Professional Documents
Culture Documents
Use of Crossflow Microfiltration in Wastewater Treatment - 1997 - Water Research
Use of Crossflow Microfiltration in Wastewater Treatment - 1997 - Water Research
3064-3072, 1997
~ Pergamon © 1996ElsevierScienceLtd. All rights reserved
Printed in Great Britain
PII: S0043-1354(96)00084-X 0043-1354/97$17.00+ 0.00
3064
Crossflow microfiltration 3065
II011 0 n
(1) (2)
Fig. 1. Schematic representation of (1) conventional and (2) crossflow filtration.
created by the less in membrane permeability, has phenomenon of concentration polarization will be
been the focus of several investigations. Resistance to discussed in a separate section. Thus, the total
the filtration flux by the membrane material and resistance (Rt) can be expressed as follows:
particles in the feed, and concentration polarization
are the two main phenomena governing the flux rate Rt = Rm + Ra + Rp + Rep + Rg. (4)
decline in crossflow microfiltration.
Concentration polarization phenomenon
Resistance phenomenon
When a solution flows through a membrane-
Darcy's Law which relates the flow rate (the flux)
bounded channel where the solvent passes through
through a porous medium to the pressure drop can
the membrane while the solute is rejected at the
be applied. Darcy's Law states that the flux is directly
solution-membrane boundary, the criterion for local
proportional to the potential pressure drop and
steady-state mass transfer of solute requires that the
inversely proportional to the resistance (l/k). The
rate of convective transport of solute toward the
resistance explicitly includes the contribution of cake
membrane surface be equal to the rate of transport
and filter medium:
(by convective and diffusive mechanisms) of solute
k AP away from the membrane surface, Michaels (1968).
v = (1)
pl This condition can be satisfied only if the solute
concentration in the layer of solution adjacent to the
l/k = Rm + R~ (2)
membrane surface is greater than that in the bulk of
where v = liquid velocity, k = proportionality con- the solution within the channel. In the simplest terms,
stant, A P = pressure drop across the medium, concentration-polarization is the accumulation, at
1 = medium thickness, p = liquid viscosity, the upstream surface of the membrane, of solute
Rm = resistance caused by filter media, and molecules which are rejected or retained by the
R. = resistance caused by cake. membrane in the course of ultra- and microfiltration.
Thus, when filtering a suspension containing a wide More details about the concentration polarization
range of particles and colloids using a microporous concept are cited in Michaels (1968) and Blatt et al.
membrane at a constant pressure, the filtration flux, (1969). The concentration polarization model in
J, can be expressed by the following resistance crossflow microfiltration was also reported by
equation: Mulder (1991), Tanny (1978), Murkes (1990),
Murkes and Carlsson (1988), Anderson et al. (1986),
AP
S = ~,R--S (3) Van Den Berg and Smolders (1990), Zydney and
Colton (1986) and Schulz and Ripperger (1989).
where J = filtration flux, Rt = total resistance to the Basically, the steady-state mass-transfer conditions
flux = Rm + Re. which must exist within the polarized boundary layer
At a constant applied pressure and, for a given are represented by the following relationship:
solution, constant viscosity (at a given temperature
and composition), the filtration flux is inversely JCx - Ds ddCx - JCp = 0 (5)
proportional to the total resistance. The total
resistance consists of the resistance caused by the where J = transmembrane solvent flux,
filter media (R.,) which is the product of the pore size, Cx = concentration at the boundary layer,
the pore density, and the pore depth; resistance due Ds = molecular diffusion coefficient, Cp = permeate
to internal colloidal fouling (R. + R,); resistance due concentration.
