wat e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 4 0 7e5 4 1 3

Avai lab le a t www.sc iencedi rec t .com

journa l homepage : www.e lsev ie r . com/ loca te /wat res

A comparison of membrane fouling under constantand variable organic loadings in submergemembrane bioreactors

Jinsong Zhang a,b, Jiti Zhou a, Yu Liu b,*, Anthony G. Fane a

aKey Laboratory of Industrial Ecology and Environmental Engineering, Ministry of Education, Dalian University of Technology,

Dalian, Chinab Singapore Membrane Technology Centre, Nanyang Technological University, Singapore

a r t i c l e i n f o

Article history:

Received 1 March 2010

Received in revised form

5 June 2010

Accepted 16 June 2010

Available online 1 July 2010

Keywords:

Membrane bioreactor

Variable organic loading

Membrane fouling

EPS

* Corresponding author.E-mail addresses: jszhang@ntu.eu.sg (J. Z

0043-1354/$ e see front matter ª 2010 Elsevdoi:10.1016/j.watres.2010.06.045

a b s t r a c t

The aim of this study is to compare the effect of constant and variable influent organic

loadings onmembrane fouling in submergedmembrane bioreactors (sMBRs). Two identical

lab-scale sMBRswere operated for 162days at an SRTof 30days,whereas the influent organic

loadingwaskept constant inoneMBR, andvaried inanother. Themicrobial characteristics of

sludge in termsofMLSS, boundEPS, EPS in thesupernatantandparticle sizedistributionwere

investigated in order to evaluate their respective effect on membrane fouling. During the

start-up period, membrane fouling in the MBR fed with variable loadings was more serious

than that in the MBR with the constant loading. However, at the stable state, the fouling

tendency was clearly reversed with less membrane fouling for variable feed strength. It was

shown that the contents of polysaccharides in the supernatant and particle size of the bio-

flocswere responsible for the observeddifferences in the fouling tendencies of the twoMBRs.

ª 2010 Elsevier Ltd. All rights reserved.

1. Introduction (Han et al., 2005; Lee et al., 2003; Nagaoka et al., 2000; Zhang

During municipal wastewater treatment by membrane

bioreactor (MBR), the variation in the influent flowrate is often

moderated by the use of an equalization basin, however the

organic loading still fluctuates substantially within a 24 h

period (Tchobanoglous et al., 2003). As the result, the Food to

Microorganisms (F/M) ratio will vary in a large range. Obvi-

ously, the variable F/M ratio would alter the microbial prop-

erties of the biomass in theMBR, and potentially impact on the

membrane fouling intensity. For instance, the variable F/M

ratio could influence the production of extracellular polymeric

substance (EPS) and solublemicrobial products (SMP). EPS and

SMP are implicated in membrane fouling (Nuengjamnong

et al., 2005; Rosenberger et al., 2006). So far, the effects of

a fixed F/M ratio onmembrane fouling have been investigated

hang), zjiti@163.com (J. Zier Ltd. All rights reserved

et al., 2006a). The results implied that higher F/M ratio

would affect the biomass to producemore EPS and SMP,which

results in greater fouling tendency. Concomitantly, the

biomass condition will also affect biofloc aggregation

behavior. This is important because the particle size distri-

bution of biomass is another factor that affects membrane

fouling in MBRs. Small particles fraction could deposit on the

membrane reversibly at sub-critical flux. Above critical flux,

the deposition could become irreversible and adversely

affecting the MBR performance (Cho and Fane, 2002; Collins

et al., 2006; Stricot et al., 2010; Zhang et al., 2006b).

However, there have been few studies, which focus on

the influence of variable F/M on membrane fouling. The aim

of the present work is to evaluate the effect of daily variations

of the influent substrate concentration on MBR performance

hou), cyliu@ntu.edu.sg (Y. Liu), agfane@ntu.edu.sg (A.G. Fane)..

Table 1 e Composition and concentration of theconcentrated synthetic wastewater.a

Nutrient mg/L

Glucose 800

Meat extract 150

wat e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 4 0 7e5 4 1 35408

in terms of fouling. In addition to fouling trends, measured as

TMP rise, the particle size distribution, EPS, and MLSS were

monitored are the 162 days of operation in each reactor.

