu K. et al., 1994). colloidal particles from the membrane surfaces. This
For wastewater treatment, membrane bioreactors method not only produces cross-flow filtration
have been examined as a combination of activated without a high discharge rate circulation pump and
sludge processes for BOD removal (Stavenger, 1971; membrane casing, but also can be set up with only
Fane et al., 1980), for removals of BOD and Nitrogen simple equipment, such as a low rate suction pump,
(Suwa et al., 1989; Chiemchaisri et al., 1993), and an air blower and a vessel.
anaerobic process for methane production (Li et al., The design of membrane bioreactor systems
1984; Kimura, 1991). In these bioreactors, mem- requires flux prediction equations for the filtration
branes are used not only as an alternative of the that relate the operating parameters, fluid character-
sedimentation tank (i.e. used as for the solid-liquid istics amd membrane properties with the flux.
separation in ordinary reactors), but also to increase For the conventional filtration method, we have
the biomass. The membrane bioreactors can maintain already proposed equations for suspensions including
higher biomass concentration, and high organic uniform and ununiform particles, respectively
carbon removal is achieved by compact systems. (Shimizu Y. et al., 1993, 1994). These equations have
Moreover, minimized production of excess sludge is been extended to the third type of filtration method,
expected by maintaining a low food to microorgan- the cross-flow microfiltration using a submerged
ism (F/M) ratio (Smith et al., 1968; Suwa et al., membrane with air bubbling, as shown in equations
1989). (1)-(3) (Shimizu Y. et al., 1996). Figure 1 shows a
Several operation methods for cross-flow filtration cross-flow microfiltration model for a submerged
can be applied to the membrane bioreactors. The membrane.
most common method uses circulation pumps to
generate cross-flow feed streams over the membrane AP 1
Z - (1)
surface. This conventional filtration method has been t/f Rm + Rp
commercially available as water reclamation system VL = K'u"°MLSS -°5 (2)
(Smith et al., 1968). The conventional method is
simple and easy to use, and the filtration flux can
be regulated by controlling pumping conditions
Z <~ VL--}J~ = Z J (3)
such as feed velocity and transmembrane pressure
(Blatt et al., 1970; Porter, 1972; Fane et al., 1981).
However, this method consumes tremendous The steady-state filtration flux, J~, is expressed in
amounts of energy and may reduce the activity of terms of the balance between the initial filtration flux,
biocatalysts, such as bacterial cells and enzymes, in Z, and the lift velocity ( = the back-transport velocity)
the circulating feed. This is because this method for the particle, VL. Z is determined by the hydraulic
requires about 10 to 20 times as much circulation feed resistances of plugging (Ro), which attributes to
as filtrate by volume to maintain a high filtration flux, membrane properties such as pore size and surface
and the use of the circulation pump with high charge, and the membrane itself (Rm). Therefore, at
discharge rate generates excessive shear stress that the initial stage of filtration, J~, that is, the flux after
should damage the biocatalysts (Shimizu Y. et al., the initial rapid decline, is dependent only on the
1992). membrane properties, as expressed in equation (1).
A second method has been proposed to eliminate The VL is generated by the shear stress originated
the disadvantages of the conventional method. In from the flow velocity gradient over the membrane
this method a cross-flow stream is generated by and is expressed as a function of the operating
rotating membranes or rotating impellers set parameters of air-liquid two-phase flow velocity (u*),
in the vicinity of membrane surfaces (Matsushita which is induced by air bubbling, and fluid
et al., 1994). Since it does not need a circulation characteristics of MLSS (equation (2)).
pump with high discharge rate, it does not incur J , is expressed by the hydraulic resistances as in
the above-mentioned disadvantages. Moreover, it equation (4). R¢ represents the resistance of
can filter highly viscous fluids that the conven- particle-packed layer on the membrane.
tional method cannot process (Matsushita et al., Ap 1
1995). J~ - (4)
t/f R m + R p + P ~
Presumably, the third and latest method is the most
energy efficient one with the lowest shear stress and When Z > V,, the particles are deposited on the
suitable for wastewater treatment membrane bio- membranes which already caused the plugging and R~
reactors (Hiasa et al., 1987; Yamamoto et al., 1993). is formed onto (Rm + Rp) until J~ = VL, as shown in
 Wastewater treatment membrane bioreactor 2387

i
Rrn , ~ bubblb
© N
M.Jbubble
Ji~---~, VL lss ~ VL

Ji= -AP~ / Rm +1 Rp 1 JSs= AP ( Rm+ Rp+ Rc

TIf '