to the formation of a highly concentrated layer By integrating, rearranging and assuming Cp = 0,
adjacent to the membrane, concentration polariz- the above relation can be simplified to:
ation (P.~); and re,sistance caused by the formation of
the gel layer (R~), due to the increasing concentration Cw
d = Km In ~ (6)
of particles near the surface of the membrane. The
3066 M.H. AI-Malack and G. K. Anderson
H 13 V5
r - * - --.n
CONCENTRATE
, i
, i
CFMF
6
AIR
LO a ooooo 0 0 mO|
0 0 o o o o o ~ L j 14
11 V4
lz_.l 7
V3
VI
PERMEATE
DRAIN
V2
10
where Km is the particle mass-transfer coefficient solution concentration, Cb, and a mass-transfer
between bulk suspension and membrane surface and coefficient, Kin, as follows:
Cb is the concentration of solids in the bulk
suspension. J =Km In C~ (7)
The gel formation process begins when, initially,
the rate of convective transport of the solute to the Km and Cs can be determined by a semi-log plot of
membrane is greater than the back-transport. This solution concentration versus limiting flux (taken as
results in an increase in the wall concentration until stabilized flux values)
a limiting concentration is reached, when further
concentration cannot take place because of the FACTORS AFFECTING CROSSFLOW FILTRATION
impermeability of the polarized macromolecular
layer to other macromolecules. This limiting value The efficiency of crossflow filtration is primarily a
of Cw is referred to as Cs, the gel concentration. function of the operating parameters. This efficiency
Once Cw reaches C~, further build-up of the solute is measured by the filtrate flow rate (flux) and its
must occur by thickening of the gel layer rather than quality. The following factors were found to affect the
by further concentration. The formed gel layer will performance of crossflow filtration: (i) crossflow
act as a hydraulic barrier in series with the velocity; (ii) transmembrane pressure; (iii) tempera-
membrane. As the wall concentration, Cw, becomes ture; (iv) pore size of membrane; and (v) feed
constant and equal to C~, the flux will become a concentration.
dependent variable which is constant for a given
Table 2. Wastewater characteristics
Constituent Value
Table 1. Characteristics of the woven fabric
Turbidity (NTU) 22
Configuration Tubular pH 7.45
Material Polyester Density (g) 1001.84
Pore size 20--40#m Viscosity (cpu) 1.07
Radius 1.25 cm (0.5 inch) Alkalinity (mg/I) 300
Length 4 x 0.98 meter Chlorides (rag/l) . 3720
Cross-sectional area 4.597 m2 Sulfates (ms/I) 640
Total surface area 0.295 ms Suspended solids (mg/l) 66
Wall thickness 0.34--0.375 mm COD (rag/l) 130
Hydraulic resistance 2.7 x 106 (l/m) BOD (rag/l) 20
pH resistance range 2-14 TOC (mg/l) 86
Crossflow microfiltration 3067
600 "
400 ---
eq
g
200 ---
I
0-
2 4 6 8 10
TIME (hrs)
300 -
C F V = 1.2 m / s e e
o C F V = 1.2 m / s e e
200 -~_
f,qo
loo - - -
0-
2 4 6 8 10 12
TIME (hrs)
15
The progressive decline o f flux with time is a Figure 2 shows a flowchart of the process. The rig is
measure of h o w m u c h reality deviates from wishful comprised of the following components:
thinking. This flux decline c a n n o t be avoided a n d (i) feed tank: this was circular and of about 3 m 3 capacity.
(ii) circulation or concentration tank: this was a rectangular
eventually the flux becomes uneconomically low a n d tank with dimensions of 30 x 70 x 80 cm.
some type o f cleaning or regeneration o f the (iii) circulation pump: a centrifugal pump.
m e m b r a n e must be carried out. The cleaning process (iv) membrane: the membrane used throughout was made
a n d its frequency depend o n the filtered p r o d u c t a n d of multifilament polyester yarn woven in the form of
interleave fabric of tubular configuration when inflated with
on the chemical resistance o f the m e m b r a n e . M o s t
a tube of 25 mm diameter. Flowboxes and manifolds were
cleaning procedures are a c o m b i n a t i o n of hydraulic fitted to both ends of the tube to direct the flow through
a n d chemical cleaning. several passes of tubes both in parallel and series.
600
Legend
4oo- .....
0 2 4 6 8 10
TIME (hr)
Fig. 6. Effect of suspended solids concentration (SS) on permeate flux.
