Critical flux and chemical cleaning-in-place during the long-term operation of a pilot-scale submerged membrane bioreactor for municipal wastewater treatment

wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 8 6 3e8 7 1

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

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

Critical flux and chemical cleaning-in-place during the long-term operation of a pilot-scale submerged membranebioreactor for municipal wastewater treatment

Chun-Hai Wei a,b, Xia Huang a,*, Roger Ben Aim c, Kazuo Yamamoto d, Gary Amy b

a State Key Joint Laboratory of Environment Simulation and Pollution Control, Department of Environmental Science and Engineering,

Tsinghua University, Beijing 100084, ChinabWater Desalination and Reuse Center, 4700 King Abdullah University of Science and Technology, Thuwal 23955-6900, Saudi ArabiacUniversite de Toulouse; INSA, LISBP, 135 Avenue de Rangueil, F-31077 Toulouse, FrancedEnvironmental Science Center, The University of Tokyo, 7-3-1 Hongo, Bunkyo-Ku, Tokyo 113-0033, Japan

a r t i c l e i n f o

Article history:

Received 27 June 2010

Received in revised form

14 September 2010

Accepted 15 September 2010

Available online 1 October 2010

Keywords:

Chemical cleaning-in-place

Membrane fouling

Municipal wastewater

Critical flux

Submerged membrane bioreactor

* Corresponding author. Tel.: þ86 10 6277232E-mail address: [email protected]

0043-1354/$ e see front matter ª 2010 Elsevdoi:10.1016/j.watres.2010.09.021

a b s t r a c t

The critical flux and chemical cleaning-in-place (CIP) in a long-term operation of a pilot-scale

submerged membrane bioreactor for municipal wastewater treatment were investigated.

Steady filtration under high flux (30 L/(m2 h)) was successfully achieved due to effective

membrane fouling control by sub-critical flux operation and chemical CIP with sodium

hypochlorite (NaClO) in both trans-membrane pressure (TMP) controlling mode (cleaning

with high concentration NaClO of 2000e3000 mg/L in terms of effective chorine was per-

formed when TMP rose to 15 kPa) and time controlling mode (cleanings were performed

weekly and monthly respectively with low concentration NaClO (500e1000 mg/L) and high

concentration NaClO (3000mg/L)). Microscopic analysis onmembrane fibers before and after

highconcentrationNaClOwasalsoconducted. Imagesof scanning electronmicroscopy (SEM)

andatomic forcemicroscopy (AFM) showed thatNaClOCIP could effectively remove gel layer,

thedominant foulingunder sub-critical fluxoperation. Porositymeasurements indicated that

NaClO CIP could partially remove pore blockage fouling. The analyses from fourier transform

infrared spectrometry (FTIR) with attenuated total reflectance accessory (ATR) and energy

dispersive spectrometer (EDS) demonstrated that protein-like macromolecular organics and

inorganics were the important components of the fouling layer. The analysis of effluent

quality before and after NaClO CIP showed no obvious effect on effluent quality.

ª 2010 Elsevier Ltd. All rights reserved.

1. Introduction 2006; Itokawa et al., 2008; Huang et al., 2010). However

The membrane bioreactor (MBR), especially the submerged

membrane bioreactor (SMBR), has been extensively investi-

gated and applied for municipal and industrial wastewater

treatment and reuse worldwide in recent years due to its

advantages (i.e., excellent effluent, small footprint and less

excess sludge) over conventional activated sludge (Yang et al.,

4; fax: þ86 10 62771472.(X. Huang).ier Ltd. All rights reserved

membrane fouling, the major factor limiting the wide appli-

cation of SMBR, reduces permeate production and increases

operational cost in long-termoperation. Thus themechanisms

of membrane fouling and control strategies have become the

focus areas in SMBR studies. In general,membrane fouling can

be classified as pore blockage, gel layer and cake layer accord-

ing to fouling formation mechanisms. Cake layer fouling,

.

wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 8 6 3e8 7 1864

mainlycausedbysludgeflocs, canusuallybepreventedbysub-

critical flux operation and removed by enhanced air scouring

(Bouhabila et al., 1998; Changand Judd, 2002; Pollice et al., 2005;

Wang et al., 2008) in SMBR. From the viewpoint of a real

application, onehypothesis is generallyaccepted that forSMBR

under steady conditions (including membrane, sludge char-

acteristics and operational parameters), there is a critical flux,

above which membrane permeability will deteriorate too

rapidly to realize steady long-term filtration. However the

common method of “flux stepwise increment” for critical flux

measurement is based on a short-term constant-flux filtration

test in terms of minutes or hours (Bouhabila et al., 1998; Cho

and Fane, 2002; Yu et al., 2003; Le Clech et al., 2003; Pollice

et al., 2005). For real SMBR operation in terms of days or

months, its applicability should be further investigated.

Gel layer and pore blockage fouling, mainly caused by

colloidal and soluble organic fractions (such as extracellular

polymeric substances and soluble microbial products), and

inorganic substances (such as calcium carbonate), are inevi-

table in long-term operation of SMBR even under sub-critical

flux operation because there are significant amounts of

potential foulants contained in activated sludge mixed liquor

(Cho and Fane, 2002; Guglielmi et al., 2007; Wang et al., 2008).