The fouling comparison for constant loading and variable

loading was performed in short- and long-term fouling tests.

The results could offer useful information for MBR design.

Peptone 200

KH2PO4 35

MgSO4 35

FeSO4 20

Sodium acetate 600

aDiluted (on average) 4.8� as feed to MBRs.

2. Materials and methods

2.1. Laboratory scale MBR

The experimental MBR system (Fig. 1) comprised 2 bioreactors

(30 L aerated tank each) equipped with submerged flat sheet

microfiltration (MF) modules (Kubota, 0.12 m2 each panel and

membrane pore size of 0.2 mm). Concentrated simulated

municipal wastewater (see Table 1) was continuously pumped

into the bioreactors at constant flowrate and variable flowrate

into the aerated tank 1 and 2, respectively. Tap water was

provided as a supplement to both aerated tanks through

solenoid valves controlled by level sensors, whichmaintained

a constant level in the bioreactor. Since the membranes

operated at constant flux, the feed rates were constant. In this

way, the concentrated feed (Table 1) was about 4.8� diluted

with tap water in the well mixed aerated tanks. An I-FIX

system control software and a WAGO Programmable Logic

Controller (PLC) were used to keep the permeates flowrate

constant. The transmembrane pressure (TMP) data was

measured by Cole-Parmer high accuracy (�0.13 kPa) pressure

transducers. Air diffusers were introduced separately to each

channel between membrane modules.

2.2. Analytical material and methods

Analytical methods for mixed liquor suspended solids (MLSS)

followed the Standard Methods (Clesceri et al., 1998). The

Fig. 1 e Schematic

supernatant samples were prepared by centrifuging the

mixed liquor samples from the bioreactors twice at 4000 rpm

for 10 min each time. The particle size distribution of the

biomass was measured by a particle sizer (MALVERN Mas-

tersizer HYDRO2000SM). The pellet EPS extraction followed

the ‘formaldehyde plus NaOH extraction method’, as reported

in our previous paper (Zhang et al., 2006a,b). The poly-

saccharide content in EPS was measured by the phe-

nolesulphate acid method (Dubois et al., 1956), using glucose

as the standard for the calibration. The protein content in EPS

was determined by the Bradford-bovine serum albumin (BSA)

method (Bradford, 1976). TOC was measured by a SHIMADHU

TOC-VCSH. Dissolved oxygen (DO) was measured by a Met-

tler-Toledo online DO meter.

2.3. Experimental conditions

The MBR 1 and 2 were seeded from lab-scale MBR which was

already run for more than 2 years at an SRT 30 days. For this

study, MBR 2 was set to variable feed and MBR 1 was kept at

constant feed. The operation conditions are summarized in

Table 2. The initial biomass concentration was set at 6e7 g/L.

of the MBR1&2.

Table 2 e operation conditions of MBR1&2.

MBR1 MBR2

SRT (days) 30 30

HRT (hours) 6 6

Operation time (days) 162 162

Reactor temperature (�C) 24e26 24e26

Aeration intensity (m3/m2 h) 0.75 0.75

pH 7e8 7e8

Fig. 3 e Supernatant and permeate TOC at constant and

variable loading.

wat e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 4 0 7e5 4 1 3 5409

From 20th day, the biomass concentrations increased to 9 g/L

in both reactors and gradually reached constant at 10� 0.5 g/L

during the experiment in the MBR 1 and 2. After 80 days of

unstable operation state (stage 1, more than 2 cycles of 30 day

SRT), a stable biomass state for 82 days was achieved for both

MBRs (stage 2).