Crossflow microflltration 3069
Table 3. Results of flax modeling when secondary effluent was valve (V5) until the desired pressure was reached. It was
treated noticed that when excessive back pressures (more than
Experimental Calculated 1 bar) were applied, pinholes and breakthroughs would
flux SS flux form. Breakthroughs were seen to have a dramatic negative
(l/m2.hr) (mg/l) ln(SS) Km (l/m2.hr) effect on the permeate quality even at marginal rates. The
42 29 3.37 - 33.53 43 transmembrane pressure was calculated based on the
29 50 3.91 - 33.53 25 average value of both inlet and outlet pressures.
14 64 4.16 - 33.53 17 The hydraulic resistance was calculated using equation (3)
assuming that the kinematic viscosity of the permeate was
unity.
In the laboratory, secondary effluent from a domestic
Table 1 shows some characteristics of the fabric used in the wastewater treatment plant was used throughout the
project. research period. The wastewater was characterized in the
(v) permeate tank: the permeate tank was rectangular with Environmental Engineering Laboratory of the Civil
dimensions of 30 x 30 x 70 cm, and used to collect the Engineering Department at the University of Newcastle
permeate. upon Tyne. Table 2 shows the general characteristics of the
(vi) pressure gauges: these were used to measure the inlet secondary effluent delivered to the laboratory, the values
and outlet pressures. being that of an average wastewater in terms of suspended
(ix) flow meter: the circulation flow rate was measured by solids content.
a Ficher flow meter. During the course of the research, the wastewater
(vii) air-injection line: air was injected into the tubes, turbidity, as an example, ranged from 15 to 57 NTU which
through the inlet pipe with a flow meter installed to measure represents a suspended solids of about 20-130 mg/l. Most of
the air flow rates. the time the wastewater was oily which could have been
The wastewater was mixed continuously, in the feed tank, caused by the tanker delivering the wastewater, and which
using a multi-rotational speed stirrer. A submersible pump was used to handle all types of waste.
was placed on the bottom of the tank to feed the wastewater Routine cleaning, at the end of each run, was carried out
to a header-tank fixed at the top of the feed tank. using brushes. This practice employs the use of brushes on
Wastewater was fed, by gravity, from the header-tank to the the outside surface of the woven fabric and took place while
filtration rig via a float. The purpose of the float was to circulating tap water at low flow rates (1500 l/hr).
maintain the wastewater level inside the circulation tank by
feeding the same amount of wastewater equal to that being
produced as permeate.
The wastewater was circulated through the woven fabric RESULTS AND DISCUSSION
using a single-speed centrifugal pump with the flow rates
being adjusted using valve (V4), and measured using the Treatment of secondary effluent
flow meter. The crossflow velocity was calculated from the
flow rate values and the cross-sectional area of the tube. S e c o n d a r y effluent f r o m a c o n v e n t i o n a l activated
The permeation rate (flux) was measured by establishing sludge p r o c e s s was used for this investigation. T h e
the time required to fill the 250 ml cylinder. The flux rates first r u n was carried o u t by circulating b o t h the
were then calculated based on the total surface area of the p e r m e a t e a n d the c o n c e n t r a t e at a crossflow velocity
woven fabric. Flux values at the end of each run were
o f 1.63 m / s a n d m o n i t o r i n g the p e r m e a t e flux with
considered as the stabilized flux.
Inlet and outlet pressures were measured using pressure time, w h i c h is s h o w n in Fig. 3. T h e flux s t a r t e d at a
gauges. The back pressure could be applied by turning off relatively high value (24001/m2.hr) a n d d e c r e a s e d
60--
40 ----
jg
PC
,..] 20 - -
[,T,,
cu )
0"-
2.$ 3.0 3.5 4.0 4.5 5.0
In (SS)
Fig. 7. Relationship between suspended solids concentration (SS) and permeate flux.
