International Biodeterioration & Biodegradation 85 (2013) 491e498 Contents lists available at SciVerse ScienceDirect International Biodeterioration & Biodegradation journal homepage: www.elsevier.com/locate/ibiod Fouling characterization and nitrogen removal in a batch granulation membrane bioreactor Bui Xuan Thanh a, *, Chettiyapan Visvanathan b, Roger Ben Aim c a Faculty of Environment, Ho Chi Minh City University of Technology, Building B9, 268 Ly Thuong Kiet Str., District 10, Ho Chi Minh City 70000, Viet Nam Environmental Engineering and Management Program, School of Environment, Resources and Development, Asian Institute of Technology, P.O Box 4, Klong Luang, Pathumthani 12120, Thailand c Université de Toulouse, INSA, UPS, INP, LISBP, 135 Avenue de Rangueil, F-31077 Toulouse, France b a r t i c l e i n f o a b s t r a c t Article history: Received 16 November 2012 Received in revised form 19 February 2013 Accepted 21 February 2013 Available online 19 March 2013 A submerged membrane bioreactor (MBR) combined with aerobic granulation reactor was investigated for the simultaneous organic/nitrogen removal and membrane fouling control Total nitrogen (TN) removal was 59% (1.76 mg TN/g VSS h) in the aerobic granulation reactor The filtration of granulation effluent or low operating F/M condition of the MBR could extend the filtration period of up to 78 days without any need for physical cleaning The soluble fraction was the main contributor to fouling compared to colloids and solids The soluble polysaccharides (sPS) had more adverse effects than that of soluble protein (sPN) The deposition on a unit of the membrane’s surface area was 11 mg sPS/L m2 and mg sPS/L m2 As a result, the BG-MBR could be an alternative treatment process for simultaneous organic/nitrogen removal and fouling control Ó 2013 Elsevier Ltd All rights reserved Keywords: Aerobic granules Membrane bioreactor Extracellular polymeric substances Fouling Introduction The aerobic granular sludge process has been known to have many advantages as compared to the conventional activated sludge operations for about a decade The aerobic granule possesses a compact spherical structure, excellent settling ability, dense biomass structure, high biomass retention, ability for simultaneous nitrification-denitrification and removal of toxic substance (Beun et al., 2002; Carvalho et al., 2006; Thanh et al., 2008; Shi et al., 2011) The sludge is more stable in batch reactors due to the existence of feast and famine conditions in each cycle (Beun et al., 2002) The organic and nitrogen removal in the granulation system is high compared to that of conventional activated sludge process (Arrojo et al., 2004; Tay et al., 2007; Thanh et al., 2009; Lotito et al., 2012) However, the single granular sludge reactor was not able to meet the effluent standards due to the high suspended solids content in the effluent The suspended solids (SS) concentrations in the effluent of the granulation reactor were high, ranging from 75 to 250 mgVSS/L (Beun et al., 2002) and 200 to * Corresponding author Tel.: þ84 907866073 E-mail addresses: bxthanh@hcmut.edu.vn (C Visvanathan) (B.X Thanh), visu@ait.ac.th 0964-8305/$ e see front matter Ó 2013 Elsevier Ltd All rights reserved http://dx.doi.org/10.1016/j.ibiod.2013.02.005 450 mgTSS/L (Arrojo et al., 2004) Thus, a post treatment such as membrane filtration could be an add-in polishing step for complete treatment and water reuse Membrane technology has been proven to be the most effective wastewater treatment system in recent decades The advantages are less footprint requirements due to a high substrate loading rate, good treated water quality which can be reused for appropriate operations, less sludge production rate, high biomass retention, and microbial diversity, among others (Visvanathan et al., 2000) Membrane fouling could be due to the deposition of suspended solids/flocs (cake/gel formation, pore blocking), colloids (Bouhabila et al., 2001) and solutes (Shane Trussell et al., 2006; Jarusutthirak and Amy, 2006; Miyoshi et al., 2012) Recently, it has been found that the fouling mechanism of the submerged MBR is mainly caused by the deposition/accumulation of soluble extracellular polymeric substances (sEPS) on the membrane if reversible fouling (cake formation) is well controlled The sEPS mainly comprises of soluble polysaccharide (sPS) and soluble protein (sPN) The fouling potential of sPS, sPN or both of them is still unclear Both sPS and sPN were some of the factors which influenced membrane fouling (Shane Trussell et al., 2006; Liang et al., 2007; Miyoshi et al., 2012) where the sPS played a major role as membrane foulant (Rosenberger et