The volumetric organic loadings for both MBRs were kept

at the same average value of 1.5 kg COD/m3day. In this work,

MBR 1was fedwith constant concentration influent, while the

feed concentration of MBR 2 was changed during the course of

a day. As shown in Fig. 2, the feed TOC concentration to MBR 1

was fixed constant at 135 � 5 mg/L, whereas the feed

concentration to MBR 2 was in a changing cycle to simulate

Singapore local municipal wastewater treatment plant. In this

cycle, from 0th to 2nd h, a high concentration of TOC of

270 mg/L wastewater was fed to the reactor. From the 2nd to

the 6th h, the feed TOC concentration was kept at 135 mg/L

and followed by 2 h of TOC 270 mg/L feed shock. From the 9th

to the 15th hour, the TOC concentration of feed was changed

to 135 mg/L. From the 15th to the 24th, the TOC concentration

of feed was changed to 100 mg/L for 8 h. From Fig. 2, some

variation in the DO can be also observed in the variable

loading run. Over 24 h, the DO fluctuatedwith the feed loading

rate. The variable range of the DO was from 1 mg/L to 3 mg/L.

3. Results and discussion

3.1. Overall performance of the MBRs

Fig. 3 shows the supernatant and permeate TOC concentra-

tions under the constant and the variable loading conditions.

In stage 1, the supernatant TOC concentrations in both reac-

tors fluctuated. It was found that the total supernatant TOC at

variable loading was slightly higher than that at constant

050

100150200250300

0 2 4 6 8 10 12 14 16 18 20 22 24hour

TOC

(mg/

L)

02468101214161820

DO

(mg/

L)

feed variable feed constant DO(mg/L) variable loading

Fig. 2 e The hourly variations in TOC of feed and DO in

aeration tank in one day.

loading at stage 1, whereas at stage 2, the supernatant TOC at

variable loadingwas only about 1/3 of that at constant loading.

It appears that under both the constant and variable

loading conditions, the performances of two MBRs in terms of

the permeate TOC concentration and TOC removal efficiency

are comparable, e.g. the permeate TOC concentration fell into

the range of 1.5 and 4mg/L, whereas 99% of total TOC removal

was achieved in both MBRs.

3.2. Bound EPS and soluble EPS production at constantand variable loadings

Fig. 4 shows the bound EPS and soluble EPS distribution in

the mixed liquor. It appears from Fig. 4 that the bound EPS

would be the major fraction of the total EPS. At stage 1 and 2,

the bound EPS concentrations both increased from about

0.24 g/L to 0.38 g/L with MLSS increased in MBR 1 and 2,

indicating no significant difference. The polysaccharide to

protein ratios (PS/PN) in bound EPS were 0.19 � 0.09 and

0.21 � 0.09 under the constant and variable loading condi-

tions, respectively.

From Fig. 4, at stage 1, the soluble EPS concentration was

0.011 � 0.004 g/L and 0.016 � 0.007 g/L under the constant and

variable loading conditions, respectively. However, at stage 2,

the soluble EPS concentration remained at around

0.013 � 0.04 g/L under the constant loading conditions, while

the soluble EPS concentration dropped to 0.006 � 0.003 g/L

under the variable loading conditions, respectively, which

indeed are pretty comparable.

Fig. 4 e EPS in supernatant and on the pellet at constant

and variable loading.

30405060708090

0 20 40 60 80 100 120 140 160 180

Parti

cle

size

(um

)

Days

Variable loading d50Constant loading d50

stage 1 stage 2

Fig. 6 e The median particle diameter (volume %) change at

constant and variable loading.

wat e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 4 0 7e5 4 1 35410

In the supernatant EPS, the protein concentration was

stable at 0.004 � 0.002 g/L and 0.005 � 0.002 g/L under the

constant and variable loading conditions respectively, at stage

1 and 2. The PS/PN was 2.3 � 1.1 at constant loading at stage 1

and 2. However, The PS/PNwas 2.3� 1.4 at constant loading at

stage 1 and dropped to 0.4 � 0.2 at stage 2, which implying

a significant drop in polysaccharide.

3.3. Soluble polysaccharide concentrations at variableand constant loadings

The colloidal and soluble organic materials in the mixed

liquor will cause membrane fouling in an MBR. Fig. 5 shows

the soluble polysaccharides concentration under the constant

and variable loading conditions. At stage 1, the soluble poly-

saccharides concentrations in both reactors fluctuated in

a wide range, while their trends seem to suggest that the total

soluble polysaccharides concentration at variable loading

would be higher than that at the constant loading. At stage 2,

the supernatant polysaccharides in both reactors were more

stable and their concentrations were 0.01 � 0.004 g/L and

0.003 � 0.001 g/L at constant loading and variable loading,

respectively.

