ARTICLE IN PRESS

Available at www.sciencedirect.com

WAT E R R E S E A R C H 4 1 ( 2 0 0 7 ) 9 5 – 1 0 1

0043-1354/$ - see frodoi:10.1016/j.watres

�Corresponding auE-mail address:

journal homepage: www.elsevier.com/locate/watres

Soluble microbial products in membrane bioreactoroperation: Behaviors, characteristics, and fouling potential

Shuang Liang, Cui Liu, Lianfa Song�

Division of Environmental Science and Engineering, National University of Singapore, 10 Kent Ridge Crescent, Singapore 119260, Singapore

a r t i c l e i n f o

Article history:

Received 6 April 2006

Received in revised form

3 October 2006

Accepted 3 October 2006

Available online 15 November 2006

Keywords:

Soluble microbial products

Membrane bioreactor

Sludge retention time

Accumulation

Characteristics

Fouling potential

nt matter & 2006 Elsevie.2006.10.008

thor. Tel.: +65 6516 8796; feseslf@nus.edu.sg (L. Son

A B S T R A C T

This paper presents an experimental study on soluble microbial products (SMP) in

membrane bioreactor (MBR) operation at different sludge retention times (SRTs). A

laboratory-scale MBR was operated at SRT of 10, 20, and 40 days for treatment of readily

biodegradable synthetic wastewater. The accumulation, composition, characteristics, and

fouling potential of SMP at each SRTwere examined. It was found that accumulation of SMP

in the MBR became more pronounced at short SRTs. Carbohydrates and proteins appeared

to be the components of SMP prone to accumulate in the MBR compared with aromatic

compounds. The proportions of SMP with large molecular weight in supernatants and in

effluents were almost identical, implying that membrane sieving did not work for most

SMP. In addition, the majority of SMP was found to be composed of hydrophobic

components, whose proportion in total SMP gradually increased as SRT lengthened.

However, fouling potentials of SMP were relatively low at long SRTs. The hydrophilic

neutrals (e.g., carbohydrates) were most likely the main foulants responsible for high

fouling potentials of SMP observed at short SRTs.

& 2006 Elsevier Ltd. All rights reserved.

1. Introduction

Soluble microbial products (SMP), a myriad of soluble organic

matter produced by mixed bacterial populations in bioreac-

tors, are of crucial importance for biological wastewater

treatment systems because of their significant impacts on

both effluent quality and treatment efficiency. It is widely

accepted that SMP constitute the majority of soluble organic

matter in effluents from biological treatment systems (Barker

and Stuckey, 1999). Their concentrations in effluents, there-

fore, essentially determine the discharge levels of chemical

oxygen demand (COD) and dissolved organic carbon (DOC). In

addition, some SMP have been found to exhibit certain

characteristics, such as toxicity and metal chelating proper-

ties, which affect metabolic activities of microorganisms both

in treatment systems and in receiving waters, in some cases,

reducing their specific respiration rates (Barker and Stuckey,

r Ltd. All rights reserved.

ax: +65 6774 4202.g).

1999). Hence, it is desirable to minimize the concentration of

SMP for a better treatment performance.

SMP can be further classified into two categories: (a)

utilization associated products that are associated with

substrate metabolism and biomass growth, and (b) biomass

associated products that are associated with biomass decay

(Barker and Stuckey, 1999). Several major components of SMP

such as humic substances (e.g., humic and fulvic acids),

carbohydrates, and proteins have been successfully identi-

fied, though their precise composition remains unclear

(DeWalle and Chian, 1974; Hejzlar and Chudoba, 1986a, b). It

has been reported that SMP have a broad spectrum of

molecular weight and their apparent molecular weight

distribution is greatly affected by process parameters such

as sludge retention time (SRT), and food-to-microorganism

ratio (Barker and Stuckey, 1999). A certain optimum range of

SRT, from 2 to 15 days, has been found for conventional

ARTICLE IN PRESS

Table 1 – Composition and concentration of componentsof synthetic wastewater

Components Concentration (mg/L)

CH3COONa 768.75

(NH4)2SO4 284

KH2PO4 26

CaCl2 � 2H2O 0.368

MgSO4 � 7H2O 5.07

MnCl2 � 4H2O 0.275

ZnSO4 � 7H2O 0.44

FeCl3 1.45

CuSO4 � 5H2O 0.391

CoCl2 � 6H2O 0.42

Na2MoO4 � 2H2O 1.26

Yeast extract 30

WAT E R R E S E A R C H 4 1 ( 2 0 0 7 ) 9 5 – 1 0 196

aerobic treatment systems in which minimum SMP produc-

tion can be achieved (Barker and Stuckey, 1999).

