Separation and Purification Technology 124 (2014) 124–133

Contents lists available at ScienceDirect

Separation and Purification Technology

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

Optimal operating strategies of SFDM formation for MBR application

1383-5866/$ - see front matter � 2014 Published by Elsevier B.V.http://dx.doi.org/10.1016/j.seppur.2014.01.028

⇑ Corresponding author. Tel.: +98 21 61112184; fax: +98 21 66954041.E-mail address: mmehrnia@ut.ac.ir (M.R. Mehrnia).

Fariba Rezvani, Mohammad Reza Mehrnia ⇑, Amir Ali PoostchiBiotechnology Group, School of Chemical Engineering, College of Engineering, University of Tehran, P.O. Box 11155-4563, Tehran, Iran

a r t i c l e i n f o a b s t r a c t

Article history:Received 27 June 2013Received in revised form 11 January 2014Accepted 17 January 2014Available online 29 January 2014

Keywords:Mesh filterMembrane bioreactorSelf-forming dynamic membraneWastewater treatment

This study develops process of self-forming dynamic membrane (SFDM) formation on mesh filter andinvestigates the capability of such process in treating municipal waste water. The formation process ismainly divided into two stages, i.e., the agitation and the aeration. At the agitation stage, the dynamicmembrane is built up by deposition of suspended sludge particles on the mesh surface. The just-completesuspension of sludge flocs is performed using mechanical axial-flow agitator. Four imposed fluxes areapplied to obtain the optimal flux for SFDM. Three parameters, i.e., effluent chemical oxygen demand(COD), filtrate turbidity, and trans-membrane pressure (TMP) are simultaneously measured to determinethe time required for SFDM formation at agitation stage. The optimal aeration rate is also determined tosteady operate the self-forming dynamic membrane bioreactor (SFDMBR). At the aeration stage, the restof dynamic membrane is gradually formed during continuous operation. The compressibility index ofdynamic membrane is also measured at the aeration stage. The results indicate that variations of thetrans-membrane pressure correspond to the variations of compressibility index. Scanning electronicmicroscopy is also utilized to observe the development of SFDM. The performance of SFDMBR is evalu-ated in terms of total dissolved solids (TDS) and COD removal. The results reveal that the SFDMBR canbe substituted for conventional membrane bioreactors. A clean surface of mesh filter is restored by usingtwo cleaning methods, i.e., air scouring and water backwash. Based on Fourier transform infrared analy-sis, proteins and carbohydrates are the dominant substances comprising the SFDM.

� 2014 Published by Elsevier B.V.

1. Introduction

The membrane bioreactor (MBR), integrating a separationprocess and biological degradation, is widely used as a proventechnology in treating both domestic and industrial wastewater[1]. MBRs provide two main advantages, including a significantlyimproved effluent quality; due to high mixed liquor suspended so-lid (MLSS) concentration, and a substantially small footprint; dueto absence of secondary settling tank [2,3]. However, a majorobstacle, inevitably limiting the application of MBRs, is the rapidjump in the trans-membrane pressure (TMP) and/or rapid declinein the permeate flux because of fouling occurrence [4]. In recentyears, there are considerable investigations focusing on develop-ment of efficient procedures to alleviate the fouling of membranein MBRs [5,6]. The dynamic membrane process can be consideredas an appropriate approach to solve the said bottleneck. The self-forming dynamic membrane coupled with a bioreactor is a newintegrated approach applied in MBR technology using a coarsepore-sized material module, such as woven, non-woven fabric,and filter-cloth, instead of conventional membranes [7,8]. TheSFDMBR has a low module cost and a high flux compared to

MBR using micro-/ultra-filtration membrane [7]. Fan and Huang[8] applied Dacron mesh polyester with 30 lm nominal pore sizein a MBR to evaluate the performance and formation of SFDM. Theyshowed that MBR equipped with mesh filter for treating municipalwastewater could remove more that 84% interring chemicaloxygen demand (COD) (97.9–371.7 mg/L COD). In this context,using low-cost filter materials including woven and non-wovencan be considered as suitable alternatives to the conventionalmembrane [9]. Chu and Li [10] reported that dynamic membraneformed in the filter-cloth material could effectively reject the par-ticular matter. They revealed that the effluent turbidity was nearlykept at stable level within 2–6 h for different concentrations ofsludge. The average COD and NH4–N removal reached 80% and74%, respectively. Liu et al. [11] separated the formation processof dynamic membrane into four stages under constant filtrationpressure, i.e., substrate formation, separation layer formation, foul-ing layer formation and filtration cake formation. They alsoobtained that COD and turbidity contents in the filtrate becamestable after 5 h, implying that the complete formation of dynamicmembrane at sludge concentration of 7540 mg/L required 5 h.Zhang et al. [12] investigated the formation of dynamic membraneon the mesh filter with average pore size of 61 lm used inanaerobic membrane bioreactor. They showed that the dynamicmembrane was completed after 20 days of operation. The results