al., 2006; Jarusutthirak and Amy, 2006; Kim and DiGiano, 2006) 492 B.X Thanh et al / International Biodeterioration & Biodegradation 85 (2013) 491e498 At the moment, there exists very limited published information related to the fouling behavior of the aerobic granular reactor effluent Some researchers studied the filterability of MBR seeding with pre-cultivated aerobic granules for a short operation period The granules used in the MBR were taken from a batch reactor Li et al (2005) reported that the permeability of the MBR seeding with pre-cultivated granules was 50% higher than that of the conventional MBR during 16 days of operation The author proposed that the compact and round shaped structure of the granule might cause less fouling due to the contact of less floc particles with the membrane’s surface Additionally, Tay et al (2007) also reported that the filterability of pre-cultivated granules was much better than that of conventional sludge flocs Granular sludge had a membrane permeability loss of 1.68-fold less than conventional sludge flocs during the constant pressure test In this study, a hybrid system includes a submerged MBR following a sequencing batch airlift reactor (SBAR) to filter the effluent This is named as a batch granulation membrane bioreactor (or BG-MBR) This combination was selected instead of inserting the membrane inside the granulation reactor because granular sludge was not stable in the continuous operation mode It is clearly proven that granules could be stable with the cyclic feast and famine conditions in a batch reactor (Beun et al., 2002; Tay et al., 2007) The advantages of this hybrid system include high organic and nitrogen removal efficiencies and fouling control This paper focuses on the investigation of high loading simultaneous organic and nitrogen removal and the fouling characteristics of the BG-MBR system Further, the fouling behavior of sludge fractions was also investigated Materials and methods 2.1 Experimental setup Fig describes the BG-MBR system including a SBAR (granulation reactor), a settler and a submerged MBR The SBAR which operated in batch mode consisted of four cycles of operation Air was supplied through a porous stone diffuser from the bottom of the reactor Each batch of operation consisted of four stages namely feeding (6 min), reaction (high aeration rate: h; and low aeration rate: 48 min), settling (3 min) and withdrawal (3 min) The high aeration rate is to achieve oxidation of organic and nitrogen compounds and granule stability Further, it was followed by low aeration to reduce the aeration cost and to enhance the nitrogen removal through the denitrification process occurring inside the core of the granule The denitrification process might be enhanced by limitation of oxygen diffusivity into the core of the granule The SRT was not controlled in this study because the suspended solids from SBAR effluent fluctuated according to time The second unit was the settler The settler is a dual purpose tank to function as both holding and settling tank (denoted as “settler”) The effluent of SBAR was transferred into the settler which was then fed into the MBR in a continuous mode of operation Settled sludge of 500 mL/ d (twice, each 250 mL) from the settler was removed periodically The final unit, the submerged MBR was used for the separation of liquid and solid fractions The remaining substrate, unsettled colloids and pin flocs could be further biologically degraded in the MBR All these systems were controlled automatically by programmable logic controller Table shows the operating conditions of BG-MBR system 2.2 Wastewater and support media The feeding wastewater contained 260 mg TOC/L (700 mg COD/ L as glucose), 190 mg N/L of NH4Cl, 50 mg/L of KH2PO4, 30 mg/L of CaCl2$2H2O, 12 mg/L of MgSO4$7H2O, and mg/L of FeCl3 throughout the experiment Trace elements were added at the rate of mL/L of wastewater as described by Thanh et al (2008) The shell carrier produced from the shell of white rose cockle was added to act as a support for microbial granule formation The carrier was used to enhance the structure, round shape, and physico-chemical characteristics of the granules The shells were dried, ground and sifted with sieve Nos 70 and 100, to reach a fraction between 150 and 212 mm The powder obtained was Influent valve Manometer Effluent valve Level control Permeate pump membrane SBAR MBR Settler Pump Air supply Fig Experimental set-up of BG-MBR system B.X Thanh et al / International Biodeterioration & Biodegradation 85 (2013) 491e498 Table Operating conditions of BG-MBR system Parameters SBAR Settler MBR Working volume (L) Size (length  diameter) 9.7 þ Down-comer: 