So far, however, most of the previous studies have focused

on SMP in conventional biological treatment systems. Few

studies have been conducted to investigate SMP in membrane

bioreactor (MBR) systems, an innovative technology having

gained immense popularity in recent years (Huang et al.,

2000; Shin and Kang, 2003). The advantages offered by MBR

have been well documented in the literature such as excellent

effluent quality, low sludge production, high treatment

efficiency, and small footprint (Brindle and Stephenson,

1996; Ng and Hermanowicz, 2005). However, membrane

fouling remains the most serious problem for widespread

application of MBR technology (Chang et al., 2002).

In an MBR, membranes are employed for solid–liquid

separation instead of secondary clarifiers commonly used in

conventional biological treatment systems. The behaviors of

SMP, therefore, become even more complicated. Apart from

affecting biodegradation processes, SMP are implicated in

membrane filtration processes. It has been reported that SMP,

as major organic foulants to the commonly used microfiltra-

tion or ultrafiltration membranes in MBRs, contribute 26–52%

of membrane fouling depending on the experimental condi-

tions (Wisniewski and Grasmick, 1998; Bouhabila et al., 2001).

Recently, Lee et al. (2003) provided more detailed informa-

tion on the characteristics and fouling behaviors of super-

natants, where SMP reside, in a submerged MBR. The

filtration resistance caused by the supernatant was found to

be independent of SRT. The characteristics of SMP such as

molecular size, hydrophobicity, and zeta potential were

measured and correlated to fouling strength, but none of

them appeared to be a remarkable fouling parameter. It was

therefore suggested that more fundamental information,

probably new characteristic, was required for a better under-

standing of supernatant fouling. On the other hand, it has

been observed that concentrations of SMP in MBR super-

natants were much higher than those in MBR effluents,

indicating that some SMP components accumulate within

MBRs (Huang et al., 2000; Shin and Kang, 2003). The extent of

SMP accumulation is reasonably expected to be heavily

dependent on their characteristics and composition, which

would significantly vary with operating conditions such as

SRT. However, at present, the effect of SRT on SMP accumula-

tion has not yet been seriously examined. In particular, little

information is available on the characteristics of SMP in MBR

effluents such as apparent molecular weight distribution,

which is of critical importance in understanding SMP

accumulation.

The primary objective of this research was, therefore, to

contribute towards a better understanding of the accumula-

tion, composition, and fouling potential of SMP in MBR

operation. Experiments were conducted in a submerged

MBR at different SRTs. Characteristics of SMP both in super-

natants and in effluents were measured and compared. The

column chromatographic method was applied to investigate

the hydrophobic/hydrophilic and charge properties of SMP.

The results reported here would provide valuable new

insights into the characteristics of SMP, and would conse-

quently further advance our knowledge on the behaviors of

SMP in MBR operation.

2. Materials and methods

2.1. Experimental setup

Experiments were performed in a laboratory-scale submerged

MBR consisting of a rectangular tank having an operating

volume of 16 L and a flat-sheet membrane module submerged

in the tank. The membrane module was made of polyolefin

membrane with a pore size of 0.4mm and an effective

filtration area of 0.1 m2 (Type 203, Kubota, Osaka, Japan).

Aeration was done through the air diffuser installed directly

beneath the membrane module to maintain desired dissolved

oxygen (DO) concentration and to mix activated sludge in the

MBR. The air bubbles generated during aeration, on the other

hand, induced a crossflow scouring the membrane surface

and so suppressing membrane fouling. Two baffle plates were

mounted above the air diffuser to optimize the contact

between air bubbles and the membrane surface. The syn-

thetic wastewater was continuously supplied into the MBR by

a peristaltic pump (Model 7553-85, Cole-Palmer, Vernon Hills,

IL, USA) from the storage tank with an effective volume

of 50 L, which was refilled everyday. The membrane-

filtered effluent was extracted by a pump of the same

model operating intermittently with a cycle of 8 min on and

2 min off.