3070 M.H. AI-Malack and G. K. Anderson
120
• s s - laa ~
• ss-2ao n~
• ss-4oo m~l
~ 80
~ -
N -
~ 40
~ --
O--
100 200 300 400 500
TIME (min)
Fig. 8. Effect of suspended solids (SS) on permeate flux when treating settled sewage.
sharply in the first few minutes of the run and by after 6 hours of running, was due to the off--on
more than 95% of its original value in the first 30 mechanism, namely switching off and on the
minutes of the experiment. After this, the flux started circulation pump. The flux rose to 482 l/m2-hr, which
to decrease at relatively slower rates until it came to is more than 19 times greater than the value before,
a value of 30 l/m 2.hr after 5 hours of running time. but decreased to a stabilized value of 301/m2"hr
The sharp decline in the flux, at the start of the run, within 3 hours. This mechanism causes the filter wall
is attributed to the fact that the filter was clean and to collapse, creating destabilization and loosening of
all pores were open. This will result in higher the bond of the particles adhering to the filter surface.
permeation rate (flux) which will in turn result in a These particles will be swept away by the crossflow
higher solids deposition rate, thus clogging the fabric shear force, once the circulation pump is switched on
more rapidly. The increase in flux to 4801/m2'hr, again. This mechanism has not been fully investigated
6O
d~ 4O
fq
,d 2O
k.
111111I 17111111
3.0 3.5 4.0 4.5 5.0 5.5 6.0
in (SS)
Fig. 9. Relationship between suspended solids (SS) and permeate flux.
Crossflow microfiltration 3071
60 m
40 '
20 ---
o--:; I I F I I l l l l l l l l I I I J l l l l l l l l l l l l l l l l l l l l l l I I I I I I I I r t l b l l l l l l l l
10 20 30 40 $0 60
PREDICTED FLUX
Fig. 10. Experimental versus model predicted flux values [Flux = 93 - 17 x In(55)].
in this study. The quality of the permeate during that solids concentrations like he did when he was
operation (off-on mechanism) was seen to be investigating the effect of solid concentration on
deteriorating but started to improve gradually ultrafiltration. Using stabilized flux values, in Fig. 6,
through the course of the run. a relationship between flux and suspended solid
The effect of crossflow velocity on the flux concentration, based on concentration-polarization
behaviour during secondary effluent microfiltration phenomena (equation (7)), was investigated and
was investigated. The investigation was carried out at Table 3 summarizes the results. The model produced
three different crossflow velocities for the same predicted flux values acceptably close to those
effluent (i.e. same', solid concentration). Figure 4 obtained experimentally.
shows the effect of' crossflow velocity on flux values. Figure 7 shows the steady-state flux values versus
It is clear that, as the crossflow velocity was the natural logarithm of the solids concentrations.
increased, the flux was stabilizing at higher values. The slope of the regression line represents the
The relationship between stabilized flux and mass-transfer coefficient, Kin, while the x-intercept
crossflow velocity is shown in Fig. 5, which was represents the natural logarithm of Cg, the gelatinous
produced using stabilized flux values from Fig. 4, at concentration.
a specific time. Figure 5 suggests a linear relationship Regarding the permeate suspended solids, it was
between flux and crossflow velocity. noticed that lower solids concentrations in the
The effect of crossflow velocity on permeate quality permeate were obtained when the feed contained a
was also investigated and showed that lower low concentration of suspended solids.
permeate suspended solids were obtained at lower
crossflow velocities. High crossflow velocities will TREATMENT OF SETTLED SEWAGE
result in higher transmembrane pressure, which have
forced some particles in the feed to penetrate through Settled sewage, after primary sedimentation, was
the membrane surface. used in this study at suspended solids concentrations
The effect of the. initial solids concentration in the between 180 and 400 mg/1. The significant effect of
secondary effluent on the performance of the such high solids concentration on flux is shown in
crossflow microfikration process was also investi- Fig. 8. At a concentration of 280 mg/l suspended
gated by studying the effect of three different solids solids, a flux value of about 10 l/m2.hr was obtained
concentrations at the same crossflow velocity. The after 5 minutes from the start of the experiment. This
results are shown in Fig. 6. It suggests that as the feed is natural and expected in microfiltration, where pore
concentration was increased, the stabilized flux values sizes are larger than those found in ultrafiltration and
decreased. Elkabir (1991), in his investigation of the crossflow velocities are lower. Thus, when treating a
effect of solids cortcentration, claimed that it has no feed such as settled sewage, where all types of particle
effect on the performance of crossflow microfiltra- sizes may be encountered, blockage of these pores
tion. This is because he did not use different initial could take place in the first few minutes of the run.