Therefore cleaning techniques are necessary for eliminating

these inevitable foulants and achieving long-term steady

operation. Physical cleaning techniques (i.e. water and/or air

scouring/backwashing, ultrasonic cleaning and mechanical

scouring) have been proven effective only for removing cake

layer fouling (Changetal., 2002;LimandBai, 2003;Adhametal.,

2004; Huang et al., 2008). Chemical cleaning techniques with

different reagents, such as an oxidant, acid, alkali and metal-

chelator, appear to bemore effective for removinggel layer and

pore blockage fouling (Liu et al., 2000; Lim and Bai, 2003; Xing

et al., 2003; Adham et al., 2004; You et al., 2006; Brepols et al.,

2008; Grelot et al., 2010). But these results on chemical clean-

ing were mainly obtained from off-line cleaning (i.e.

membranemodules are taken out of bioreactor and immersed

in a tank of cleaning reagent, or membrane modules are

immersed directly in membrane tank full of cleaning agent

after draining off sludge). Compared with off-line chemical

cleaning, chemical cleaning-in-place (CIP) (i.e. cleaning

reagent is injected into the membrane in reverse to normal

filtration while membrane modules are still submerged in

bioreactor) is simpler and cheaper. Thus the other hypothesis,

that membrane fouling under sub-critical flux can be

controlled by chemical cleaning-in-place and thus long-term

steady filtration can be achieved, appears to be feasible.

Level sensor

Influent

Raw wastewater

Waste sludge

Feed pump Bioreactor

Membrane module

Fig. 1 e Diagram of the pilo

Although chemical cleaning is used in most full-scale SMBR

plants (Fatone et al., 2007; Lyko et al., 2008; Itokawa et al., 2008;

Kraume et al., 2009), little information about chemical CIP is

available in literature, especially in terms of performance and

mechanisms in long-term operation.

Inourpreviousstudy (Weietal., 2006), steadyfiltrationunder

high flux of 30 L/(m2 h) for 190 dwas successfully achievedwith

sub-critical flux operation and chemical CIP in a pilot-scale

SMBR for municipal wastewater treatment. In order to address

the former two hypotheses, during the following continuous

operation of this SMBR (up to 750 d), different fluxes were

adopted to check the applicability of critical flux derived from

short-term constant-flux filtration tests (Wei et al., 2006). At the

same time, chemical CIP, including its macroscopic perfor-

mance formembrane fouling control with different operational

parameters (cleaning reagent and cleaning frequency) and

microscopic mechanism on removing membrane fouling, was

investigated in detail.

2. Materials and methods

2.1. Experimental set-up

A pilot-scale SMBR with an effective volume of 2.14 m3 was

used in this study. (Fig. 1) A hollow fiber membrane module

made of polyvinylidene fluoride (PVDF) (Mitsubishi Rayon Co.

Ltd, Japan) with a nominal pore size of 0.4 mm and filtration

area of 29m2 was submerged in the riser of the bioreactor. The

membrane effluent was intermittently (13 min on, 2 min off)

extracted with a suction pump at a constant-flux mode. The

suction pump could also function as the cleaning pump during

chemical CIP. Municipal wastewater after primary sedimen-

tation from Qinghe wastewater treatment plant located in

Beijing was pumped into the bioreactor automatically by level

sensor control. Trans-membrane pressure (TMP) was moni-

tored through two U-type mercury manometers as an indi-

cator of membrane fouling evolution. A perforated pipe was

located below the membrane module for aeration, both for

oxygen supply and membrane fouling control through cross-

flow scouring.

2.2. Method of chemical CIP

Sodium hypochlorite (NaClO) with effective chlorine

concentration of 500e3000 mg/L was used as the dominant

cleaning reagent due to its widespread application in MBR for

Timer

Effluent

Cleaning reagent

Suction or backwashing pump Air blower

Flow meter

Valve

U-type mercury

manometer

t-scale SMBR system.

wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 8 6 3e8 7 1 865

membrane cleaning. Hydrochloric acid (HCl, pH ¼ 1) and

sodium hydroxide (NaOH, pH ¼ 12) were used as supple-

mentary cleaning reagents. All doses were applied at 2 L per

square meter membrane filtration area. This membrane is

tolerant to all cleaning agents used in this study according to

the manual provided by manufacturer. Two types of cleaning

modes with primary emphasis on NaClO were mainly

investigated. One was TMP controlling mode, i.e., cleaning

with high concentration NaClO (2000e3000 mg/L) was per-

formed when TMP rose to a pre-determined value (15 kPa).

The other was time controlling mode, i.e., cleanings were

weekly and monthly performed, respectively, with low

concentration NaClO (500e1000 mg/L) and high concentra-

tion NaClO (3000 mg/L). HCl and NaOH were aperiodically

used for the supplementary cleaning. TMP change before and

after chemical CIP was regarded as the macroscopic index for

chemical CIP efficiency. Each cleaning process consisted of

several steps including stopping influent and effluent, stop-

ping aeration, switching corresponding valves, pumping

cleaning reagent in 30 min (flow rate of 116 L/h and corre-

sponding flux of 4 L/(m2 h)) into membrane fibers in reverse

direction of normal filtration, idle for 30 min and 90 min for

low concentration NaClO cleaning and HCl/NaOH cleaning

and high concentration NaClO cleaning respectively,

switching valves, starting aeration, and starting influent and

effluent. The whole cleaning process required about 60 min

for low concentration NaClO cleaning and HCl/NaOH clean-

ing and about 120 min for high concentration NaClO cleaning

respectively.

2.3. Operational conditions

From our previous study (Wei et al., 2006), critical flux zone of

this pilot-scale SMBR was about 30e35 L/(m2 h) under a mixed

liquor suspended solid concentration (MLSS) of lower than

13 g/L and air flow of more than 13.7 m3/h, according to the

adoptedfluxstepwise incrementmethodwithafluxstepof 5 L/

(m2h) and thecriterion for identifying the critical fluxzonewas

when the TMP increment during 120 min filtration time

exceeded 133 Pa (i.e., the precision of a U-type mercury

Table 1 e Operational parameters in all runs.