2.2. Synthetic wastewater and operating conditions

The composition of the synthetic wastewater is listed in Table

1. It should be noted that no technique is currently available

to rigorously identify SMP due to the difficulty in tracing the

origins of various soluble organic compounds in a treatment

system. In the present study, sodium acetate was chosen to be

the carbon source of the synthetic wastewater because it

could be regarded as completely removed in biological

degradation (Onuki et al., 2002; Slavica et al., 2004). As a

result, the concentrations of SMP in supernatants and in

effluents can be simply indicated by DOC measurements.

Nitrogen and phosphorus were provided by ammonium

ARTICLE IN PRESS

WAT E R R ES E A R C H 41 (2007) 95– 101 97

sulfate and monopotassium phosphate, respectively. The

influent COD concentration was 600720 mg/L with a COD:N:P

mass ratio of 100:10:1. The synthetic wastewater was freshly

made daily and the storage tank was thoroughly cleaned

every two days to prevent microbial growth.

Seed sludge was obtained from the aeration tank of a local

pilot MBR for municipal wastewater treatment. After trans-

ferring into the laboratory-scale MBR, the sludge was allowed

to acclimate to the synthetic wastewater for five weeks.

During this start-up period, the MBR was operated at the

same condition as that used in the experimental period

except no sludge wastage. The experiments were performed

in three phases according to the change of SRT in the order of

40, 20 and 10 days. The SRT of 40 days was investigated first in

consideration of minimizing the loss of acclimated sludge.

Before transferring to a new phase, a period of at least two

times of the new SRT was provided for MBR stabilization. In

each phase, a steady-state of four weeks was maintained,

during which 13–26 measurements were evenly conducted for

the parameters of interest. The number of measurements for

each SMP parameter was indicated in the captions of the

tables and figures.

The hydraulic retention time (HRT) of 10 h and DO

concentration of around 5 mg/L were maintained during the

entire experimental period of 256 days. The MBR was

operated under ambient temperature (2872 1C) and the pH

was controlled within a range of 7.0–8.0. Fouling develop-

ment, indicated by the increase in suction pressure, was

monitored using a digital pressure switch (ZSE50F-T2-22L,

SMC, Japan). Membrane cleaning was required in about 35–50

days when the suction pressure increased beyond 35 kPa.

The membrane module was taken out of the MBR. It

was rigorously rinsed with tap water to remove the attached

cake layer followed by backwashing with 0.05% sodium

hypochlorite solution for 2 h to further remove the foulants

adsorbed within membrane pores. The membrane module

was thoroughly cleaned again with tap water before it was

mounted back in the MBR. Since there is no significant

irreversible fouling observed, the same membrane was used

during the steady-state at all investigated SRTs for a fair

comparison.

2.3. SMP fouling test

The fouling potential of SMP at the investigated SRT was

examined for both supernatant and effluent at the end of

each experimental phase. The fouling test was conducted in a

stirred-cell system (Model 8200, Amicon, Beverly, MA, USA) in

connection with an external reservoir. A plane membrane

made of the same material as that used in the MBR was

employed in fouling test. The sample volumes used

for fouling test were 4 and 5 L for supernatant and effluent,

respectively, and the filtration time was 25 min. TMP

was maintained constant at 51.7 kPa (7.5 Psi) and stirring

speed was set at 180 rpm. A top-loading digital mass balance

(PG8001-S, Mettler Toledo, Greifensee, Switzerland) was

used to measure membrane filtrate. Based on the experi-

mental data, the fouling potential of SMP can be calculated

according to the normalization method developed by Song

et al. (2004).

2.4. Analytical methods

COD, NH4+–N, mixed liquor volatile suspended solids/sus-

pended solids (MLVSS/SS), and specific oxygen uptake rate

(SOUR) were measured in accordance with the Standard

Methods (APHA-AWWA-WEF, 1998). Supernatant samples

were obtained by centrifuging MBR mixed liquor at

10,000 rpm (11,000g) for 10 min at 4 1C and then filtering

through 0.45mm membranes. Concentrations of DOC, which

is basically SMP, in supernatants and in effluents were

determined by 1010 Total Organic Carbon Analyzer (O. I.

Analytical, College Station, TX, USA). Ultraviolet absorbance

at 254 nm (UVA254) of SMP was measured using DR/4000U

Spectrophotometer (HACH, Loveland, CO, USA). Specific

UVA254 (SUVA), indicating the aromaticity of SMP, was

calculated as the ratio of UVA254 to DOC.