(Initial stage) (Steady state)

~"
•~
Ji(a}fl~
VL _L__'-
(a)

~ Ji(b)[ ~ ( b )
Operating time
Fig. 1. Cross-flow microfiltration model of submerged membrane with air bubbling.

Fig. l(a). That is, if the optimum membrane pore these types of membranes could be used for
size, which is not too small to restrict the permeability constructing the compact filtration systems.
o f filtrate and prevent the plugging o f the particles, In this study, we examined the adaptability of our
is selected and enough transmembrane pressure is models for hollow fiber membrane modules packed in
gained, J= is controlled only by VL. high density which is used in the wastewater
When Ji ~ VL, Rc does not develop during the treatment membrane bioreactors.
filtration, and the flux does not decline after the initial
stage. J~ can be expressed as J~ = Z, as shown in
MATERIALS AND METHODS
Fig. l(b), when a high-resistance membrane or a
membrane with pores larger than the particles that Activated sludge
cause plugging is used or low transmembrane Experimental colloidal particles were obtained from
pressure operation. domestic wastewater-derived activated sludge grown in a
0.45 m 3 membrane bioreactor. Two hollow fiber microfiltra-
In this method, eilLher tubular, plate and frame, or tion membrane modules (Mitsubishi Rayon Co. Ltd,
hollow fiber membrane modules can be applied. The Japan), having pore size, Dp, of 0.1/zm and membrane area
membrane modules will be classified into two groups, of 4 m 2 per module, and air diffusers were placed in a
(1) membrane modules with low packing density (i.e. bioreactor vessel. Domestic wastewater, with a BOD of
0.2 kg" m -3 and total nitrogen content of 0.05 kg • m -3, was
membrane surface area per unit module volume
fed into the bioreactor at the hydraulic retention time of
is small) using rigid membrane elements and 12 h.The bioreactor was aerated at a flow rate of 4.2 m 3. h -~
(2) membrane modules packed in high density with by an air blower for 25 min with the interval of 35 min. The
small and flexible elements. The tubular module membrane was sucked by a fixed delivery pump at a rate of
having a membrane packing density, A/V, of several 0.14 m 3. h -~. The filtration was conducted only during the
aeration. This has been described in detail in a previous
hundreds m 2- m 3 and plate and frame module having paper (Tsuji et al., 1996).
A / V o f tens m 2. m -3 belong to the first group
(Gutman, 1987). In tlhese modules, uniform fluidity of Microfiltration membranes
feed over the membrane surface can be achieved Four types of hollow fiber membrane modules and a
because of adequate feed space in the modules. The tubular alumina ceramic membrane module were used for
above-mentioned filtration models were constructed the filtration experiments to evaluate the effect of membrane
with tubular modules. configuration on the filtration flux. The hollow fiber
membrane modules are shown in Fig. 2.
On the other hand, the membrane packing density A-type: A bundle of hollow fiber membrane elements
o f hollow fiber modules that belongs to the second were folded to make both ends meet and the cut-ends were
group, is very high (several thousands m 2. m-3), and fixed to collect the filtrate. This module is produced by
2388 Y. Shimizu et al.

A C f =p

t t =o.9 i ;i
0.8

B D

=1.0

Fig. 2. Submerged type hollow fiber membrane modules and their geometric hindrance coefficients, q~:
Dp = 0.1 #m, AP = 30kPa, V, = 7.2m ~. m -2. h -I, MLSS = 3 kg. m -j, T = 20°C.