F. Rezvani et al. / Separation and Purification Technology 124 (2014) 124–133 125

presented by them showed that the dynamic membrane wasformed by the sludge particles and the solutes and colloids contentsuch as SMP and EPS. Ren et al. [13] demonstrated that the nonwo-ven fabric filter with a biomass layer effectively removed activatedsludge. Kiso et al. [14] studied the performance of mesh filtrationbioreactor equipped with a nylon mesh filter as the filter materialwith three pore sizes 100, 200, and 500 lm for activated sludgeseparation instead of microfiltration membrane in a MBR. Theyconcluded that the mesh filter bioreactor retained suspended sol-ids up to 9 g/L. They also investigated the performance of com-bined process of sequencing batch reactor and mesh filtrationbioreactor [15]. Zahid and El-Shafai [16] evaluated the perfor-mance of cloth-media filters made of three materials (Acrylate,Polyester, and Nylon) for MBR treating municipal wastewater.The average COD was in the range of 30–37 mg/L for all the threefilters. Poostchi et al. [17] also investigated the filtration character-istics and the hydraulic resistance of the polyester mesh filter withaverage pore size of 30 lm, submerged in MBR. In all studies re-lated to the formation of SFDM the dynamic membrane was grad-ually formed during the operation thus the performance ofdynamic membrane was being promoted with operation time. Inorder to obtain the good performance of dynamic membrane atthe beginning stage of the filtration, in the present study, theformation of dynamic membrane is hastened by designing an addi-tional stage here in after known as the agitation. The act of agita-tion helps to quickly-uniformly form the dynamic membrane inthe polymeric mesh filter. The effects of operational conditionsincluding aeration intensity and imposed flux on the formationof SFDM are also investigated in order to obtain the optimal valuesof these operational parameters for the SFDM formation at theagitation stage. Moreover, the rest of SFDM formation is theninvestigated at the aeration stage where the long-term operationof the SFDMBR is performed for wastewater treatment.

2. Material and methods

2.1. Experimental set up

The experimental setup is illustrated in Fig. 1(a). The rectangu-lar Plexiglas tank is used as the bioreactor with 0.15 m width,0.2 m length and 0.4 m height. A polyester flat sheet mono-filamentous mesh filter is used with mean pore size of 30 lm withan initial hydraulic resistance of 7.36 � 107 m�1. The area of themesh is roughly about 0.016 m2. The filter medium with its holdingframe, as depicted in Fig. 1(b), is vertically submerged in the centerof the bioreactor. Pressure drop across the filter medium is mea-sured by an on-line pressure sensor. To adjust the permeate flowrate a peristaltic pump is used. The bioreactor is equipped withfour fine bubbles air pipe distributors evenly placed at the bottomof the bioreactor, one of which is exactly below the filter modulefor intermittent cleaning of filter surface. A stainless steel four-blade agitator is used to suspend the solid while the dynamicmembrane is being formed. The mechanical-driven agitator isplaced at a distance of 3 cm below the filter to provide appropriateflow-pattern.

2.2. Operation conditions of continuous bioreactor

The working volume of the bioreactor is 6 L. The bioreactortemperature is set in the range of 23–25 �C. The hydraulic reten-tion time (HRT) and sludge residence time (SRT) are set on 12 hand 32 days, respectively. The four imposed fluxes are used to opti-mize the formation of the SFDM at the agitation stage. The four im-posed fluxes of 62.47, 112.5, 150, and 190 LMH are used tooptimize the formation of the SFDM at the agitation stage. The

permeate flux is then kept constant at 30 LMH for long time oper-ation of the SFDM at the aeration stage. The bioreactor operateswith synthetic wastewater to avoid any fluctuation in the influentloading. The COD:N:P ratio of influent is 100:5:1 containing1125 mg/L glucose, 231.76 mg/L ammonium sulfate, and51.09 mg/L ammonium phosphate. Four aeration rates of 0.01,0.05, 0.09, and 0.15 m3 h�1 are applied to observe the SFDM stabil-ity against the aeration intensity. Consequently, dissolved oxygen(DO) is varied in the range of 2–6.5 mgO2/L. The influent total dis-solved solid (TDS) and turbidity are 687 mg/L and 288 NTU, respec-tively. NaHCO3 and H2SO4 are used to set pH in bioreactor neutral.Seed sludge is taken from the operating MBR plant (installed byHuber technology). The sludge suspensions are acclimatized for107 days with the same operating parameters of the operatingMBR. Mixed liquor suspended solids (MLSS) concentration is keptalmost constant between 8.5 ± 0.8 g/L. When the TMP reaches26 kPa after 15 days of operation, the SFDM is removed by usingtwo physical cleaning methods, i.e., air scouring and waterbackwash.

2.3. Analytical procedures

The analytical parameters such as COD, turbidity, MLSS, andMixed liquor volatile suspended solids (MLVSS) are analyzed usingMerck reagents, according to APHA standard methods [18]. DO andtemperature are measured with multi-meter (WTW 340 i,Germany). TDS is measured with TDS-meter (HM digital, America).Mass of suspended solids is measured with balance (Tecator,Germany) at a resolution of 0.0001 g. In order to minimize the er-rors due to the experimental conditions, all results are duplicatedin two times and the average of the results is considered.

2.4. Field emission scanning electron microscopic (FESEM)

The surface of fresh mesh filter and surface of cake layer formedon the used mesh filter with corresponding cross-section structurewere observed with scanning electron microscopy (S4160, Hitachi,Japan). At each sampling, the suction pump is stopped and the per-meate valve is normally closed to prevent the back flow of the fil-trate due to driving force owing to relatively high pressuredifference between filter module and suction line. The filter withthe SFDM is separated from the holding frame. The sample is fixedwith 2% (v/v) glutaraldehyde in 0.1 M phosphate buffer at pH 7.2for 2 h and then washed for 10 min and again immersed for 1 hin 0.1 M phosphate buffer. The fixed sample is dehydrated withethanol. When the drying is accomplished, the sample of dynamicmembrane is precisely obtained and it is accurately fixed to thewelfare of SEM.