The phenol–sulfuric acid method (Dubois et al., 1956) was

used to measure the content of carbohydrate in SMP with

glucose as the standard reference, whereas the modified

Lowry method (Lowry et al., 1951; Hartree, 1972) was used for

protein determination with bovine serum albumin (BSA) as

the standard reference. Apparent molecular weight distribu-

tion of SMP was determined by ultrafiltration fractionation

method using YM series ultrafiltration membranes with

nominal molecular weight cut-offs of 3, 10 and 30 kDa

(Millipore, Bedford, MA, USA).

The hydrophobic/hydrophilic and charge properties of SMP

were investigated using column chromatographic method as

described by Namour and Muller (1998). SMP can be fractio-

nated into four more homogeneous components, namely,

hydrophobic aquatic humic substances (AHS), hydrophilic

bases (HiB), hydrophilic acids (HiA), and hydrophilic neutrals

(HiN). The fractionation was performed using borosilicate

glass chromatography columns (006-CC-15-15-FF, Omnifit,

Cambridge, UK) with a series of resin adsorbents, including

non-ionic DAX-8 resin (Supelco, Bellefonte, PA, USA), AG MP-

50 cation exchange resin (Bio-Rad, Hercules, CA, USA) and

IRA-96 anion exchange resin (Rohm and Haas, Philadelphia,

PA, USA).

3. Results and discussion

The data presented here were based on the measurements

conducted in each experimental phase after the MBR reached

steady-state. The steady-state, herein, referred to the experi-

mental period approximately after two SRT when the

concentrations of both activated sludge and supernatant

DOC were generally stable. The error bars in all figures

indicate the sample standard deviations determined from

replicate measurements.

3.1. Overall MBR performance at different SRTs

The general performance of the MBR in terms of COD and

NH4+–N removal at different SRTs is summarized in Table 2.

The COD removal efficiencies were excellent and stable with

an average over 95% at all investigated SRTs, which proves the

substantial capacities of MBRs in wastewater treatment. With

respect to NH4+–N, average removal efficiencies were main-

ARTICLE IN PRESS

WAT E R R E S E A R C H 4 1 ( 2 0 0 7 ) 9 5 – 1 0 198

tained over 90% even at short SRTs. The high rate of

nitrification achieved by the MBR can be attributed to the

effective membrane retention of slow-growing nitrifying

microorganisms, which cannot be fulfilled by gravity clarifi-

cation in conventional biological treatment systems (Chang

et al., 2002).

Table 3 shows sludge concentrations and properties in the

MBR at different SRTs. It can be seen that, as SRT shortened,

the average MLSS concentration decreased accordingly from

7.82 g/L at SRT of 40 days to 3.07 g/L at SRT of 10 days.

However, the ratios of VSS/SS were very high with average

values over 96% and almost independent of SRT. This

indicates no considerable accumulation of inorganic matter

in the MBR. On the other hand, it was noted that the

metabolic activity of sludge, characterized by SOUR, slightly

decreased as SRT lengthened. This can be attributed to the

increase of inert biomass (i.e., metabolic products mainly

form endogenous respiration) at long SRTs and possibly to the

potential inhibition of SMP (Huang et al., 2000). Nevertheless,

as shown in Table 2, the reduction of specific respiration rates

of activated sludge had no significant effect on the general

performance of the MBR.

3.2. Concentration of SMP at different SRTs

Fig. 1 shows total concentrations of SMP, indicated by DOC, in

supernatants and in effluents at different SRTs. It was noted

that concentrations of SMP in supernatants significantly

increased as SRT shortened, implying that the potential effect

of SMP on system performances (e.g., membrane fouling)

might be more striking at short SRTs. In comparison,

Table 2 – COD and NH4+–N removal efficiencies of MBR at

different SRTsa

SRT (days) 10 20 40

COD Removal (%) 95.271.5 96.171.3 96.571.2

NH4+–N Removal (%) 90.273.3 92.172.6 94.371.9

a Sample mean 7 standard deviation, number of measurements:

n ¼ 24 (COD); n ¼ 25 (NH4+�N).