Run No. Time(d)

Flux(L/(m2 h))

Ch

1 1e200 27.6e30.3 TMP controlli

2 201e265 29.0e30.3 Time controll

3 266e406 29.7e35.9 Combination

controlling m

4 407e484 34.9e42.8 Time controll

5 485e584 23.5e38.1 Enhanced cle

concentratio

6 585e750 29.0e36.6 Time controll

cleaning with

a Time controlling mode (weekly cleaning with low concentration NaCl

NaClO of 2000e3000mg/L) was conductedwhen TMPwas less than 20 kPa.

2000e3000 mg/L, was conducted once TMP exceeded 20 kPa.

manometer). Although steady filtration was already achieved

under sub-critical flux (30 L/(m2 h)) operation (Wei et al., 2006),

higher flux was attempted to investigate further the accuracy

of critical flux in the following experiment because the criteria

for identifying critical flux in this studywasmore conservative

than in the literature (Pollice et al., 2005). Table 1 lists the

operational parameters in all runs. During most of the time in

the long-term (up to 750 d) operation, MLSS was 8e12 g/L by

adjusting sludgeretention time (SRT) (10e15d)andairflowwas

14e20 m3/h. Therefore, the critical flux zone during the long-

term operation could be still regarded as 30e35 L/(m2 h) based

on the assumption of constant membrane characteristics.

2.4. Analytical methods

In order to characterize themicroscopic effects of chemical CIP

on membrane fouling, several analytical techniques, such as

scanning electron microscopy (SEM, FEI QUANTA 200, FEI

Company, USA), atomic force microscopy (AFM, SPA-300 HV,

Seiko Instrument Inc., Japan), mercury porosimeter (AUTO-

PORE II 9220, Micromeritics, USA), fourier transform infrared

spectrometry with attenuated total reflectance accessory

(ATR-FTIR, Nicolet 560, Thermo Electron Corporation, USA)

and energy dispersive spectrometer (EDS) by field emission

gun-scanning electron microscopy (FEG-SEM, JSM 6301F, JEOL,

Japan), were employed in this study. This hollow fiber with

large wall thickness of about 0.9 mm consists of three layers e

outer active layer (200e250 mm), intermediate porous connec-

tion layer and inner support layer. The fibers were cut to

separate the flat-sheet active layer alone for the measure-

ments of AFM, mercury porosimeter, ATR-FTIR and EDS.

Before the analysis, membrane samples were dried under

vacuum and 50 �C for 8 h to eliminate the moisture. Chemical

oxygen demand (CODCr), biochemical oxygen demand (BOD5),

ammonia nitrogen (NH4þeN), total nitrogen (TN), total phos-

phorus (TP) and total Escherichia coli for effluent samples before

and after chemical CIP were monitored according to Standard

Methods (APHA-AWWA-WEF, 1995) to investigate the effect of

chemical CIP on effluent quality.

emical CIP Note

ng mode Sub-critical flux operation

ing mode Sub-critical flux operation

of TMP and time

odeaCritical flux operation

ing mode Super-critical flux operation

aning with high

n NaClO and acid/alkali

Variable flux operation for

restoring membrane permeability

ing mode and enhanced

acid/alkali

Critical flux operation

O of 500e1000 mg/L and monthly cleaning with high concentration

TMP controllingmode, i.e. cleaningwith high concentration NaClO of

Fig. 2 e TMP and flux changes of the pilot SMBR during the

long-term operation.

wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 8 6 3e8 7 1866

3. Results and discussion

3.1. Membrane performance under different fluxes andchemical CIP modes

Fig. 2 shows the evolution of TMP and flux in the long-term

operation. Table 2 lists the membrane performance in terms

of flux, TMP and its rising rate between two cleanings in all

runs. In Run 1 of 200 d, TMP and its rising rate were in the low

range of 4.6e17.6 kPa and 0.15e2.09 kPa/d respectively, indi-

cating steady filtration was successfully achieved in the long-

term operation. This depressed TMP evolution was mainly

attributed to both synergistic effectiveness of chemical CIP,

with a TMP controlling mode, and sub-critical flux operation.

Sub-critical flux operation could reduce the cake layer caused

by sludge flocs, resulting in slow TMP rise (such as the

average of 0.79 kPa/d in Run 1 in this study) with time (Pollice

et al., 2005). After every chemical CIP, TMP decreased almost

down to the original value (5e7 kPa), indicating that chemical

CIP could effectively remove the fouling in terms of

membrane pore blockage and gel layer caused by colloids and

soluble organic substances. Cleaning intervals were from 2 to

3 weeks to nearly 3 months due to fluctuation of sludge

characteristics. In Run 2, TMP and its rising rate were also in

the low range of 5.1e18.2 kPa and 0.50e1.48 kPa/d respec-

tively under effective membrane fouling control by sub-crit-

ical flux operation and chemical CIP, but with a time

controlling mode. Although TMP did not decrease to the

original value after weekly cleaning with low concentration

Table 2 e Membrane performance in all runs.

Run No. Time (d) Flux (L/(m2 h)) TMP (k

1 1e200 27.6e30.3 (30.1) 4.6e17

2 201e265 29.0e30.3 (30.0) 5.1e18.2

3 266e406 29.7e35.9 (34.2) 5.3e26.6

4 407e484 34.9e42.8 (40.1) 6.7e43.2

5 485e584 23.5e38.1 (30.4) 10.1e61.3

6 585e750 29.0e36.6 (33.3) 5.8e61.3

Values in parentheses are the average values.