2.5. Filtration resistance analysis

The filter fouling phenomena can be described by using a theo-retical model known as the resistance-in-series model [19,20]:

R ¼ DPlJ

ð1Þ

Rt ¼ Rm þ Rc þ Rf ð2Þ

where J is the permeate flux (m3/m2 s), DP is the trans-filter pres-sure (Pa), l is viscosity of permeate (Pa s), Rt is total filtrationresistance (m�1), Rm is the filter resistance (m�1), Rc is the cakeresistance (m�1), and Rf is the gel layer resistance (m�1). The men-tioned resistances are calculated using the following equations:

Rm ¼DPw

lJð3Þ

Permeate flowParticleof sludge

FA

FD

FL

Mixing intensity

Mesh filterFour-bladeagitator

Fluid

c

b

a

Fig. 1. (a) Schematic diagram of mesh filter bioreactor system, (b) sketch of mesh filter module. (c) Schematic of the suspended solids deposit near the module.

126 F. Rezvani et al. / Separation and Purification Technology 124 (2014) 124–133

Rf ¼DPwf

lJ� Rm ð4Þ

Rc ¼DPAS

lJ� Rm � Rf ð5Þ

where DPw is the initial of trans-filter pressure, DPwf is the final oftrans-filter pressure after removing the cake layer by flushing withtap water and DPAS is the trans-filter pressure of the activatedsludge at the steady state [21].

F. Rezvani et al. / Separation and Purification Technology 124 (2014) 124–133 127

2.6. Fourier transform infrared (FT-IR) spectroscopy

The fouled mesh filter module is taken out from the bioreactorat the end of cycle when TMP reaches about 26 kPa. The SFDM onthe mesh filter surface is scraped off and then rinsed with purewater. The collected sample is placed on a magnetic stirrer wherea good mixture is performed. KBr pellet of the dried foulant sampleis analyzed by an FT-IR spectrophotometer (Tensor 27, BrukerOptics, Germany) at a resolution of 4 cm�1.

3. Results and discussion

3.1. The SFDM formed at agitation stage

3.1.1. Formation process of SFDMSchematic of process flow diagram for SFDM formation and the

dismantled filter module is shown in Fig. 1(a, b). The formationprocess of SFDM is performed in the bioreactor at the agitationstage. As shown in Fig. 1(c), the suspended solids (SS) located nearthe membrane module are influenced by three forces, includingdrag force due to the permeate flow (FD), the shear stress due tothe agitation (FA), and the friction force on the filter surface (FL).When the shear stress and friction force are equivalent, solid par-ticles near the membrane module tend to be immobilized, leadingto deposition of particles on the surface of the mesh surface.During the SFDM formation process, the MLSS concentration isadjusted to 8.5 ± 0.8 g/L where the influent COD and turbidity isset at 1200 mg/L and 288 NTU, respectively. When the mixture le-vel inside the bioreactor reaches the top of the filter module, theinput flow is stopped and permeate peristaltic pump is switchedon. Recycle flow is supplied by the permeate flow, thus the levelof bioreactor remains constant. Due to providing an appropriateflow-pattern, having a better settling of sludge on the surface offilter, and promoting the SFDM formation, the agitation techniqueis selected at the beginning stage of the SFDM formation. Further-more, the agitation provides the oxygen required for sludge activ-ity in COD removal. The axial-flow agitator with the rotationalspeed of 350 rpm is used both to perform a just-complete suspen-sion of the solids and to provide sufficient dissolved oxygen (DO).In all studies previously done in the field of the formation of SFDM[9–12], aeration intensity created the energy dissipation requiredfor suspension of the activated sludge particles throughout the vol-ume of the MBR system. In such system the sludge flocs suspendedby aeration gradually deposited upon the porous media as well aswoven and non-woven fabrics over the extended period of theoperation of MBR. The SFDM was previously formed without help-ing any additional accessories such as agitator and the aerationwas the only method to suspend the particles in the bioreactor.The main differences between the formation of SFDM with andwithout agitator are the flow-pattern of suspension flocs and thepresence of air bubbles in the bioreactor. In other word, with noagitator, the dynamic membrane is formed in the three-phasegas–liquid–solid membrane bioreactor since the dynamicmembrane is formed in the two-phase solid–liquid membrane bio-reactor with agitator. It is obvious that three-phase mixture hasmuch more complex interactions rather than two-phase mixture.It can be concluded that when the agitation method is applied,the more reliability for the formation of SFDM is obtained as com-pared with aeration method. The main advantages of using agitatorfor the SFDM formation include the following: (1) to create regularflow-pattern throughout the volume of MBR system, (2) to gener-ate the axial flow next to filter surface, depending on the agitatorgeometry. Since the axial-flow agitator uniformly creates theappropriate flow-pattern at the close proximity of mesh surface,the use of mechanical agitator expedites the time of the SFDM

formation and improves the uniformity of the structure the SFDM.When the permeate flow is being returned back to the bioreactorthe suspension of sludge flocs is gradually deposited on the filtersurface. The time in which the effluent quality becomes stable isknown as the required time for SFDM formation at the agitationstage. It can be expected that the mesh filter effectively adsorbsactivated sludge when the binding of the biomass particles and fil-ter surface happens. To provide a better understanding of develop-ment in formation of SFDM, SEM images is obtained from thecross-section of dynamic membrane at different periods of the agi-tation stage as shown in Fig. 2(a–d). Fig. 2(b) shows that colloidalmatters attached to the intersection of yarns produced a gel layerknown as substrate layer [12]. According to Fig. 2(c), flocs embed-ded in pores, grow rapidly and generate a separation layer attachedloosely to the filter. Due to increasing the thickness of SFDM duringthe operation time at the aeration stage, the formation of SFDM isnot completed at the agitation period. Certainly, the fouling layer isformed during the operation time at the aeration stage.