Table 3 – Biomass concentration and metabolic activity inMBR at different SRTsa

Parameters SRT (days)

10 20 40

MLSS (g/L) 3.0771.28 4.9871.16 7.8271.22

VSS/SS (%) 97.172.3 97.271.9 96.373.2

SOUR (mg O2/g VSS h) 13.9572.05 11.2371.58 9.5871.57

a Sample mean7standard deviation, number of measurements:

n ¼ 25 (MLSS and VSS/SS); n ¼ 18 (SOUR).

concentrations of SMP in effluents were relatively stable with

merely slight increases at short SRTs. Furthermore, it was

found that concentrations of SMP in supernatants were

always higher than those in effluents. This means that the

membrane serves as a selective barrier for a certain portion of

SMP, resulting in their accumulation inside the MBR.

SMP accumulation in an MBR is, indeed, a complicated

phenomenon, heavily dependent on the characteristics of

SMP and the properties of membranes. It was noted that

accumulation of SMP was more salient at short SRTs. Since

the same type of membrane was used over the entire

experimental period, the extent of SMP accumulation was

primarily determined by the characteristics of SMP. It appears

that SMP generated at short SRTs are more prone to

accumulate in MBRs, which leads to higher SMP concentra-

tions. In the following sections, compositions and character-

istics of SMP both in supernatants and in effluents were

measured and compared at different SRTs to better under-

stand the behaviors of SMP in MBR operation.

3.3. Composition of SMP at different SRTs

It is well known that SMP represent a myriad of structurally

complex organics with distinctly different characteristics. In

addition to measuring the gross concentration of SMP, the

concentrations of carbohydrate and protein, the known

components of SMP, were examined in order to get more

detailed insights into the composition of SMP at different

SRTs. As presented in Table 4, concentrations of both

carbohydrate and protein increased as SRT shortened, which

corresponded well to the variation of total SMP concentration.

It can be, therefore, inferred that the proportions of both

carbohydrate and protein in total SMP maintain approxi-

mately the same at different SRTs. Furthermore, similar to

the case of total SMP concentration, concentrations of both

carbohydrate and protein in supernatants were found to be

always higher than those in effluents. This indicates that

carbohydrates and proteins are the components of SMP

accumulating in the MBR.

On the other hand, the aromaticity of SMP in supernatants

and in effluents was also measured at different SRTs. The

0.0

5.0

10.0

15.0

20.0

10 20 40

SRT (days)

DO

C (

mg/

L)

Supernatant Effluent

Fig. 1 – Total concentrations of SMP in supernatants and in

effluents at different SRTs (number of measurements:

n ¼ 26).

ARTICLE IN PRESS

Table 4 – Carbohydrate and protein concentrations and SUVA values of SMP at different SRTsa

SRT (days) Carbohydrate (mg/L glucose-equivalent) Protein (mg/L BSA-equivalent) SUVA (L/mg DOC m)

Supernatant Effluent Supernatant Effluent Supernatant Effluent

10 12.2974.31 5.1571.75 9.1372.55 4.2571.23 1.9370.29 2.4370.34

20 10.1473.04 4.9371.87 7.1871.94 3.6471.09 2.5870.44 3.0670.37

40 8.0372.25 4.8671.79 5.6271.68 3.1270.95 2.8270.42 3.4770.36

a Sample mean7standard deviation, number of measurements: n ¼ 25 (carbohydrate and protein); n ¼ 26 (SUVA).

0

20

40

60

80

<3k 3-10k 10-30k >30k

Molecular weight (Da)

Per

cent

age

(%)

SRT 10d SRT 20d

SRT 40d

SMP in Supernatants

0

20

40

60

80

<3k 3-10k 10-30k >30k

Molecular weight (Da)

Per

cent

age

(%)

SRT 10d SRT 20d

SRT 40d

SMP in Effluents

(a)

(b)

Fig. 2 – Apparent molecular weight distributions of SMP at

different SRTs: (a) apparent molecular weight distributions

of SMP in supernatants; (b) apparent molecular weight

distributions of SMP in effluents (number of measurements:

n ¼ 15).

WAT E R R ES E A R C H 41 (2007) 95– 101 99

results are summarized in Table 4. It was interestingly noted

that the SUVA value decreased as SRT shortened, though the

total SMP concentrations were higher at short SRTs. Lee et al.