3.4. Particle sizes at variable loading and constantloading conditions

Fig. 6 illustrates the size profile of the sludge particle at the

constant and variable loadings. At stage 1, themedian particle

diameter (d50, volume %) at constant loading was similar to

that at variable loading in the first 80 days, e.g. 60 � 3 mm and

56 � 3 mm at constant and variable loadings respectively. In

stage 2, the median particle diameter at constant loading

remained at the same level, whereas the median particle

diameter increased from 65 mm to 80 mm at the variable

loading. The sludge particle size number distributions under

two loading conditions were presented in Fig. 7. At stage 1, the

mean particle sizes in terms of number (d50) were 1.1 and

1.26 mm at variable loading and constant loading conditions

respectively.

In stage 2, the mean particle sizes in terms of number (d50)

at variable loading increased from 1.1 mm to 1.58 mm and the

distribution curve shifted to the right, while the number (d50)

at constant loading remain the same at 1.26 mm. The number

concentration of particles at variable loading was also lower

00.0050.01

0.0150.02

0.0250.03

0.0350.04

0 20 40 60 80 100 120 140 160 180

Poly

in s

uper

nata

nt (g

/L)

Days

polysaccharide in supernatant of constant loading polysaccharide in supernatant of variable loading

stage 2

stage 1

Fig. 5 e Polysaccharide in supernatant at constant and

variable loading.

than that at constant loading. The particle sizes based on

volume percentage (Fig. 6), is a good indicator for large parti-

cles distributed in the mixed liquor, because the volume of

one large particle could be equal to a large number of smaller

particles. The number concentration of particles more effec-

tively presents the small particles distribution in the mixed

liquor. In stage 1: from Figs. 6 and 7, the median particle

diameter based on the volume percentage was similar at

constant and variable loading, but the mean size of the

smaller particles based on the number concentration was

different. The mean size of the smaller particles at constant

loading was slightly bigger than that at variable loading, and

the number concentration of the smaller particles at variable

loading was higher than that at constant loading (9 � 104 VS

7.5 � 104). This is relevant because the small particles fraction

in the mixed liquor could affect the critical flux significantly

(Zhang et al., 2006) and the concentrationwould affect the rate

of accumulation (fouling).

3.5. Membrane fouling rates at constant and variableloadings

The flux stepping test is a method used to determine the

short-term “critical flux” and TMP increasing rate (fouling

rate) at each flux step. In the test, the flux was increased from

10 to 50 L/m2 h (LMH) by a flux step of 5 LMH. Before each step

test, the membrane was cleaned with 0.6% NaClO solution

and the clean water permeability was checked to be constant.

Each flux step was maintained for 10 h and TMP increasing

rate was measured.

Fig. 7 e Particle size distributions at constant and variable

loading.

Fig. 9 e TMP VS time at flux 20 LMH in stage 1.

wat e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 4 0 7e5 4 1 3 5411

Fig. 8 shows the plots of the TMP increase rate (dTMP/dt)

against flux under the variable and constant loading condi-

tions. As the flux was stepwise increased, the fouling rate

(dTMP/dt) also tended to increase throughout two operation

stages. At stage 1, the dTMP/dt under the variable loading

condition was higher than that under the constant loading

condition. The ‘critical flux’ at dTMP/dtw zero was estimated

at about 15e20 LMH under the variable loading condition and

at about 20 LMH under the constant loading condition.

However, in stage 2, the dTMP/dt under the variable loading

conditionwas consistently lower than that under the constant

loading condition at each flux step. The ‘critical flux’ was

approximately 25 LMH under the variable loading.

These results suggest that the MBR performance under

the variable loading condition was poorer than that under the

constant loading condition at stage 1. However, after the

biomass had become stabilized, the reverse was observed

at stage 2.