3.1.2. Optimization of the imposed flux for SFDMTo optimize the imposed flux during the formation of SFDM at

the agitation stage, four imposed fluxes of 62.47, 112.5, 150, and190 LMH are examined. Three parameters including the effluentCOD, the filtrate turbidity, and the TMP are also measured to obtainthe optimal value for the imposed flux. Fig. 3(a–c) depicts thevariations of the effluent COD, the filtrate turbidity, and the TMPprofile at aforementioned fluxes. For all four imposed fluxes thepermeate quality increases with the SFDM formation time andfinally reaches constant value. As shown in Fig. 3(a, b), when theimposed flux increases from 62.74 to 190 LMH, the ultimate per-meate quality increases until 150 LMH and thereafter the qualityof filtrate decreases. It may be due to the fact when increasingthe imposed flux from 150 to 190 LMH, drag force overcomes shearstress so that colloids pass through the SFDM, and thus leading toincrease in the permeate quality (effluent COD and turbidity). Thefinal effluent COD and turbidity becomes equal to 81 mg/L and43 NTU for the imposed flux 150 LMH, respectively. Correspond-ingly, the required time for the SFDM formation is equal to 840 swhen the thickness of the SFDM becomes equal to 248 lm (asshown in Fig. 2(d)). It is also observed when increasing the im-posed flux the time required for the SFDM formation is decreased.The monitoring of TMP profile is also performed for the appliedfluxes as shown in Fig. 3(c). The TMP profile is close to each otherfor the imposed fluxes between 62.74 and 150 LMH. It means thatno significant change in the TMP is observed during the SFDM for-mation. At the imposed flux of 190 LMH the TMP profile severelyincreases due to increasing the compressibility index of the SFDM,thus the over-compacting of the SFDM happens. According to thevariations of the effluent quality and the TMP profile for the im-posed fluxes, it can be resulted that the imposed flux of 150 LMHis the optimum flux whereas the effluent quality reaches thelowest limit and the TMP profile remained stable without anysignificant changes. Fan and Huang [8] reported that the effluentsuspended solid concentrations were more than 1000 mg/L at theinitial stages of operation. It should be pointed out that thethickness of cake layer plays a major role in filtration capabilityof dynamic membrane [12]. Kiso et al. [15] indicated that in theoperation of mesh filtration the effluent suspended solid concen-trations were less than 10 mg/L in only 600 s of the filtration.

3.1.3. Optimization of aeration rate for stability of SFDMWhen the SFDM is built up at the agitation stage, the axial-flow

agitator is switched off and the permeate flux is set to 30 LMH forlong-term operation of the SFDMBR. The effect of aeration intensityon the SFDM stability is investigated. Aeration imparted to themembrane is denoted SADm, the specific aeration demand in the

Fig. 2. SEM photographs: (a) pristine mesh filter, (b) sub layer (gel), (c) the beginning of separation layer, and (d) the surface of SFDM when the turbidity reaches a constantvalue (magnification 2000�).

128 F. Rezvani et al. / Separation and Purification Technology 124 (2014) 124–133

volumetric rate of air per unit membrane area. The specificaeration demands (SADm) are varied in the range of 0.625–9.375 m3 m�2 h�1 in order to obtain the optimum condition forstability of the SFDM previously formed at the agitation stage.The stability of SFDM is evaluated on the basis of its capability toretain the suspended particles and colloidal substances. Fig. 4 de-picts a comparison of the final values of the effluent COD and tur-bidity at the short period of the applied aeration. According to dataobtained, with increasing the SADm from 0.625 to 5.625 m3 m�2 -h�1, no significant change in effluent quality values is observed.It means that the spatial structure of SFDM remains stable byincreasing SADm up to 5.625 m3 m�2 h�1. A sudden increase inthe effluent quality appears when the SADm is increased from5.625 to 9.375 m3 m�2 h�1 as shown in Fig. 4. It can be expectedthat the SADm as the representative of aeration intensity impacton the structure of the SFDM, thus affecting the effluent quality.In addition to the proper stability of the SFDM, the aeration rateprovides a good aerobic condition to the microorganisms forbiodegradation. Fig. 5 shows the average concentrations of DOwith the SADm. As depicted in Fig. 5, the DO values increase withincreasing the SADm; the amount of DO is sufficient to microbialactivity for biodegradation. In this case, the SADm of 5.625m3 m�2 h�1 can be expected as the optimum value in which thestability of the SFDM with the sufficient oxygen occurs.