(2003) also observed the decrease of SUVA value of MBR

supernatant as SRT shortened from 40 to 20 days. This

implies that the SMP generated at short SRTs contain smaller

percentage of aromatic compounds. It appears that produc-

tion of aromatic SMP are more favored at long SRTs where the

food-to-microorganism ratio is low. Furthermore, it was

found that SMP in effluents exhibited higher SUVA values

than those in supernatants. This means that the percentage

of aromatic compounds in total SMP increased after passing

through the membrane. It is therefore inferred that, unlike

carbohydrates and proteins, aromatic SMP seem much less

prone to accumulate in the MBR.

3.4. Apparent molecular weight distributions of SMP atdifferent SRTs

Fig. 2 shows the apparent molecular weight distributions of

SMP in supernatants and in effluents at different SRTs. It can

be seen that SMP in the MBR had a broad spectrum of

molecular weight. The majority of SMP, accounting for around

57%, had molecular weight of smaller than 3 kDa, whereas the

components with large molecule weight (430 kDa) formed

the second largest fraction, constituting 23–32% of SMP. Each

of the two fractions with molecule weight in the range

between 3 and 30 kDa, however, only represented a very small

amount of SMP. Lee et al. (2003) studied the apparent

molecular weight distributions of MBR supernatants at

different SRTs and observed a smaller fraction of small

molecules but larger fractions of intermediate and large

molecules. This may be attributed to the different methods

used for supernatant separation. Since further purification

with 0.45mm membranes was not conducted, the supernatant

samples in the previous study may contain some large

components in addition to SMP.

It was noted that apparent molecular weight distributions

of SMP were quite similar at each SRT, even though the

concentrations of SMP were significantly different. The

results are somewhat inconsistent with those observed in

conventional biological treatment systems where apparent

molecular weight distribution of SMP has been found to be

greatly affected by SRT with large molecular weight compo-

nents become more evident at long SRTs (Barker and Stuckey,

1999). It is clear that the findings with respect to SMP in

conventional biological treatment systems are not directly

applicable to the case of MBRs. In addition, it was found that

apparent molecular weight distributions of SMP in super-

natants and in effluents were almost identical at all investi-

gated SRTs, indicating that membrane sieving may not work

for most SMP. It is therefore inferred that SMP accumulated in

the MBR based not mainly on their molecular size but on

other characteristics.

3.5. Hydrophobic/hydrophilic and charge properties ofSMP at different SRTs

Apart from molecular size, the hydrophobic/hydrophilic and

charge properties of SMP are of particular interest in studying

the fouling potential and accumulation of SMP in MBRs. It has

ARTICLE IN PRESS

WAT E R R E S E A R C H 4 1 ( 2 0 0 7 ) 9 5 – 1 0 1100

been well accepted that hydrophobic/hydrophilic and charge

properties of soluble organic matter have great effects on

their interactions with membranes (Carroll et al., 2000; Fan

et al., 2001). The SMP fractionation results according to these

characteristics are shown in Fig. 3. It can be seen that

hydrophobic AHS were the most abundant fraction of SMP,

though their proportion significantly varied with SRT. This

implies that SMP in MBRs are mainly composed of hydro-

phobic components, probably humic and fulvic acids. In

addition, it was noted that the proportion of AHS in total SMP

gradually increased as SRT lengthened, suggesting that SMP

generated at long SRTs tend to be more hydrophobic.

Typically, biomass associated products are more predominant

in total SMP at long SRTs as a result of the elevated biomass

concentration (Namkung and Rittmann, 1986; Barker and

Stuckey, 1999). It is therefore inferred that biomass associated

products may contain relatively large proportion of hydro-

phobic components compared to utilization associated pro-

ducts. Moreover, it was noteworthy that proportions of AHS in

supernatants were always lower than those in effluents.

Apparently, hydrophobicity has significant effect on SMP

accumulation in MBRs.

The distributions of hydrophilic components were quite

complex. As shown in Fig. 3a, neutral components constitute

the major fraction of hydrophilic SMP in the MBR, especially

at short SRTs. However, the proportion of HiN in total SMP

0

20

40

60

80

100

10 20 40SRT (days)

Per

cent

age

(%)

AHS HiA HiB HiNSMP in Supernatants

0

20

40

60

80

100

10 20 40

SRT (days)

Per

cent

age

(%)

AHS HiA HiB HiNSMP in Effluents

(b)

(a)

Fig. 3 – Hydrophobic/hydrophilic and charge properties of

SMP at different SRTs: (a) hydrophobic/hydrophilic and

charge properties of SMP in supernatants; (b) hydrophobic/

hydrophilic and charge properties of SMP in effluents

(number of measurements: n ¼ 13).

decreased significantly as SRT lengthened. In contrast,

proportions of HiA and HiB were relatively stable and

independent of SRT. Since most MBRs are operated at long

SRTs in practice, the distributions of hydrophilic components

at SRTs of 20 and 40 days may be closer to the real case. On

the other hand, it was noted that the proportion of HiN

considerably reduced after passing through the membrane.