Mitsubishi Rayon Co. (Japan) as Sterapore-S with flux value after 4-h operation was regarded as the
membrane surface area of 8 m 2.
steady-state value, Jss.
B-type: A modified module of A-type. Folded ends were
dispersed and spread by a wire frame. D u r i n g the filtration, colloidal particles were
C-type: A modified module of A-type. Folded ends were deposited over the m e m b r a n e surface a n d formed a
cut and separately sealed by thermal treatment to let them particle-packed layer t h a t reduced the flux. T h e layer
move individually. thickness, 6c, was less t h a n 100 # m as observed by
D-type: A remodeled module of A-type, which is
produced by Mitsubishi Rayon Co. As Sterapore-L with scanning electron microscopy (data not shown). The
membrane surface area of 8 m 2. The folded elements were excessive tSc f o r m a t i o n over the m e m b r a n e surfaces to
aligned in two rows and the cut-ends were fixed to seal. s u r r o u n d i n g the n u m b e r s o f elements should reduce
Polyethylene membrane elements, having an inner the effective m e m b r a n e surface area for the filtration
diameter of 0.27 ram, an outer diameter of 0.41 mm and a
(Fig. 5).
pore size, Dp, of 0.1/~m, were used for these modules.
A tubular alumina membrane AS005U from TOTO Ltd G i v e n the distance between the elements of the
(Japan) has a Dp of 0.5 gm, an outer diameter of 10 mm, an modules, which was designed as 1 . 5 m m , the
inner diameter of 7 mm and a length of 200 mm. reduction o f effective m e m b r a n e surface area was
Cross-flow microfiltration assumed to be avoided.
Cross-flow microfiltration was carried out with the
filtration apparatus shown in Fig. 3. The modules were
submerged in the activated sludge with MLSS of 4
1-20 kg. m-L A cross-flow stream was produced by air
bubbling generated by a diffuser situated below the
submerged membrane. The bubbling strengths (V~), defined
as the bubbling air flow rates per projection membrane area
for the base, were set between 3 and 22 m 3 • m -2 • h-L The
1
cross-flow stream velocity, u*, for air-liquid two-phase flow
was measured by a submersible electromagnetic current
meter from Kenek Ltd (Japan).
Filtration was conducted for 4 h at a transmembrane
pressure AP of 4-50 kPa and temperature of 20°C. The
filtrate was weighed on an electric balance every 30 s.

RESULTS AND DISCUSSION

Filtration flux Fig. 3. Schematic diagram of filtration apparatus: 1, 10 1
filtration vessel; 2, tubular alumina membrane; 3, diffuser;
The filtration flux declined rapidly within initial 4, pressure gauge; 5, filtrate reservoir; 6, electric balance; 7,
1 h. F u r t h e r decline was quite small (Fig. 4), a n d the suction pump; 8, air flow meter; 9, blower.
 Wastewater treatment membrane bioreactor 2389

1.0 I membrane elements, which permits a rapid crowding

and a reduction of the effective membrane surface

._0"8t
"r..~
area due to high Ap, as shown in Fig. 6.
The rapid crowding was also affected by the fluidity
of the feed over the membrane surface, i.e. air-liquid
.,E 0.6 two phase flow velocity, u*. Low fluidity caused the

Jss rapid crowding and reduced the flux (Fig. 7). These
results suggest that the characteristics which control
the fluidity, such as MLSS and viscosity, affect the