NaClO and rose, to some extent, in one month, TMP

decreased to the original value after monthly cleaning with

high concentration NaClO. From the low average TMP and its

rising rate in Run 1 and Run 2 with the similar flux, steady

filtration in the long-term operation could be achieved under

sub-critical flux operation and chemical CIP (with TMP or

time controlling mode).

Comparedwith Run 1 and Run 2, TMP (5.3e26.6 kPa) and its

rising rate (0.39e3.05 kPa/d) between two cleanings in Run 3

were higher due to critical flux operation. Although filtration

stability was not as good as in Run 1 and Run 2, TMP could be

kept below 30 kPa in Run 3, showing a possible steady long-

term filtration under critical flux operation and chemical CIP

with combined TMP and time controlling modes. Especially

TMP increase in the second half of Run 3 with flux of

29.7e35.9 L/(m2 h) was comparable to that in initial Run 1 with

flux of 27.6e30.3 L/(m2 h), indicating the possible seasonal

effects on membrane fouling. Both initial Run 1 and the

second half of Run 3 were in summer with good sludge

filterability. High temperature not only decreased the sludge

viscosity but also affected microbial activity especially under

low SRT, thus resulting in less fouling than low temperature.

Similar results have been reported by several researchers (Al-

Halbouni et al., 2008; Lyko et al., 2008; Miyoshi et al., 2009). In

addition, the relative steady filtration in Run 3 also showed

that the criterion for identifying critical flux in this study was

a little conservative.

In Run 4, under super-critical flux operation and chemical

CIP with time controlling mode, both TMP and its rising rate

were in the higher range of 6.7e43.2 kPa and 0.61e4.69 kPa/

d compared to Run 1e3. Especially at the secondmonth of Run

4, TMP rising rate was higher than 3 kPa/d, showing charac-

teristics of cake layer fouling. After weekly cleaning with low

concentrationNaClO, virtually noTMPdecreasewas observed.

Although TMP could be depressed aftermonthly cleaningwith

highconcentrationNaClO,TMP roseup tohigher than40kPaat

the end of Run 4 and flux decreasedwith increased TMP due to

the limit of suction pump. This indicated that steady filtration

under super-critical flux operation could not be achieved even

adopting chemical CIP. Adham et al. (2004) also got similar

results from different flux operation of a pilot-scale US Filter

MBR for municipal wastewater treatment.

In Run 5, variable flux operation was conducted in order to

restore membrane permeability. Although enhanced chem-

ical CIP (NaClO and HCl or NaOH) was performed, low TMP

(less than 20 kPa) was only adequate under low flux (about

25 L/(m2 h)) due to accumulative fouling in Run 4. In addition,

Pa) TMP rising rate between two cleanings (kPa/d)

.6 (9.7) 0.15e2.09 (0.79)

(11.0) 0.50e1.48 (0.84)

(12.1) 0.39e3.05 (1.22)

(17.2) 0.61e4.69 (2.11)

(27.3) 0.99e9.95 (3.89)

(20.3) 0.30e5.26 (1.63)

Table 3 e Cleaning performance of different reagents.

Reagent TMP decrease aftercleaning (kPa)

NaClO (2000e3000 mg/L) 1.8e31.9 (12.7), n ¼ 31

NaClO (500e1000 mg/L) 1.4e13.0 (5.1), n ¼ 25

HCl (pH ¼ 1) 5.0e12.1 (9.8), n ¼ 4

NaOH (pH ¼ 12) 4.7e7.6 (6.2), n ¼ 2

HCl (pH ¼ 1) followed by

NaClO (3000 mg/L)

7.4e49.2 (22.4), n ¼ 3

Mixture of NaClO (3000 mg/L)

and NaOH (pH ¼ 12)

14.6e28.4 (21.7), n ¼ 3

Values in parentheses are the average values. n is the number of

cleaning.

wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 8 6 3e8 7 1 867

the relative poor filterability of activated sludge under winter

temperature (10e15 �C) in Run 5 might be another factor for

low permeability recovery (Al-Halbouni et al., 2008; Lyko et al.,

2008; Miyoshi et al., 2009).

InRun6undercritical fluxoperation,TMPshowedagradual

decrease followed by a gradual increase due to enhanced

chemical CIP and fluctuation of sludge characteristics. In the

second half, TMPwas obviously lower than in Run 4 and Run 5

and was close to that in Run 3, indicating membrane perme-

ability could be recovered from serious fouling, caused by

super-critical flux operation in Run 4, to some extent by

chemical CIP.

From all runs, long-term steady filtration was successfully

achieved under sub-critical flux operation (30 L/(m2 h)) and

chemical CIP with both cleaning modes. Critical flux operation

(30e35 L/(m2h)) appeared to be feasible for achieving long-term

steady filtration under chemical CIP with combined TMP and

time controlling mode because this combined cleaning mode

could provide the more intensive protection for membrane.

However super-critical flux operation (35e42 L/(m2 h)) was not

possible forachieving long-termsteadyfiltrationevenadopting

chemical CIP. The results from the long-term operation of the

pilot-scale SMBR in this study also demonstrated that the crit-

ical flux should be a key parameter for realizing long-term

steady filtration for real SMBR application and the method for

critical flux measurement through short-term constant-flux

filtration test used in this study was feasible for long-term

operation.