This implies that neutral components are more prone to

accumulate in the MBR than charged components, though

they are all hydrophilic in nature. It should be pointed out

that the amount and nature of SMP in supernatants and in

effluents are affected not only by operating conditions like

SRT, but also by the properties of membranes. Future research

is therefore needed to explore the effects of membrane

properties on the behaviors of SMP in MBR operation.

3.6. Fouling potential of SMP at different SRTs

The fouling potentials of SMP at different SRTs were

examined at an equivalent DOC concentration of 5 mg/L to

eliminate the concentration effect on membrane fouling. The

results are presented in Fig. 4. It can be seen that the fouling

potential of SMP considerably increased as SRT shortened.

The differences in fouling potential are supposed to originate

from the different characteristics of SMP at different SRTs.

Although SMP generated at different SRTs had similar

apparent molecular weight distributions, it was noted that

the hydrophobic/hydrophilic and charge properties of SMP

varied significantly with SRT. In particular, SMP generated at

short SRTs were found to have high proportions of HiN. It is

therefore inferred that HiN are most likely the key foulants of

SMP responsible for high fouling potentials of supernatant

SMP at short SRTs. It is noteworthy that, in the real case,

fouling potentials of SMP at short SRTs would be even higher

than those at long SRTs due to their high concentration.

On the other hand, fouling potentials of SMP in effluents

were found to be lower than those in supernatants to a

certain extent, especially at short SRTs. This indicates that

organic compounds prone to accumulate in the MBR are the

major components of SMP responsible for membrane fouling.

It is inferred that SMP accumulated in the MBR (e.g.,

carbohydrate and protein) have relatively high fouling poten-

tial. The dominance of carbohydrates and proteins in

5.0E+04

1.5E+05

2.5E+05

3.5E+05

4.5E+05

5.5E+05

6.5E+05

10 20 40

SRT (days)

Fou

ling

pote

ntia

l (P

a s/

m2 )

Supernatant Effluent

Fig. 4 – Fouling potential of SMP in supernatants and in

effluents at different SRTs.

ARTICLE IN PRESS

WAT E R R ES E A R C H 41 (2007) 95– 101 101

membrane foulants were also reported in a recent pilot-scale

MBR study (Kimura et al., 2005). It is therefore suggested that

MBRs should be operated at long SRTs to minimize the

amount of carbohydrate and protein for SMP fouling control.

4. Conclusions

The research presented here focused on the behaviors,

characteristics, and fouling potential of SMP in MBR operation

at different SRTs. The following specific conclusions were

drawn:

(1)

Accumulation of SMP in the MBR was more pronounced at

short SRTs. Carbohydrates and proteins appeared to be

the components of SMP prone to accumulate in the MBR

compared with aromatic compounds.

(2)

Apparent molecular weight distributions of SMP were

similar at different SRTs and almost identical in super-

natants and in effluents. The results indicate that

membrane sieving may not work for most SMP.

(3)

The majority of SMP was found to be hydrophobic AHS,

whose proportion in total SMP gradually increased as SRT

lengthened. Moreover, the proportions of AHS in super-

natants were found to be always lower than those in

effluents.

(4)

The fouling potential of SMP considerably increased as

SRT shortened. The hydrophilic neutrals (e.g., carbohy-

drates) were most likely the main foulants of SMP.

R E F E R E N C E S

APHA-AWWA-WEF, 1998. Standard Methods for Examination ofWater and Wastewater, 20th ed. APHA-AWWA-WEF, Wa-shington, DC, USA.

Barker, D.J., Stuckey, D.C., 1999. A review of soluble microbialproducts (SMP) in wastewater treatment systems. Water Res.33 (14), 3063–3082.

Bouhabila, E., Aim, R., Buisson, H., 2001. Fouling characterisationin membrane bioreactors. Sep. Purif. Technol. 22–23 (1–3),123–132.