3.6. Long-term fouling comparison

Long-term runs were carried out to study the effect of variable

loading on membrane fouling. In stage 1, a constant flux of

20 LMH was maintained under both loading conditions. It

should be noted that the flux level of 20 LMH was over the

‘critical flux’ under the variable loading condition, but below

the critical flux under the constant loading condition (based

on Fig. 8 data). It can be observed in Fig. 9 that the TMP under

the variable loading condition increased faster than that

under the constant loading condition. Under the variable

loading condition, the TMP increased gradually at a rate of

approximately 0.26 kPa/day for 30 days, then followed by

a rapid increase in TMP (‘TMP jump’). The dTMP/dt at the

constant loading increased more slowly at a rate of around

0.09 kPa/day without a TMP jump, even after 33 days of

operation. These imply that themembrane fouling propensity

would be lower under the constant loading condition in

stage 1.

In stage 2, the constant flux operations at 20, 30 and 40 LMH

were tested and the fouling developments were shown in

Fig. 10. At the constant flux of 20 LMH in both reactors, it can

be observed that the TMP under the variable loading condition

increased slightly slower than that under the constant loading

condition, e.g. the dTMP/dt increased at a respective rate of

about 0.1 kPa/day and 0.08 kPa/day under the constant and the

variable loading conditions over 44 days of operation. This

00.010.020.030.040.050.060.070.08

10 15 20 25 30 35 40 45 50

dTM

P/dt

(kPa

/h)

Flux (LMH)

dTMP/dt at variable loading in stage 1dTMP/dt at constant loading in stage 1&2dTMP/dt at variable loading in stage 2

Fig. 8 e dTMP/dt VS flux at constant and variable loading.

may provide an explanation for the long-term fouling

observed at stage 2. At the flux of 30 LMH, the slower increase

in TMP was observed under the variable loading condition.

The dTMP/dt under the constant loading condition increased

gradually at a rate of 0.5 kPa/day for 15 days and followed by

a ‘TMP jump’. Under the variable loading condition, the dTMP/

dt increased gradually at a rate of 0.42 kPa/day for 28 days

without the ‘TMP jump’. At the flux of 40 LMH, the TMP

reached 40 kPa within 4 days of operation under the constant

loading condition; the increase of dTMP/dt was approximately

3.33 kPa/day for 3 days, followed by a ‘TMP jump’. Under the

variable loading condition, the dTMP/dt increased at slower

rate of approximately 0.48 kPa/day for 6.5 days without the

‘TMP jump’. Thus, in all the long-term fouling tests (Fig. 10),

the variable loading showed a lower fouling tendency.

4. Discussion

Previous study showed a clear relationship between fouling

rate and supernatant components, such as EPS and SMP

(Drews et al., 2008; Lesjean et al., 2005). The EPS abundance of

the MBR sludge is related to the environmental variations,

such as changes in organic loading rate (Liu and Fang, 2002).

Compared to a constant loading process, variable loadingmay

result in a periodical starvation in the reactor. In this study,

changes in feeding mode to the AS resulted in significant

changes in the soluble EPS concentration.

In the start-up phase (stage 1) under the variable loading

condition, the soluble EPS (Fig. 4), especially the soluble

polysaccharide concentrations were all higher than the

constant loading condition. The production of carbohydrate

by activated sludge was reported to increase significantly

during the periodical famine-feast (Chen et al., 2001; Yang

and Li, 2009), while the EPS content would change during

the transition period from steady state to unsteady state

(Drews et al., 2006; Lebegue et al., 2008; Sponza, 2002). The

production of EPS is a response of a microbial community to

changes in culture conditions (Liu et al., 2004). Thus, the

observed dynamic changes in the soluble EPS concentration

would result in active responses of microbial community to

changes of organic loading at stage 1. However, at the steady

state (stage 2), the soluble EPS content of the sludge

decreased to a low level, e.g. 70% lower than that observed in

stage 1. When microbial community gradually adapted to the

variable feeding scheme and reached the steady state (stage

Fig. 10 e TMP VS time at flux 20, 30 and 40 LMH in long-term operation at constant and variable loading in stage 2.

wat e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 4 0 7e5 4 1 35412

2), microbial response to the quasi-stable culture conditions

would be at the lowest level, thus the production of soluble

EPS might not be induced significantly as discussed above.

Furthermore, during starvation, part of the soluble EPS

or SMP would be degraded by their own producers or

other microorganisms (Nagaoka and Akoh, 2008; Wang et al.,

2006, 2007; Yang and Li, 2009). In stage 2, after the biomass

has acclimatized to the variable feed concentration of

substrate, the excess supernatant EPS will be quickly

consumed during the low feed concentration period. This

result is consistent with the previous report by (Zhang and

Bishop, 2003).