3072 M.H. Al-Malack and G. K. Anderson
The permeate quality in terms of suspended solids of Petroleum and Minerals (Dhahran, Saudi Arabia) for the
was also investigated. As the suspended solid content technical support provided during the preparation of the
manuscript.
of the feed increases, blockage of the pores increases
and at a faster rate, which results in build-up of a
thicker deposited layer. Such a layer was seen to REFERENCES
almost completely block the fabric pores (from the
very low flux values) which would certainly result in Anderson G. K., James A. and Saw C. B. (1986) Crossflow
retaining the solid particles in the feed, thus microfiltration--a membrane for biomass retention in
anaerobic digestion. 4th World Filtration Congress,
producing a permeate of high quality (low suspended Ostend, Belgium.
solids). The modeling of crossflow microfiltration Blatt W. F., David A., MichaelsA. S. and Nelson L. (1970).
with respect to suspended solids concentration was Solute polarization and cake formation in membrane
investigated using the experimental data obtained ultrafiltration: Causes, consequencesand control technol-
ogy. In Membrane Science and Technology (Edited by
from both secondary and settled sewage effluents. Flinn J.). Plenum Press, New York.
Simple regression was adopted in the model by taking Elkebir A. A. (1991). Application of crossflow filtration for
flux values as the dependent variable and the biomass retention in aeration tanks. Ph.D. Thesis,
suspended solids concentration as the independent University of Newcastle Upon Tyne, U.K.
variable. The results are shown in Fig. 9, which Gekas V. and Hallstrom B. (1990) Microfiltration
membranes, crossflow transport and fouling studies.
suggests a linear relationship between stabilized flux Desalination 77, 195-218.
and the initial suspended solids concentration. Figure Michaels A. S. (1968) New separation technique for the CPI.
l0 shows the experimental flux values versus those Chem. Eng. Technol. 6402), 31-44.
predicted by the model. Michaels S. L. (1989) Crossflow microfilters: The ins and
outs. Chem. Eng. 96(1), 84-91.
Mulder M. (1991) Basic Principles of Membrane Technol-
CONCLUSIONS ogy. Kluwer Academic Publishers, Dordrecht, The
Netherlands.
It could be concluded that using the woven Murkes J. (1990) Fundamentals of crossflow filtration.
polyester fabric without dynamically formed mem- Separation and Purification Methods 19(1), 1-29.
branes was found to produce impractically low Murkes J. and Carlsson C. G. (t988) Crossflow Filtration.
John Wiley and Sons Ltd, Chichester, U.K.
permeate flux values due to either pore size of the Schulz G. and Ripperger S. 0989) Concentration polariz-
membrane or the physical interactions between the ation in crossflow microfiltration. J. Membr. Sci. 40,
particles in the feed and the primary membrane. The 173-187.
stabilized flux was seen to vary between 2 and Tanny G. B. 0978). Dynamic membranes in ultrafiltration
42 l/m 2.hr and was found to depend mainly upon the and reverse osmosis. Separation and Purification Methods
7(2), 183-220.
feed suspended solids concentration and the Van Den Berg G. B. and Smolders C. A. (1990) Flux decline
crossflow velocity, while the suspended solids in ultrafiltration processes. Desalination 77, 101-133.
removal was also low. Chemical cleaning was, Van Der Horst H. C. and Hanemaaijer J. H. 0990)
occasionally, necessary depending on the feed solids Crossflow microfiltration in the food industry: State of
the art. Desalination 77, 235-258.
content. Zydney A. L. and Colton C. K. (1986) A concentration
polarization model for the filtration flux in cross-flow
Acknowledgements--The authors would like to express their microfiltration of particulate suspension. Chem. Eng.
gratitude to the Research Institute of King Fahd University Commun. 47, 1-21.