3.2. Chemical CIP performance of different cleaningreagents

Fig. 3 and Table 3 show chemical CIP performance of various

cleaning reagents in terms of TMP decrease after cleaning

under the same flux, i.e. the simple index for permeability

recovery. From Table 3, average TMP decrease after cleaning

for high concentration NaClO (2000e3000 mg/L), low concen-

tration NaClO (500e1000 mg/L), HCl (pH ¼ 1) and NaOH

(pH ¼ 12) was 12.7 kPa, 5.1 kPa, 9.8 kPa and 6.2 kPa respec-

tively, indicating high concentration NaClO was the best

among all of the reagents. Grelot et al. (2008) also found that

0

5

10

15

20

25

30

35

40

45

50

0 75 150 225 300 375 450 525 600 675 750Time (d)

ΔTM

P (k

Pa)

NaClO (2000-3000 mg/L)NaClO (500-1000 mg/L)HCl (pH=1)NaOH (pH=12)

HCl (pH=1) followed by NaClO (3000 mg/L)Mixture of NaClO (3000 mg/L) and NaOH (pH=12)

Fig. 3 e TMP decrease after chemical CIP with different

cleaning reagents.

high concentration NaClO (2000 mg/L) was better than other

chemicals (such as H2O2, NaOH, HCl, citric acid and enzymes)

used in their study. The recovery of membrane permeability

by NaClO cleaning has also been reported in many publica-

tions (Liu et al., 2000; Xing et al., 2003; Adham et al., 2004;

Trussell et al., 2005; Le-Clech et al., 2006; Fatone et al., 2007;

Qin et al., 2009). HCl and NaOH showed results similar to

that of low concentration NaClO. The combination of NaClO

and HCl or NaOH appeared to be better than NaClO only,

especially when serious fouling occurred. Similar results have

been reported by Xing et al. (2003), Adham et al. (2004), Zhang

et al. (2005), Brepols et al. (2008) and Matosic et al. (2009). In

addition, for the performance of each cleaning in terms of

TMP decrease, it depended not only on cleaning reagent but

also on fouling condition before cleaning. Under the same

cleaning reagent, generally the higher TMP before cleaning

was, the higher TMP decrease after cleaning was. Serious

fouling occurred at the second half of Run 4, Run 5 and initial

Run 6, was partially the reason for the large fluctuation of

cleaning performance. Finally, it should be noted that HCl,

NaOH, combined NaClO and HCl, and combined NaClO and

NaOH cleaning were performed only 4, 2, 3 and 3 times,

respectively, in this study. Therefore, further investigation is

necessary to optimize CIP strategies for MBR.

3.3. Changes of membrane surface morphology andproperties by chemical CIP

In order to characterize the microscopic changes of a mem-

brane by chemical CIP, membrane fibers before and after

cleaning with high concentration NaClO (3000 mg/L) on day

623 in Run 6 were cut off for various analyses including

membrane surface morphology, membrane material charac-

teristics and foulants. TMP values under a flux of 30 L/(m2 h)

before and after cleaning were 31 kPa and 20 kPa, respectively,

indicating the limited performance for fouling removal

because TMP for a newmembrane was only about 5 kPa under

the sameflux. This large difference ofmembrane permeability

was mainly caused by accumulative fouling in Run 4 under

super-critical flux operation.

3.3.1. Membrane surface morphologyFrom theappearance,newmembranefiberswerewhite, fouled

membrane fibers appeared dark and cleaned membrane fibers

Fig. 4 e SEM images of membrane fibers (30003) (A: new membrane; B: fouled membrane; C: cleaned membrane).

Fig. 5 e Porosity changes of new, fouled and cleaned

membrane fibers.

wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 8 6 3e8 7 1868

looked like mixed yellowewhitemore close to newmembrane

than fouled membrane, indicating the macroscopic effect of

removing membrane fouling by chemical CIP.

Fig. 4 shows the SEM images of membrane fibers. There

was a widespread gel layer and no obvious cake layer on the

fouled membrane surface, indicating that deposition of flocs

could be avoided by sub-critical flux operation. Gel layer dis-

appeared virtually after cleaning, indicating that chemical CIP

could remove gel layer fouling effectively. Compared with

a newmembrane, fouling (mainly pore blockage foulants) still

existed after cleaning. This explanation was supported by the

measurement results of mercury porosimeter in the next

section.

FromAFManalysis (imageno shown), for anewmembrane,

fouled membrane and cleaned membrane, Ra (the average

roughness) and RMS (the square mean roughness) of the AFM

images, were 37.39 nm, 126.4 nm, 83.74 nm and 49.69 nm,

156.8 nm, 113.9 nm, respectively. The sequence of both Ra and

RMS was fouled membrane > cleaned membrane > new

membrane, indicating that the height of fouling layer

decreased after cleaning. This was in agreement with the SEM

images.

3.3.2. Membrane material characteristicsFig. 5 shows the changes of porosity in terms of pore volume

per unitmass ofmembranematerial with pore size. Themean

pore diameter of new membrane was about 0.4 mm, in agree-

ment with the nominal pore size from membrane manufac-

turer. In comparison with a new membrane, the porosity of

a fouled membrane and a chemically cleaned membrane

decreased noticeably, and pore diameter distribution also

narrowed to some extent. This demonstrated the existence of

pore blockage fouling. Comparedwith a fouledmembrane, the

porosity increased slightly after cleaning, indicating that

chemical CIP could remove pore blockage fouling to a certain

extent. There was apparently no difference at pore size less

than 0.2 mmbetween new and fouled/cleanedmembrane from

Fig. 5. However pore blockage should also occur in smaller

pores because the potential pollutants with smaller size in

activated sludge could enter the smaller pore and block it.

Partial pore blockage might be the possible reason. Generally

membrane pore is not blocked completely. The partial blocked

pore can be regarded as a smaller pore during measurement.