3.2. Operation of SFDM at aeration stage

3.2.1. Permeate qualityThe recycling flow continues until the level of effluent quality

remains constant. At that point, stirring is stopped and aerationwill start. Simultaneously, peristaltic pump is set on flux of30 LMH and the permeate flow is totally transferred to permeatetank. The treatment performance of the dynamic membrane isevaluated in terms of COD, turbidity and TDS. Fig. 6(a) showsvariation of MLSS and MLVSS/MLSS in the mentioned period.Fig. 6(b–c) presents the turbidity, average COD concentrations,

and efficiency of COD removal in the effluent during the operation,respectively. According to Fig. 6(b), SFDM can reduce turbidityfrom 288 NTU to 15 ± 1 NTU (before and after physical cleaning).Turbidity as an important index for colloidal particles (0.1–10 lm) reduces significantly through the process. Due to the factthat turbidity reduction is indicative of microfiltration process, asignificant turbidity reduction indicates that the dynamic mem-brane is operated as a microfiltration process. This shows the ben-efit of using mesh filter instead of conventional membrane. Asshown in Fig. 6(c), the SFDM is able to reduce COD from1200 mg/L to 22 ± 2 mg/L (before and after physical cleaning).COD removal efficiency is about 98 ± 0.7% (before and after physi-cal cleaning) which indicates that SFDM can approximately be re-formed with the same performance. Fig. 6(d) shows TDS in theeffluent before and after physical cleaning. According to Figs. 5,6(d), TDS reduction during the operation can also be consideredas an implication of dynamic membrane compression. This param-eter represents the number of soluble solid particles which aresmaller than 100 nm [4], are separated by ultra-filtration in MBR.Therefore, at the beginning of the operation, TDS does not changesignificantly, but from the second day it reduces to 550 ± 5 mg/L(before and after physical cleaning). As it is shown in Fig. 6(b–d),the variations in the values of measured parameters are relatedto the first 4 days of each 15-day operation and afterwards the val-ues fluctuated in the same range so such that they can be consid-ered constant. It can be explained that when forming the SFDM onthe mesh filter during aeration stage, the biomass layer can effec-tively retain the fine particles and colloids. It can be concluded thatas the capacity of the SFDM in rejection of solutes and colloids isbalanced, the filtrate quality is kept constant at a stable level, butthe TDS of the filtrate declines slightly because of the attachmentof some solutes in the interior layer of SFDM. In fact, as filtrationis processed, the increasing in dynamic membrane compressibilityand thickness is observed, contributing to a higher pressure drop.To compare the operational performance of the SFDM formed byusing agitator and the dynamic membrane formed by using

a

b

c

Fig. 3. (a) COD decline, (b) turbidity decline, and (c) TMP profiles during theformation of SFDM using various fluxes of recycle flow.

Turb

idity

(NTU

)

SADm (m3 m-2 h-1)

Fig. 4. The variation of COD and turbidity values with aeration rate.

Aver

age

DO

(mgO

2/L)

SADm (m3 m2 h-1)

Fig. 5. Average concentration of DO against the SADm.

F. Rezvani et al. / Separation and Purification Technology 124 (2014) 124–133 129

aeration, data obtained from current study and the results of pre-vious studies are presented in Table 1.

3.2.2. TMP monitoring and compressibility indexTMP diagram through the time is shown in Fig. 7. According to

Fig. 7 there are three stages of pressure drop in each of the diagram

before and after physical cleaning which are related to the dynamicmembrane development and compressibility index. In each 15-dayoperation, when forming the dynamic membrane at agitationstage, the deposition rate of particles on the mesh filter surface in-creases till the second day, resulting in the development of dy-namic membrane. From the second day, pressure drop increasesdue to cake layer compression but solid particles deposition re-duces. From the eighth day, thickness of the dynamic membraneremains constant. As a result of membrane thickness getting con-stant and particles deposit decline, colloidal particles penetrateinto membrane pores and bring about their blockage. In otherword, from the eighth day, membrane compression increasesbesides, its porosity reduction gives rise to a drastic pressure drop.The layer compression and membrane pores blocking continueuntil the 15th day when the membrane thickness decreases to380 lm and pressure drop reaches its maximum value.Fig. 8(a–d) shows the thickness of dynamic membrane at 4 pointsof TMP diagram. As shown in Fig. 8, at the early stage of the oper-ation, the increase of dynamic membrane thickness has a succeed-ing tendency while it turns to be falling in such way that at the endof the operation not only membrane thickness does not increasebut also it decreases and gets more compressed in comparison toits thickness in the eighth day. The compressibility index, i.e. theindex of SS/cake volume, indicates the compaction of SFDM; thisis evaluated at three points as specified in Fig. 7. Liu et al. [11] pre-sented the compressibility factor to evaluate the formation mech-anism and the structure of dynamic membrane bioreactor. Zhanget al. [12] also represented that the index of SS/cake volume couldpredict the compactness and the voidage of the cake layer. Afterthe dynamic membrane is formed at the agitation stage asabove-mentioned, the rest particles are gradually deposited atthe aeration stage under the drag force of permeate flow contribut-ing to membrane thickness incensement. The dynamic membraneis not only formed by sludge particles, but also by the interceptionof the soluble and colloidal contents such as proteins and carbohy-drates in dynamic membrane during filtration [23]. Fig. 9(a, b)presents the mass of suspended solid (SS), thickness, and com-pressibility index values at certain points, respectively. As can beseen, membrane thickness increases from 248 lm (after the agita-tion stage) to 361 lm (the second day), indicating that, thicknessgrowth rate is high while compressibility index growth is not sig-nificant due to more porosity of the cake layer. At this stage, SFDMgrowth trend outweighs the tendency for compressibility indexincrease. However, from the second day till the eighth day, mem-brane thickness does not change much due to the fact that mem-brane pores are already filled with the soluble particles such thatmembrane thickness does merely increase from 361 lm to385 lm on the eighth day which is trivial as compared to theprevious level. As the thickness rate declines with operation time,

0.4

0.5

0.6

0.7

0.8

0.9

1

2

4

6

8

10

0 4 8 12 16 20 24 28 32

MLV

SS /

MLS

S

MLS

S (g

/L)

Time (d)

MLSS

MLVSS/MLSS

a

05

101520253035404550

0 4 8 12 16 20 24 28 32

Efflu

ent t

urbi

dity

( (N

TU)

Time (d)

b

80

85

90

95

100

0102030405060708090

0 4 8 12 16 20 24 28 32

CO

D re

mov

al e

ffeci

ency

%

Efflu

ent C

OD

(mg/

L)

Time (d)

COD before cleaning COD After cleaningCOD removal before cleaning COD removal after cleaning

c

400

450

500

550

600

650

700

0 4 8 12 16 20 24 28 32

Efflu

ent T

DS

(mg/

L)

Time (d)

Before cleaning After cleaning

d

Fig. 6. (a) MLSS and MLVSS/MLSS in continuous bioreactor, (b) effluent turbidity,(c) effluent COD and removal efficiency, and (d) effluent TDS before and afterphysical cleaning.