Several parameters have been identified which affect

particle sizes, such as hydrodynamics and cell attached EPS in

the biofilm and aerobic granules studies (Liu et al., 2004;

Skillman et al., 1999; Sutherland, 2001; Wang et al., 2005). In

this study, the hydrodynamics in the two MBRs were same

and the EPS result may also imply no correlation between the

floc size and pellet EPS production. Themechanismof variable

loading effect on floc size change remains unclear.

5. Conclusions

This study investigated the effect of variable loading on

the long-term performance of a lab-scale MBR. It was found

that membrane fouling in the MBR receiving variable loading

was more significant than in the MBR fed with a constant

loading during the start-up period. However, after two SRTs,

when the MBR systems gradually stabilized in terms of

biomass concentration and TOC removal, less membrane

fouling was observed in the MBR run under the variable

loading condition as compared with that operated at the

constant loading. The observed phenomena could be

adequately explained by changes in EPS and particle size of

bioflocs over the operation time.

It appears that variable loading would be an alternative

operation strategy for controlling membrane fouling in MBR

during long-term operation.

Acknowledgement

The authors acknowledge support from ASTAR Singapore for

the Temasek Professor Programme at Nanyang Technological

University (NTU). The work was completed in the Singapore

Membrane Technology Centre, at NTU, funded by the Envi-

ronment & Water Industry Development Council(EWI) of

Singapore. The author also thanks Ms Shuwen Goh for useful

discussion.

r e f e r e n c e s

Bradford, M.M., 1976. A rapid and sensitive method for thequantitation of microgram quantities of protein utilizing theprinciple of proteinedye binding. Analytical Biochemistry 72,248e254.

Chen, G.H., Yip, W.K., Mo, H.K., Liu, Y., 2001. Effect of sludgefasting/feasting on growth of activated sludge cultures. WaterResearch 35 (4), 1029e1037.

Cho, B.D., Fane, A.G., 2002. Fouling transients in nominally sub-critical flux operation of a membrane bioreactor. Journal ofMembrane Science 209 (2), 391e403.

Clesceri, L.S.,Greenberg,A.E., Eaton,A.D., 1998. StandardMethods forthe Examination ofWater andWastwater, twentieth ed. APHA.

Collins, J.H., Yoon, S.H., Musale, D., Kong, J.F., Koppes, J.,Sundararajan, S., Tsai, S.P., Hallsby, G.A., Cachia, P.,Kronoveter, K., 2006. Membrane performance enhancerevaluations on pilot- and full-scale membrane bioreactors.Water and Environment Journal 20, 43e47.

Drews, A., Vocks, M., Bracklow, U., Iversen, V., Kraume, M., 2008.Doesfouling inMBRsdependonSMP?Desalination231,141e149.

Drews, A., Vocks, M., Iversen, V., Lesjean, B., Kraume, M., 2006.Influence of unsteady membrane bioreactor operation on EPSformation and filtration resistance. Desalination 192, 1e9.

Dubois, M., Gilles, K.A., Hamilton, J.K., Rebers, P.A., Smith, F.,1956. Colorimetric method for determination of sugars andrelated substances. Analytical Chemistry 28 (3), 350e356.

Han, S.S., Bae, T.H., Jang, G.G., Tak, T.M., 2005. Influence of sludgeretention time on membrane fouling and bioactivities inmembrane bioreactor system. Process Biochemistry 40 (7),2393e2400.

wat e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 4 0 7e5 4 1 3 5413

Lebegue, J., Heran, M., Grasmick, A., 2008. MBR functioning understeady and unsteady state conditions: impact onperformances and membrane fouling dynamics. Desalination231, 209e218.

Lee, W., Kang, S., Shin, H., 2003. Sludge characteristics andtheir contribution to microfiltration in submergedmembrane bioreactors. Journal of Membrane Science 216(1e2), 217e227.