Thus for fouled/cleaned membrane, smaller pores were really

blocked but the apparent porosity at smaller pores did not

decrease or even increased because partial blocked big pores

would be measured as small pores.

3.3.3. Analysis of membrane foulantsFig. 6 shows themeasuredATR-FTIR spectra of new, fouled and

cleanedmembrane fibers. Two characteristic peaks (1070/1180/

1280 cm�1, CeF str.; 1400 cm�1, CH2 def.) of membrane material

(e[eCF2eCH2e]en) were clearly seen for all membrane fibers.

Compared with the new membrane, the peaks of amide I/II

(1643 cm�1, C]O str.; 1546 cm�1, NeH def.) and the peak

(3300 cm�1, NeH str.) on the fouled membrane demonstrated

the existence of protein-like substances in the fouling layer. In

addition, the peak at 1000e1100 cm�1 was obviously broadened

and strengthened on the fouled membrane, indicating that

some foulants (possibly polysaccharide-like substances or

inorganics) had overlapping peaks with membrane material.

Kimura et al. (2005) also found that protein-like and poly-

saccharide-like substances were themain foulants in SMBR for

municipal wastewater treatment through FTIR analysis of off-

line alkali cleaning solution of the fouled membrane modules.

In comparison with the fouled membrane, the peak at

1000e1100 cm�1 was obviously narrowed and weakened and

the peaks of amide I/II were also weakened after cleaning,

Fig. 6 e ATR-FTIR spectrums of new, fouled and cleaned

membrane fibers.

Table 4 e Effluent quality before and after the CIP withNaClO (3000 mg/L) on day 471.

Item Background valuebefore cleaning

Initial valueafter cleaning

COD (mg/L) 40.2 42.3

BOD5 (mg/L) 1.0 1.4

NH4þeN (mg/L) 3.4 3.5

TN (mg/L) 13.5 7.5

TP (mg/L) 1.5 4.1

Total E. coli (CFU/ml) Not detected Not detected

wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 8 6 3e8 7 1 869

indicating that NaClO CIP could remove some foulants

including protein-like substances to some extent.

From the qualitative EDS spectrum (data not shown), metal

elements such as Ca, Fe, Mg, Al, Na and Mn were detected,

indicating that inorganic substances also contributed to

membrane fouling. Metal ions or clusters exist widely in

municipal wastewater and activated sludge could be the

origin of inorganic foulants. These cations (Ca2þ, Fe3þ, Mg2þ,Al3þ) could be easily precipitated by the biopolymers with

anion groups (SO42�, CO3

2�, PO43�, OH�), thus enhancing

membrane fouling (Seidel and Elimelech, 2002). Inorganics

were generally the major foulants in anaerobic MBR (Yoon

et al., 1999; Kang et al., 2002; You et al., 2006) and were also

detected as the important foulants in aerobic MBR (Ognier

et al., 2002; Adham et al., 2004; Meng et al., 2007; Wang

et al., 2008). In addition, no obvious change was found for

metal elements before and after NaClO CIP. This indicated

that NaClO CIP was not effective in removing inorganic fou-

lants. In general acid (Yoon et al., 1999; Adham et al., 2004; Lee

and Kim, 2009) and metal-chelator (You et al., 2006) cleaning

agents were more effective in removing inorganic fouling.

Considering the former analysis, the residual fouling in terms

of pore blocking and partial gel layer after NaClOCIPwas likely

related to inorganics. The existence of inorganic foulants also

demonstrated the necessity of acid cleaning for restoring

membrane permeability completely.

According to theaboveanalysis, apreliminaryexplanationon

membrane fouling in Run 1 and Run 2 under sub-critical flux

operation and NaClO CIP could be concluded. Sub-critical flux

operation prevented sludge flocs from depositing on the

membrane surface and avoided the formation of cake layer

fouling. Thus gel layer and pore blockage, formed mainly by

macromolecular organics and inorganics, became the dominant

membrane fouling. With the accumulation of gel layer and pore

blockage fouling, membrane resistance and TMP increased

under constant-flux operation. NaClO CIP could removemost of

the gel layer and partial pore blockage fouling formedmainly by

macromolecular organics. Therefore TMP clearly decreased due

to the restoration of membrane filterability after NaClO CIP.

Steady filtration in Run 1e3 without acid cleaning was possibly

attributed to negligible inorganic fouling caused by low concen-

trationsofmetal ions inmunicipalwastewater and/or short time

operation. Chemical CIP with NaClO as a primary agent and

supplementaryHCl should be adopted inorder to achieve steady

filtration for much longer operation (at least 1 year).

3.4. Effects of chemical CIP on effluent quality

Table 4 shows the effluent quality before and after one

chemical CIP performed with high concentration NaClO

(3000 mg/L) on day 471. CODCr, BOD5 and NH4þeN exhibited no

obvious changes before and after cleaning. TN decreased and

TP increased in the initial time after cleaning. This wasmainly

attributed by the non-aerated period in the bioreactor during

cleaning, providing the anoxic and anaerobic environment for

denitrification and phosphorus releasing. Subsequently, TN

and TP returned to a normal value after about 1 h operation. In

general, chemical CIP had no obvious effect on effluent

quality.