Table 1Comparison of data obtained and the results of similar works related to the SFDM.

Type of medium Pore size(lm)

Dynamic membraneformation

Formation time(min)

Dacron mesh 100 Aeration and recycling 60Terylene filter-cloth . . .. Aeration in batch test 200–300Nonwoven polyester

filter100 Aeration 130

Dacron mesh 61 Anaerobic 20 (day)Stainless steel mesh 58 Aeration and recycling 15Polyester mesh 30 Agitation and aeration 14

-5

0

5

10

15

20

25

30

0 4 8 12 16 20 24 28 32

TMP

(kpa

)

Operation time (d)

physicalcleaning

1 2

3

1 2

3

Fig. 7. Profile of TMP during the aeration stage.

130 F. Rezvani et al. / Separation and Purification Technology 124 (2014) 124–133

due to prevalence of drag force over shear stress, the dynamicmembrane thickness remains somewhat constant and colloidalparticles penetrate into the membrane pores which results indynamic membrane mass incensement. After that, as a result ofdynamic membrane compressibility, the thickness reduces to380 lm which leads to a drastic reduction of membrane porosityand a sharp increase in TMP. At this stage, membrane thickness de-creases due to compression resulted from the drag force; thus itcauses that the index reaches its highest value. These results areconsistent with the study reported by Zhang et al. [12].

3.2.3. Filtration resistance analysisThe filtrations resistance involves in a MBR include membrane

resistance (Rm), cake resistance (Rc), and blocking and irremovablefouling resistance (Rf) [21]. The hydraulic resistance analysis isconducted according to the resistance-in-series model as previ-ously explained. Fig. 10 quantitatively shows the filtration resis-tances measured for the imposed flux of 30 LMH at three pointsspecified in Fig. 7. Until the second day, the total hydraulic resis-tance is 1.8 � 1011 m�1 which contributions of cake and gel layerare 87% and more than 12%, respectively. When operation time isprolonged, pressure drop increases as a result of total resistanceenhancement such that the obtained total resistance in the eighthday is 5.65 � 1011 m�1 out of which 91% and less than 10% are re-lated to cake and gel, respectively. After that, total resistance gets1.11 � 1012 m�1 out of which more than 95% and less than 5%are related to cake and gel, respectively, which are nearly similarto the resistance values reported in the previous studies for meshmedia [24,25]. It is obvious that the accumulation of suspendedsolids major factor responsible for the increase of filtration resis-tance. Besides that, another important factor affecting membraneresistance is the layer of compaction [26].

3.2.4. SFDM cleaningPermeability recovery is investigated by the ratio of water

permeability after cleaning the initial water permeability. Two

Influent COD(mg/L)

Average COD removal(%)

Operation time(day)

Reference

97.9–371.7 84.2 90 [8]328 81 30–35 [10]286.2 ± 10.86 85.5–89.3 160 [13]

302.1 ± 87.9 57.3 ± 6.1 100 [12]140–380 40–60 7–20 (h) [22]1200 97 32 This

study

Fig. 8. SEM photographs: (a) the thickness of SFDM when the turbidity reaches a constant value, (b) the second day of filtration operation, (c) the eighth day of filtrationoperation; and (d) the fifteenth day of filtration operation.

0

1

2

3

4

5

6

7

8

9

0 3 6 9 12 15

a

0

50

100

150

200

250

300

350

400

450

0

200

400

600

800

1000

1200

1400

0 3 6 9 12 15

Thic

knes

s (µ

m)

Com

pres

sibi

lity

Inde

x (k

g/m

3 )

Time (d)

Time (d)

Index Tickness

b

Mas

s of

SS

(gr)

Fig. 9. (a) Mass of SFDM, (b) thickness and compressibility index of SFDM duringaeration stage.

0

2E+11

4E+11

6E+11

8E+11

1E+12

1.2E+12

1.4E+12

2 8 15

Res

ista

nce

(m-1

)

Time (d)

R cake R gel

Fig. 10. Resistances through gel and bio-cake at the corresponding points in Fig. 7.

F. Rezvani et al. / Separation and Purification Technology 124 (2014) 124–133 131

cleaning methods including air scouring and backwash are used toclean up the fouling layer. When the TMP reaches 26 kPa, the suc-tion pump is set off and then the cleaning is started. The air scour-ing is performed by coarse sparger located below the filter module.The pressurized air is introduced for 30 min at an air flux of0.3 m3 h�1. The on-line water backwash is performed by returning

back permeate flow to the filter module. The water backwash isrun for 40 s at a water flux of 30 LMH [27]. The permeability recov-eries are shown in Fig. 11. The cleaning recovery for air scouringand water backwash is restored to 98% and 96%, respectively. It re-sults that the physical cleaning can recover effectively the mesh fil-ter permeability, indicating that decrease in the TMP is mainlycaused by the deposition of sludge flocs. The microscopic observa-tion of the mesh filter cleaned shows that the some colloids andsolutes still attach between warp and weft fabrics (pictures notshown). In fact, it is concluded that mesh filter fouling is reversibleso that its permeability recovery is more than 98%.