o.,L!'_,
o.a ...........
o.o rapid crowding in a similar manner.
In the region without the rapid crowding, J,,
I exhibited a 1.0 power exponential relationship with
200 300 u* for the hollow fiber membrane, as well as for the
T i m e [min] tubular membrane (Fig. 7). Therefore, the membrane
Fig. 4. Filtration flux of intermittently aerated activated surface of the hollow fiber module packed in high
sludge through A-type hollow fiber membrane module: density could be effectively applied to the filtration
Dp = 0.1/Lm, AP = 30 kPa, V, = 7.2 m3 • m -2. h -j, as well as that of tubular module packed in low
MLSS = 3 kg- m -3, T = 20°C. density.
As the filtration flux of the movable and flexible
hollow fiber membrane was not larger than that of
Flux prediction model for hollow fiber membranes the unmovable and rigid tubular ceramic membrane,
The effects of transmembrane pressure AP on the the improvement of flux by the movement or
flux of A-type hollow fiber and the tubular membrane vibration of membrane elements, induced by a
modules are shown in Fig. 6. As known already in the turbulence flow of air bubbles over the membrane
conventional cross-flow microfiltration method using elements, may be small.
regid membranes such as ceramic tubular type, there The effect of MLSS on Jss was a - 0 . 5 power
were two regions in the correlation of transmembrane exponential relationship under 8 kg - m -3 of MLSS
pressure with steady-state flux (I and II in Fig. 6). The (Fig. 8). This relationship is identical with a
flux depended on AP in the region I, and not in the previously reported one for the tubular module, and
region II. In the region I, the flux is determined by should support our assumption mentioned above
membrane properties and controlled by Rm and Rp, (Shimizu Y. et al., 1996).
as shown in Fig. l(b). The AP-independent flux in the Consequently, J~ of the flexible hollow fiber
region II could be explained as the balance between membrane modules packed in high density can be
the flux and the lift w.~locity, VL, as shown in Fig. 1(a). expressed as Jss= VLOCU*I'°'MLSS-°5, when it is
The increment of A.r' induces the growth of Re. How- operated by sufficient transmembrane pressure such
ever, the effect of At' increment on J , is neglected by as several kPa and a particle-packed layer is formed
the growth of R¢ (Shimizu Y. et al., 1993). over the membrane surface without the rapid
The region III in Fig. 6 was newly observed only crowding. This relationship is identical with the
in the flexible hollow fiber membrane module packed cross-flow microfiltration model for rigid submerged
in high density, where the flux reduced from the membranes, which is noted for the motion of particles
AP-independent value. This phenomenon is thought in the vicinity of the membrane surface, that the
to be caused only by the movable property of the particles are controlled by the balance between the

uneffective membrane area

Pre~
eiYective membrane area
<

(D particle packed lay~

o ,
(Before Filtration) (After Filtration)
Fig. 5. Schematicdiagram of the reduction of effective membrane surface area for filtration caused by
the formation of a filtered particle-packed layer over the hollow fiber membrane elements.
2390 Y. Shimizu et al.

@@
go~ o° o , ; o ° o° 0o g o ° o° ° o
o @ o

,_,.__0.5-- =/ @ =_ -

.----0 . . . . . . . . . 0--.%...
"9

O~
-I
==~

0 I I
0 50
A P [kPa]
Fig. 6. Effect of transmembrane pressure, AP, on steady-state flux, Js, of intermittently aerated activated
sludge: V, = 7.2 m 3. m -2. h -t, MLSS = 3 kg. m -3, T = 20°C. Symbols: O, A-type hollow fiber
membrane (Dp = 0.1/~m); 0 , tubular alumina ceramic membrane (Dp = 0.5/~m).

filtration flux and the lift velocity, VL (Shimizu Y. filtration flux was examined. In these modules,
et al., 1996). module volumes were regarded as the cylinder or
rectangular prism volumes being accommodated in
Effect of membrane configuration on filtration flux them, and membrane packing densities were calcu-
Using four types of hollow fiber membrane lated by using these values. Figure 9 shows the
modules, the effect of membrane configuration on filtration fluxes of the modules against the bubbling
strength, V.. Here we could see that the higher the

2 ' I
I I

~'E "O
'~E 0.5
Of}
0~ "E 0.1
¢,o
O

0.2- 0 I I - O
0.2 0.5 1 2 0.01 I I
u'[m.s'] 5 10 20
Fig. 7. Effect of air-liquid two-phase flow velocity, u*, on
MESS [kg.ms]
steady-state flux, J,, of intermittently aerated activated Fig. 8. Effect of MLSS on steady-state flux, J,,, in
sludge: A P = 3 0 k P a , M L S S = 3 k g . m -3, T=20°C. intermittently aerated activated sludge for A-type hollow
Symbols: O, A-type hollow fiber membrane (Dp = 0.1/~m); fiber membrane module: Dp = 0.1 t~m, AP = 30 kPa,
O, tubular alumina ceramic membrane (Dp = 0.5 #m). V~=7.2m3.m-2.h -~, T=20°C.
 Wastewater treatment membrane bioreactor 2391

i I i for the filtration as well as that of tubular ceramic

°°o....°°,°°" membrane packed in low density.