4. Conclusions

During the long-term (750 d) operation of a pilot-scale

submerged membrane bioreactor for municipal wastewater

treatment, steady filtration for at least 265 d under high flux

(30 L/(m2 h)) was successfully achieved due to effective

membrane fouling control by sub-critical flux operation and

chemical CIP with NaClO. Sub-critical flux operation pre-

vented rapid fouling caused by cake layer formation. NaClO

CIP, in both a TMP controlling mode (cleaning with high

concentration NaClO (2000e3000 mg/L) was performed when

TMP rose to 15 kPa) and a time controlling mode (cleanings

were performed weekly and monthly respectively with low

concentration NaClO (500e1000 mg/L) and high concentration

NaClO (3000 mg/L)), effectively controlled slow fouling caused

by pore blockage and gel layer. Critical flux operation (30e35 L/

(m2 h)) under chemical CIP with combined TMP and time

controlling mode achieved steady filtration up to 140 d and

appeared promising for achieving longer steady filtration

although it was less stable than sub-critical flux operation.

However super-critical flux operation (35e42 L/(m2 h)) was not

attainable for achieving long-term steady filtration even

adopting chemical CIP. In addition, the results also demon-

strated that themethod for critical fluxmeasurement through

short-term constant-flux filtration test used in this study was

feasible for long-term operation.

wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 8 6 3e8 7 1870

From macroscopic cleaning performance, high concentra-

tion NaClO was the best. HCl (pH ¼ 1) and NaOH (pH ¼ 12)

showed results similar to low concentration NaClO. The

combination of NaClO and HCl or NaOH appeared to be better

than NaClO alone, especially when serious fouling occurred.

Microscopic analysis onmembranefibers before andafter high

concentrationNaClOCIPwas conducted in order to investigate

themechanism of NaClO CIP for removingmembrane fouling.

SEM and AFM images showed that NaClO cleaning could

effectively remove the gel layer, the dominant fouling under

sub-critical flux operation. Porosity measurements indicated

that NaClO cleaning could partially remove pore blockage

fouling. ATR-FTIR and EDS analyses demonstrated that

protein-likemacromolecular organics and inorganicswere the

important components of the fouling layer. The analysis of

effluent quality before and after NaClO CIP showed no obvious

negative effect on effluent quality.

Acknowledgements

This work was supported by the National Science Fund for

Distinguished Young Scholars (No. 50725827) and 863 program

(No. 2009AA062901).

r e f e r e n c e s

Adham, S., DeCarolis, J.F., Pearce, W., 2004. Optimization ofvarious MBR systems for water reclamation-Phase III.Desalination and Water Purification Research andDevelopment Program Final Report, No. 103.

Al-Halbouni, D., Traber, J., Lyko, S., Wintgens, T., Melin, T.,Tacke, D., Janot, A., Dott, W., Hollender, J., 2008. Correlation ofEPS content in activated sludge at different sludge retentiontimes with membrane fouling phenomena. Water Research 42(6e7), 1475e1488.

APHA-AWWA-WEF, 1995. Standard methods for the examinationof water and wastewater, nineteenth ed. American PublicHealth Association, American Water Works Association andWater Environment Federation, Washington, DC, USA.

Bouhabila, E.H., Ben Aim, R., Buisson, H., 1998. Microfiltration ofactivated sludge using submerged membrane with airbubbling (application to wastewater treatment). Desalination118 (1e3), 315e322.

Brepols, C., Drensla, K., Janot, A., Trimborn, M., Engelhardt, N.,2008. Strategies for chemical cleaning in large scale membranebioreactors. Water Science and Technology 57 (3), 457e463.

Chang, I.S., Judd, S.J., 2002. Air sparging of a submerged MBR formunicipal wastewater treatment. Process Biochemistry 37 (8),915e920.

Chang, I.S., LeClech, P., Jefferson, B., Judd, S., 2002. Membranefouling in membrane bioreactors for wastewater treatment.Journal of Environmental Engineering-ASCE 128 (11), 1018e1029.

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.

Fatone, F., Battistoni, P., Pavan, P., Cecchi, F., 2007. Operation andmaintenance of full-scale municipal membrane biologicalreactors: a detailed overview on a case study. Industrial &Engineering Chemistry Research 46 (21), 6688e6695.

Grelot, A., Grelier, P., Vincelet, C., Bruss, U., Grasmick, A., 2010.Fouling characterisation of a PVDF membrane. Desalination250 (2), 707e711.

Grelot, A., Machinal, C., Drouet, K., Tazi-Pain, A., Schrotter, J.C.,Grasmick, A., Grinwis, S., 2008. In the search of alternativecleaning solutions for MBR plants. Water Science andTechnology 58 (10), 2041e2049.

Guglielmi, G., Saroj, D.P., Chiarani, D., Andreottola, G., 2007. Sub-critical fouling in a membrane bioreactor for municipalwastewater treatment: experimental investigation andmathematical modeling. Water Research 41 (17), 3903e3914.

Huang, X., Wei, C.H., Yu, K.C., 2008. Mechanism of membranefoulingcontrolbysuspendedcarriers ina submergedmembranebioreactor. Journal of Membrane Science 309 (1e2), 7e16.

Huang, X., Xiao, K., Shen, Y.X., 2010. Recent advances inmembrane bioreactor technology for wastewater treatment inChina. Frontiers of Environmental Science & Engineering inChina 4 (3), 245e271.

Itokawa, H., Thiemig, C., Pinnekamp, J., 2008. Design andoperating experiences of municipal MBRs in Europe. WaterScience and Technology 58 (12), 2319e2327.

Kang, I.J., Yoon, S.H., Lee, C.H., 2002. Comparison of the filtrationcharacteristics of organic and inorganic membranes ina membrane-coupled anaerobic bioreactor. Water Research 36(7), 1803e1813.