3.3. FT-IR analysis

FT-IR spectra analysis is performed to identify the biomassfunctional groups responsible for the dynamic membrane formedon the mesh filter. According to FT-IR results of virgin mesh filterand dynamic membrane, it is well accepted that the dominant

Fouled mesh

Air scouringWater backwash

0

20

40

60

80

100

Perm

eabi

lity

reco

very

%

Fig. 11. Permeability recovery by two methods of physical cleaning.

3418

1638

1412

1050

620

0.40

0.50

0.60

0.70

0.80

0.90

1.00

40080012001600200024002800320036004000

Tran

smitt

ance

Wavelength (cm-1)

Fig. 12. FTIR analysis of SFDM.

132 F. Rezvani et al. / Separation and Purification Technology 124 (2014) 124–133

membrane foulants in MBR are protein and carbohydrate [28].Fig. 12 illustrates the FT-IR results of virgin mesh filter and dy-namic membrane. The broad peak of 1050 cm�1 is ascribed toCAO bonds associated with polysaccharide or polysaccharide-likesubstances [29]. The peak at 1412 cm�1 (CAN stretch) impliesthe secondary amines, representing proteins [30]. The peak of1638 cm�1 (NAH bend and NAC@O stretch) in the spectrum isindicative of the protein secondary structure, called primaryamines [31,32]. This result confirms the presence of carbohydratesand proteins in the cake layer formed on the surface of mesh filter.The peak of 3418 cm�1 is attributed to NAH stretch and OAHstretch of humic acids that cause clogging in the bioreactor[33,34]. The band of 620 cm�1 is observed in the fingerprint region(<1000 cm�1), corresponding to the phosphate and sulfate groupsarising from nucleic acids [35].

4. Conclusions

In this study, the suitability of a monofilament mesh filter as amembrane media in synthetic wastewater treatment is investi-gated. The self-forming dynamic membrane is in situ formed attwo stages, i.e., the agitation and the aeration. The dynamic mem-brane quickly built-up at the agitation stage provides a good capac-ity in rejecting the particular matter. The advantage of the additionof agitation stage is to expedite the SFDM formation as uniformlyas possible. By comparing the effluent quality and the TMP profile,the optimum value of the permeate flux is found to be equal to150 LMH for dynamic membrane formation at the agitation stage.The aeration intensity equal to 0.09 m3 h�1 is obtained as the opti-mum value in which the SFDM formed at the agitation stageremains stable. Three sharp steeps are seen in trans-membrane

pressure profile which can be attributed to the dynamic membranethickness and compressibility index. Results showed that com-pressibility index increases through the aeration stage which im-plies SFDM compression. The efficiency of dynamic membrane inremoving COD is outstanding, i.e. more than 97% at the hydraulicretention time of 12 h. Dynamic membrane also improves particleremoval efficiency of mesh filter via reducing turbidity as much asmicrofiltration membrane does. During the aeration stage, no sig-nificant change in TDS value is observed. The results obtained fromthe analysis of filtration resistance show that the cake layer resis-tance increases throughout the aeration stage while gel layerresistance decreases. Mesh filter is completely recovered afterphysical cleaning to re-build up SFDM. The preferable method forcleaning is the air scouring. The results of the FT-IR analysis alsoreveal that the dynamic membrane is characterized as proteinsand polysaccharides.

References

[1] E.J. McAdam, S.J. Judd, A review of membrane bioreactor potential for nitrateremoval from drinking water, Desalination 196 (2006) 135–148.

[2] K. Brindle, T. Stephenson, The application of membrane biological reactors forthe treatment of wastewaters, Biotechnol. Bioeng. 49 (1996) 601–610.

[3] S. Rosenberger, M. Kraume, Filterability of activated sludge in membranebioreactors, Desalination 146 (2002) 373–379.

[4] T. Stephenson, S. Judd, B. Jefferson, K. Brindle, Membrane Bioreactors forWastewater Treatment, IWA Publishing, London, 2000.

[5] P. Le-Clech, V. Chen, T.A.G. Fane, Fouling in membrane bioreactors used inwastewater treatment, J. Membr. Sci. 284 (2006) 17–53.

[6] F. Meng, S.R. Chae, A. Drews, M. Kraume, H.S. Shin, F. Yang, Recent advances inmembrane bioreactors (MBRs): membrane fouling and membrane material,Water Res. 43 (2009) 1489–1512.

[7] X.H. Zhou, H.C. Shi, Q. Cai, M. He, Y.X. Wu, Function of self-forming dynamicmembrane and biokinetic parameters’ determination by microelectrode,Water Res. 42 (2008) 2369–2376.

[8] B. Fan, X. Huang, Characteristics of a self-forming dynamic membrane coupledwith a bioreactor for municipal wastewater treatment, Environ. Sci. Technol.36 (2002) 5245–5251.

[9] Y. Satyawali, M. Balakrishnan, Treatment of distillery effluent in a membranebioreactor (MBR) equipped with mesh filter, Sep. Purif. Technol. 63 (2008)278–286.

[10] L. Chu, S. Li, Filtration capability and operational characteristics of dynamicmembrane bioreactor for municipal waste water treatment, Sep. Purif.Technol. 51 (2006) 173–179.