~,~ 0.5
CONCLUSIONS

The effect of membrane configuration on the

filtration characteristics for activated sludge was
--j studied, using four types of hollow fiber membrane
modules and a ceramic membrane module. The
following conclusions were made.
(1) The movable and flexible properties of hollow
0.1 i I i
fiber membrane elements cause the rapid crowding of
5 10 50 100
elements and reduce the effective membrane surface
Va [ .r4.fi area at high-rate filtration operation (high AP) or
Fig. 9. Effect of membrane module configuration on under the low fluidity conditions such as low u* and
steady-state flux, J~, of intermittently aerated activated high MLSS.
sludge: AP=30kP~., M L S S = 3 k g . m -j, T=20°C. (2) The improvement of flux by the movement or
Symbols: (3, A-type (A/V = 1.4 x 103m2. m-J); O, B-type vibration of hollow fiber membrane elements is small.
(A/V =0.1 x 103m2.1aa-3); I-q, C-type (A/V= 3.0 x (3) The nonuniformity of fluidity over the
10am2" m-3); I1, D-type hollow fiber membrane modules
with Dp = 0.1 #m (A/V = 0.7 x 103m2. m-3). membrane surface due to the geometric hindrance of
hollow fiber membrane elements is defined as
geometric hindrance coefficient, ~b.
(4) A highly packed membrane surface of hollow
membrane packing density is set, the easier the rapid
fiber module can be effectively used for filtration. By
crowding becomes.
introducing the q~, the filtration flux of the hollow
For the filtration of a fluid with low particle
fiber membrane module could be applied to an earlier
concentration, such as the fluids in desalination or
filtration model that had been constructed for a
water purification, membrane modules packed in
ceramic tubular membrane packed in low density.
high density are effective. However, for the filtration
of a fluid with high particle concentration, for
instance, filtration in which the membrane is directly REFERENCES
submerged in activated sludge suspension, the
Blatt W. F., Dravid A., Micheals A. S. and Nelse L. M.
packing density of membranes must be optimized to (1970) Solute polarization and cake formation in
avoid the rapid crowding and use the membrane membrane ultrafiltration: cause, consequences and con-
surface area effectively. trol techniques. In Membrane Science and Technology
The nonuniformity of fluidity over the membrane (Edited by Finn J. E.), pp. 47-97. Plenum Press, New
surface due to the geometric hindrance of membrane York.
Chiemchaisri C., Wong Y. K., Urase T. and Yamamoto K.
elements may also affect the flux of the membrane (1993) Organic stabilization and nitrogen removal in a
modules with high packing density. To evaluate the membrane separation bioreactor for domestic wastewater
effect quantitatively, the geometric hindrance co- treatment. Filtration & Separation 247-252.
efficient, q~, is defined as the flux ratio, ~b = J~m)/J~,); Fane A. G., Fell C. J. D. and Nor M. T. (1980)
Ultrafiltration/activated sludge system--~evelopment of
the measured flux for hollow fiber membrane a predictive model. In Ultrafiltration Membranes and
modules, J,~m), against the flux obtained under the Applications (Edited by Cooper A. R.), pp. 631-658.
uniform fluidity of feed over the membrane surface, Plenum Press, New York.
J~,). J,~,) is regard,~d as the value for a tubular Fane A. G., Fell C. J. D. and Waiters A. G. (1981) The
relationship between membrane surface pore character-
ceramic membrane module.
istics and flux for ultrafiltration membranes. J. Membrane
Taking into account the newly introduced geo- Sci. 9, 245-262.
metric hindrance coefficient, the flux prediction Gutman R. G. (1987) Membrane Filtration, the Technology
equations for the hollow fiber membrane modules are of Pressure-Driven Crossflow Processes. Adam Hilger,
equations (1), (3) and (5), where the filtration Bristol, 54 pp.
Hiasa M., Yamamoto K. and Matsuo, T. (1987) Proc. 42nd
constant K' of the intermittently aerated activated Annual Conf. Jpn Soc. Civil Eng., pp. 825-825.
sludge was the same as the previously reported value Kimura S. (1991) Japan's aqua renaissance '90 project. Wat.
of 2.6 × 10-5 kg°5. m -15 (Shimizu Y. et al., 1996). Sci. Tech. 23, 1573-1582.
Li A., Kothari D. and Corrado J. J. (1985) Application of
VL = K' cku *t'°" LSS -°'5 (5) membrane anaerobic reactor system for the treatment of
industrial wastewaters. Ind. Waste Conf. Purdue Univ. 39,
The ~b values of the modules are shown in Fig. 2. 627-636.
The effect of the geo~aaetric hindrance of hollow fiber Matsushita K., Shimizu Y. and Watanabe A. (1994)
Cross-flow filtration of yeast extract with multi-tubular
membrane modules on the flux was small. The membrane module and rotating-disk membrane
membrane surfaces of hollow fiber membrane module. Kagakukogaku Ronbunshu 20, 725-728 (in
modules packed in high density were effectively used Japanese).
2392 Y. Shimizu et al.