Brindle, K., Stephenson, T., 1996. The application of membranebiological reactors for the treatment of wastewaters. Biotech-nol. Bioeng. 49 (6), 601–610.

Carroll, T., King, S., Gray, S.R., Bolto, B.A., Booker, N.A., 2000. Thefouling of microfiltration membranes by NOM after coagula-tion treatment. Water Res. 34 (11), 2861–2868.

Chang, I.-S., Clech, P.L., Jefferson, B., Judd, S., 2002. Membranefouling in membrane bioreactors for wastewater treatment. J.Environ. Eng.—ASCE 128 (11), 1018–1029.

DeWalle, F.B., Chian, E.S.K., 1974. Kinetics of formation of humicsubstances in activated sludge systems and their effect onflocculation. Biotechnol. Bioeng. 16 (6), 739–755.

Dubois, M., Gilles, K.A., Hamilton, J.K., Rebers, P.A., Smith, F., 1956.Colorimetric method for determination of sugars and relatedsubstances. Anal. Chem. 28 (3), 350–356.

Fan, L., Harris, J.L., Roddick, F.A., Booker, N.A., 2001. Influence ofthe characteristics of natural organic matter on the fouling ofmicrofiltration membranes. Water Res. 35 (18), 4455–4463.

Hartree, E.F., 1972. Determination of protein: a modification of theLowry method that gives a linear photometric response. Anal.Biochem. 48 (2), 422–427.

Hejzlar, J., Chudoba, J., 1986a. Microbial polymers in the aquaticenvironment: I. Production by activated sludge microorgan-isms under different conditions. Water Res. 20 (10), 1209–1216.

Hejzlar, J., Chudoba, J., 1986b. Microbial polymers in the aquaticenvironment: II. Isolation from biologically non-purified andpurified municipal waste water analysis. Water Res. 20 (10),1217–1221.

Huang, X., Liu, R., Qian, Y., 2000. Behavior of soluble microbialproducts in a membrane bioreactor. Process Biochem. 36 (5),401–406.

Kimura, K., Yamato, N., Yamamura, H., Watanabe, Y., 2005.Membrane fouling in pilot-scale membrane bioreactors(MBRs) treating municipal wastewater. Environ. Sci. Technol.39 (16), 6293–6299.

Lee, W., Kang, S., Shin, H., 2003. Sludge characteristics and theircontribution to microfiltration in submerged membranebioreactors. J. Membr. Sci. 216 (1–2), 217–227.

Lowry, O.H., Rosebrough, N.J., Farr, A.L., Randall, R.J., 1951. Proteinmeasurement with the Folin phenol reagent. J. Biol. Chem. 193(1), 265–275.

Namkung, E., Rittmann, B.E., 1986. Soluble microbial products(SMP) formation kinetics by biofilms. Water Res. 20 (6),795–806.

Namour, P., Muller, M.C., 1998. Fractionation of organic matterfrom wastewater treatment plants before and after a 21-daybiodegradability test: physical–chemical method for mea-surement of the refractory part of effluents. Water Res. 32 (7),2224–2231.

Ng, H.Y., Hermanowicz, S.W., 2005. Membrane bioreactor opera-tion at short solids retention times: performance and biomasscharacteristics. Water Res. 39 (6), 981–992.

Onuki, M., Satoh, H., Mino, T., 2002. Analysis of microbialcommunity that performs enhanced biological phosphorusremoval in activated sludge fed with acetate. Water Sci.Technol. 46 (1–2), 145–153.

Shin, H.-S., Kang, S.-T., 2003. Characteristics and fates of solublemicrobial products in ceramic membrane bioreactor at varioussludge retention times. Water Res. 37 (1), 121–127.

Slavica, A., Santek, B., Novak, S., Maric, V., 2004. Microbial acetateoxidation in horizontal rotating tubular bioreactor. J. Biosci. 29(2), 169–177.

Song, L., Chen, K.L., Ong, S.L., Ng, W.J., 2004. A new normalizationmethod for determination of colloidal fouling potential inmembrane processes. J. Colloid Interface Sci. 271 (2), 426–433.

Wisniewski, C., Grasmick, A., 1998. Floc size distribution in amembrane bioreactor and consequences for membrane foul-ing. Colloids Surf. A—Pysicochem. Eng. Asp. 138 (2–3), 403–411.