[11] H.B. Liu, C.Z. Yang, W.H. Pu, D.J. Zhang, Formation mechanism and structure ofdynamic membrane in the dynamic membrane bioreactor, Chem. Eng. J. 148(2009) 290–295.

[12] X. Zhang, Z. Wang, Z. Wu, F. Lu, J. Tong, L. Zang, Formation of dynamicmembrane in an anaerobic membrane bioreactor for municipal wastewatertreatment, Chem. Eng. J. 165 (2010) 175–183.

[13] X. Ren, H.K. Shon, N. Jang, Y.G. Lee, M. Bae, J. Le, K. Ch, I.S. Kim, Novelmembrane bioreactor (MBR) coupled with a nonwoven fabric filter forhousehold wastewater treatment, Water Res. 44 (2010) 751–760.

[14] Y. Kiso, Y.J. Jung, T. Ichinari, M. Park, T. Kito, K. Nishimura, K.S. Min,Wastewater treatment performance of a filtration bio-reactor equipped witha mesh as a filter material, Water Res. 17 (2000) 4143–4150.

[15] Y. Kiso, Y.J. Jung, T. Ichinari, T. Kito, K. Nishimura, K.S. Min, Coupling ofsequencing batch reactor and mesh filtration: operational parameters andwastewater treatment performance, Water Res. 39 (2005) 4887–4898.

[16] W.M. Zahid, S.A. El-shafai, Use of cloth-media filter for membrane bioreactortreating municipal wastewater, Bioresour. Technol. 102 (2011) 2193–2198.

[17] A.A. Poostchi, M.R. Mehrnia, F. Rezvani, M.H. Sarrafzadeh, Low-costmonofilament mesh filter used in membrane bioreactor process: filtrationcharacteristics and resistance analysis, Desalination 286 (2012) 429–435.

[18] Standard Methods for the Examination of Water and Wastewater, 20th ed.,American Public Health Association/American Water Works Association/Water Environment Federation, Washington, DC, USA, 1999.

[19] I.S. Chang, C.H. Lee, Membrane filtration characteristics in membrane-coupledactivated sludge system – the effect of physiological states of activated sludgeon membrane fouling, Desalination 120 (1998) 221–233.

[20] M. Cheryan, Ultrafiltration and Microfiltration Handbook, TechnimicPublishing, 1998.

[21] J.M. Lee, W.Y. Ahn, C.H. Lee, Comparison of the filtration characteristicsbetween attached and suspended growth microorganisms in submergedmembrane bioreactor, Water Res. 35 (2001) 2435–2445.

[22] H. Chu, Y. Zhang, X. Zhou, B. Dong, Bio-enhanced powder-activated carbondynamic membrane reactor for municipal waste water treatment, J. Membr.Sci. 433 (2013) 126–134.

[23] D. Jeison, I. Diaz, J.B. van Lier, Anaerobic membrane bioreactors: aremembranes really necessary?, Electron J. Biotechnol. 11 (2008) 1–7.

F. Rezvani et al. / Separation and Purification Technology 124 (2014) 124–133 133

[24] M.C. Chang, R.Y. Horng, H. Shao, Y.J. Hu, Performance and filtrationcharacteristics of non-woven membranes used in a submerged membranebioreactor for synthetic wastewater treatment, Desalination 191 (2006) 8–15.

[25] P.K. Tewari, R.K. Singh, V.S. Batra, M. Balakrishnan, Membrane bioreactor(MBR) for waste water treatment: filtration performance evaluation of lowcost polymeric and ceramic membranes, Sep. Purif. Technol. 71 (2010) 200–204.

[26] K.J. Hwang, Y.H. Cheng, The role of dynamic membrane in cross-flowmicrofiltration of macromolecules, Sep. Sci. Technol. 38 (2003) 779–795.

[27] J. Wu, P. Le-Clech, R.M. Stuetz, A.G. Fane, V. Chen, Effects of relaxation andbackwashing conditions on fouling in membrane bioreactor, J. Membr. Sci. 324(2008) 26–32.

[28] S. Kim, N. Jang, The effect of calcium on the membrane biofouling in themembrane bioreactor (MBR), Water Res. 40 (2006) 2756–2764.

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

[30] S. Gur-Reznik, I. Katz, C.G. Dosoretz, Removal of dissolved organic matter bygranular-activated carbon adsorption as a pretreatment to reverse osmosis ofmembrane bioreactor effluents, Water Res. 42 (2008) 1595–1605.

[31] T. Maruyama, S. Katoh, M. Nakajima, H. Nabetani, T.P. Abbott, A. Shono, K.Satoh, FT-IR analysis of BSA fouled on ultrafiltration and microfiltrationmembranes, J. Membr. Sci. 192 (2001) 201–207.

[32] K.J. Howe, K.P. Ishida, M.M. Clark, Use of ATR/FTIR spectrometry to studyfouling of microfiltration membranes by natural waters, Desalination 147(2002) 251–255.

[33] J. Cho, G. Amy, J. Pellegrino, Y. Yoon, Characterization of clean and naturalorganic matter (NOM) fouled NF and UF membranes, and foulantscharacterization, Desalination 118 (1998) 101–108.

[34] Y. Satyawali, M. Balakrishnan, Effect of PAC addition on sludge properties in anMBR treating high strength wastewater, Water Res. 43 (2009) 1577–1588.

[35] G. Guibaud, N. Tixier, A. Bouju, M. Baudu, Relation between extracellularpolymers’ composition and its ability to complex Cd, Cu and Pb, Chemosphere52 (2003) 1701–1710.