Matsushita K., Kanekuni N., Itakura H., Shimizu Y. and Shimizu Y., Uryu K., Okuno Y. and Watanabe A. (1996)
Watanabe A. (1995) Microfiltration of soy sauce sediment Cross-flow microfiltration of activated sludge using
with rotary disk membrane module. Kagakukogaku submerged membrane with air bubbling. J. Ferment.
Ronbunshu 21, 66-73 (in Japanese). Bioeng. 81, 55-60.
Porter M. C. (1972) Concentration polarization with Smith C. V. Jr, Gregorio D. D. and Talcot R. M. (1968) The
membrane ultrafiltration. Ind. Eng. Chem. Prod. Res. use of ultrafiltration membrane for activated sludge
Develop. 11, 234-248. separation. Proc. 24th Annual Purdue Industrial Waste
Shimizu K., Furuya K. and Taniguchi M. (1994) Optimal Conf., 1300-1310.
operation derived by Green's theorem for the cell-recycle Stavenger P. L. (1971) Putting semipermeable membranes to
filter fermentation focusing on the efficient use of the work. Chem. Eng. Progr. 67, 30-36.
medium. Biotechnol. Prog. 10, 258-262. Suwa Y., Yamaguchi T., Urushigawa Y. and Hirai M.
Shimizu Y., Matsushita K., Shimodera K. and Watanabe A. (1989) Simultaneous organic carbon removal-nitrification
(1992) Shear breakage of bacterial cells with cross- by an activated sludge process with cross-flow filtration.
flow filtration. In Biochemical Engineering for 2001 J. Ferment. Bioeng. 67, 119-125.
(Edited by Furasaki S.), pp. 578-580. Springer-Verlag, Taniguchi M., Kotani, N. and Kobayashi T. (1987)
Tokyo. High-concentration cultivation of lactic acid bacteria in
Shimizu Y., Shimodera K. and Watanabe A. (1993) fermentor with cross-flow filtration. J. Ferment. Technol.
Cross-flow microfiltration of bacterial cells. J. Ferment. 65, 179-184.
Bioeng. 76, 493-500. Yamamoto Y., Doi H., Okumura S., Matsuda T. and
Shimizu Y., Matsushita K. and Watanabe A. (1994) Narukami Y. (1993) Removal of nitrogen and phos-
Influence of shear breakage of microbial cells on phorus of domestic wastewater by ceramic membrane
cross-flow microfiltration flux, J. Ferment. Bioeng. 78, filter submerged reactor. Proc. Jpn Soc. on Water
170-174. Environ. pp. 408-409.