Kimura, K., Yamato, N., Yamamura, H., Watanabe, Y., 2005.Membrane fouling in pilot-scale membrane Bioreactors(MBRs) treating municipal wastewater. Environmental Science& Technology 39 (16), 6293e6299.

Kraume, M., Wedi, D., Schaller, J., Iversen, V., Drews, A., 2009.Fouling in MBR: what use are lab investigations for full scaleoperation? Desalination 236 (1e3), 94e103.

Le-Clech, P., Chen, V., Fane, T.A.G., 2006. Fouling in membranebioreactors used in wastewater treatment. Journal ofMembrane Science 284 (1e2), 17e53.

Le Clech, P., Jefferson, B., Chang, I.S., Judd, S.J., 2003. Critical fluxdetermination by the flux-stepmethod in a submergedmembranebioreactor. Journal of Membrane Science 227 (1e2), 81e93.

Lee, M., Kim, J., 2009. Membrane autopsy to investigate CaCO3

scale formation in pilot-scale, submerged membranebioreactor treating calcium-rich wastewater. Journal ofChemical Technology and Biotechnology 84 (9), 1397e1404.

Lim, A.L., Bai, R., 2003. Membrane fouling and cleaning inmicrofiltration of activated sludge wastewater. Journal ofMembrane Science 216 (1e2), 279e290.

Liu, R., Huang, X., Chen, L.J., Wang, C.W., Qian, Y., 2000. A pilotstudy on a submerged membrane bioreactor for domesticwastewater treatment. Journal of Environmental Science andHealth Part A Toxic/Hazardous Substances & EnvironmentalEngineering 35 (10), 1761e1772.

Lyko, S., Wintgens, T., Al-Halbouni, D., Baumgarten, S., Tacke, D.,Drensla, K., Janot, A., Dott, W., Pinnekamp, J., Melin, T., 2008.Long-term monitoring of a full-scale municipal membranebioreactor e characterisation of foulants and operationalperformance. Journal of Membrane Science 317 (1e2), 78e87.

Matosic, M., Prstec, I., Jakopovic, H.K., Mijatovic, I., 2009.Treatment of beverage production wastewater by membranebioreactor. Desalination 246 (1e3), 285e293.

Meng, F.G., Zhang, H.M., Yang, F.L., Liu, L.F., 2007. Characterizationof cake layer in submerged membrane bioreactor.Environmental Science & Technology 41 (11), 4065e4070.

Miyoshi, T., Tsuyuhara, T., Ogyu, R., Kimura, K., Watanabe, Y.,2009. Seasonal variation in membrane fouling in membranebioreactors (MBRs) treating municipal wastewater. WaterResearch 43 (20), 5109e5118.

Ognier, S., Wisniewski, C., Grasmick, A., 2002. Characterisationand modeling of fouling in membrane bioreactors.Desalination 146 (1e3), 141e147.

wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 8 6 3e8 7 1 871

Pollice, A., Brookes, A., Jefferson, B., Judd, S., 2005. Sub-criticalflux fouling in membrane bioreactors e a review of recentliterature. Desalination 174 (3), 221e230.

Qin, J.J., Oo, M.H., Tao, G.H., Kekre, K.A., Hashimoto, T., 2009. Pilotstudy of a submerged membrane bioreactor for waterreclamation. Water Science and Technology 60 (12),3269e3274.

Seidel, A., Elimelech, M., 2002. Coupling between chemical andphysical interactions in natural organic matter (NOM) foulingof nanofiltration membranes: implications for fouling control.Journal of Membrane Science 203 (1e2), 245e255.

Trussell, R.S., Adham, S., Trussell, R.R., 2005. Process limits ofmunicipal wastewater treatment with the submergedmembrane bioreactor. Journal of Environmental Engineering-ASCE 131 (3), 410e416.

Wang, Z.W., Wu, Z.C., Yin, X., Tian, L.M., 2008. Membrane foulingin a submerged membrane bioreactor (MBR) under sub-criticalflux operation: membrane foulant and gel layercharacterization. Journal of Membrane Science 325 (1),238e244.

Wei, C.H., Huang, X., Wen, X.H., 2006. Pilot study on municipalwastewater treatment by a modified submerged membranebioreactor. Water Science and Technology 53 (9), 103e110.

Xing, C.H., Wen, X.H., Qian, Y., Wu, W.Z., Klose, P.S., 2003. Foulingand cleaning in an ultrafiltration membrane bioreactor formunicipal wastewater treatment. Separation Science andTechnology 38 (8), 1773e1789.

Yang, W.B., Cicek, N., Ilg, J., 2006. State-of-the-art of membranebioreactors:worldwide researchand commercial applications inNorthAmerica. Journal ofMembraneScience270 (1e2), 201e211.

Yoon, S.H., Kang, I.J., Lee, C.H., 1999. Fouling of inorganicmembrane and flux enhancement in membrane-coupledanaerobic bioreactor. Separation Science and Technology34 (5), 709e724.

You, H.S., Huang, C.P., Pan, J.R., Chang, S.C., 2006. Behavior ofmembrane scaling during crossflow filtration in the anaerobicMBR system. Separation Science and Technology 41 (7),1265e1278.

Yu, K.C., Wen, X.H., Bu, Q.J., Xia, H., 2003. Critical fluxenhancements with air sparging in axial hollow fibers cross-flow microfiltration of biologically treated wastewater. Journalof Membrane Science 224 (1e2), 69e79.

Zhang, S.T., Qu, Y.B., Liu, Y.H., Yang, F.L., Zhang, X.W.,Furukawa, K., Yamada, Y., 2005. Experimental study ofdomestic sewage treatment with a metal membranebioreactor. Desalination 177 (1e3), 83e93.