Lesjean, B., Rosenberger, S., Laabs, C., Jekel, M., Gnirrs, R., Amy, G., 2005. Correlation between membrane fouling and soluble/colloidal organic substances in membrane bioreactors formunicipal wastewater treatment. Water Science andTechnology 51 (6e7), 1e8.

Liu, H., Fang, H.H.P., 2002. Extraction of extracellular polymericsubstances (EPS) of sludges. Journal of Biotechnology 95 (3),249e256.

Liu,Y.Q., Liu,Y.,Tay, J.H., 2004.Theeffectsofextracellularpolymericsubstances on the formation and stability of biogranules.Applied Microbiology and Biotechnology 65, 143e148.

Nagaoka, H., Akoh, H., 2008. Decomposition of EPS on themembrane surface and its influence on the foulingmechanism in MBRs. Desalination 231, 150e155.

Nagaoka, H., Kono, S., Yamanishi, S., Miya, A., 2000. Influence oforganic loading rate on membrane fouling in membraneseparation activated sludge process. Water Science andTechnology 41 (10e11), 355e362.

Nuengjamnong, C., Kweon, J.H., Cho, J., Polprasert, C., Ahn, K.H.,2005. Membrane fouling caused by extracellular polymericsubstances during microfiltration processes. Desalination(Membranes in Drinking and Industrial Water Production) 179(1e3), 117e124.

Rosenberger, S., Laabs, C., Lesjean, B., Gnirss, R., Amy, G., Jekel, M.,Schrotter, J.C., 2006. Impact of colloidal and soluble organicmaterial on membrane performance in membrane bioreactorsfor municipal wastewater treatment. Water Research 40 (4),710e720.

Skillman, L.C., Sutherland, I.W., Jones, M.V., 1999. The role ofexopolysaccharides in dual species biofilm development.Journal of Applied Microbiology 85 (1), 13e18.

Sponza, D.T., 2002. Extracellular polymer substances andphysicochemicalpropertiesofflocs insteadyandunsteady-stateactivated sludge systems. Process Biochemistry 37 (9), 983e998.

Stricot, M., Filali, A., Lesage, N., Sperandio, M., Cabassud, C., 2010.Side-streammembrane bioreactors: influence of stressgenerated by hydrodynamics on floc structure, supernatantqualityandfoulingpropensity.WaterResearch44 (7), 2113e2124.

Sutherland, I.W., 2001. Biofilm exopolysaccharides: a strong andsticky framework. Microbiology 147, 3e9.

Tchobanoglous, G., Burton, F.L., Stensel, H.D., 2003. WastewaterEngineering Treatment and Resuse, fourth ed.

Wang, Z.W., Li, Y., Zhou, J.Q., Liu, Y., 2006. The influence of short-term starvation on aerobic granules. Process Biochemistry 41,2373e2378.

Wang, Z.W., Liu, Y., Tay, J.H., 2007. Biodegradability of extracellularpolymeric substances produced by aerobic granules. AppliedMicrobiology and Biotechnology 74, 462e466.

Wang, Z.W., Liu, Y., Tay, J.H., 2005. Distribution of EPS and cellsurface hydrophobicity in aerobic granules. AppliedMicrobiology and Biotechnology 69, 469e473.

Yang, S.F., Li, X.Y., 2009. Influences of extracellular polymericsubstances (EPS) on the characteristics of activated sludgeunder non-steady-state conditions. Process Biochemistry 44,91e96.

Zhang, J.S., Chua, H.C., Zhou, J., Fane, A.G., 2006a. Effect of sludgeretention time on membrane bio-fouling intensity ina submerged membrane bioreactor. Separation Science andTechnology 41 (7), 1313e1329.

Zhang, J.S., Chua, H.C., Zhou, J.T., Fane, A.G., 2006b. Factorsaffecting the membrane performance in submergedmembrane bioreactors. Journal of Membrane Science 284(1e2), 54e66.

Zhang, X., Bishop, P.L., 2003. Biodegradability of biofilmextracellular polymeric substances. Chemosphere 50 (1),63e69.

Zhang, Y.P., Fane, A.G., Law, A.W.K., 2006. Critical flux andparticle deposition of bidisperse suspensions during crossflowmicrofiltration. Journal of Membrane Science 282 (1e2),189e197.