120 cm  11.5 cm þ Raiser: 90 cm  cm 0.86 Ỉ 0.22 0.6 Ỉ 0.1 þ batches/d; h/batch: À Feeding: min; À Reaction: high aeration rate (3 h) followed by low aeration rate (48 min); À Settling: À Withdrawal: ỵ High aeration: 1.67 ỵ Low aeration: 0.08 7.3 w24 e e 53 cm  10 cm e e e e e Intermittent suction (7 on/ off) e 0.3 e e e e 3.4 20 PE 0.1 mm, 0.42 m2, Mitsubishi, Japan 2.8 L/m2 h OLR (kg TOC/m3 d) NLR (kg N/m3 d) Operating mode Air velocity (cm/s) HRT (h) SRT (d) Membrane module Membrane flux (net) then washed and dried before used in the experiment The carrier had a bulk density of 1.45 g/cm3 and a weight loss of 2% at 550 C within 20 Initially, the amount of the carrier added was 200 g (20 g/L) to SBAR Ten grams were added every month to compensate for the loss through sampling and effluent discharge 2.3 Analytical methods The supernatant samples from the settler and the MBR mixed liquor were prepared once a day by centrifugation at 4500 rpm for 15 (Centrifuge M/c Universal 320R, Germany) The procedure for getting supernatant sample was described by Bouhabila et al (2001) Dissolved organic carbon (DOC) was determined by a TOC analyzer (TOC-VCSN, Shimadzu, Japan) Parameters of ammonia, nitrite and nitrate were measured according to standard methods (APHA, 1998) (Total Nitrogen or TN ¼ NH3eN ỵ NO2eN ỵ NO3eN) In addition, the PN and PS were analyzed by methods of Lowry et al (1951) and Dubois et al (1956), respectively (EPS ẳ PS ỵ PN) The UVA254 was measured by using 1-cm quartz cell by UV/Visible Spectrophotometer (U-2001, Hitachi, Japan) where specific ultraviolet absorbance (SUVA) was calculated from the ratio of UV254 and DOC During the operation, sludge characterization was conducted for samples of granular sludge, mixed effluent from SBAR and MBR sludge SVI and MLSS measurements were determined according to standard methods MLVSS of shell granular biomass was not able to measure accurately by gravitational method according to standard methods For this kind of shell granules, measurement of MLVSS of gravitational method is not accurate due to the loss of shell carriers when mixed biomass (cell and carrier) are burned at temperature more than 450 C Thus the authors selected the indirect measurement method which measures the total organic carbon (TOC) of cell biomass and then converts into cell mass based on the cell formulae To measure MLVSS, sludge samples were ground for with the Ultra-Turrax machine (Ika-Werk, Germany) before homogenous sonication at 100 hz for (Ultra Sonic processor, CP130, USA) The sample was then diluted with milli-Q water and stirred in a volumetric flask for 10e20 at 493 500 rpm until homogenization occured Biomass in terms of MLVSS was determined by measuring TOC of the homogenized sample Then, the value of TOC was converted to MLVSS (multiplied by the factor 2.05) (Tijhuis et al., 1994) The particle size distribution of samples for the MBR sludge and settler were determined by the laser diffraction technique (MastersizerS, Malvern, UK, and detection range of 0.05e900 mm) The size of colloidal fraction was examined by the zetasizer nanoZS after centrifuging at 4500 rpm and C for (detection range of 0.6e6000 nm) The size of the granules was measured by a digital camera with a transparent scale located under the beaker containing the granules The bound EPS (bEPS) of granular sludge, the MBR sludge and fouling layer sample were extracted using the cation exchange resin technique (Dowex HCR-S/S, 16e50 mesh, sodium form, Dow Chemical Company) according to Frølund et al (1996) For granular sludge sample, it was ground by the Ultra-Turrax equipment for one minute before carrying out resin extraction The extraction was conducted with resin dosage (60 g/gVSS) and stirring speed of 600 rpm for 45 Then, the bEPS solution was centrifuged at 4500 rpm and C for 15 (twice) The amount of PS deposition on the membrane was quantified with the same method adopted by Kim and DiGiano (2006) Two fibres (about 10e30 cm) were cut off from the fouled membrane and washed with tap water until the membrane fibre became white/clean like cleaned (initial) membrane (removal of entire fouling layer attached on the fibre) The fibres were cut into small segments and immersed into a test tube containing ml milli-Q water Then the color reagent (1 ml of phenol 5% and ml of concentrated H2SO4) was added into the test tube This analytical procedure is similar to Dubois’ measurement method The PS deposition on the membrane was measured at a wavelength of 490 nm and converted to the unit of mg PS/cm2 of fibre Modified fouling index (MFI) and cake resistance was measured by a stirred cell (AMICON 8400 USA, diameter 67 mm, area ¼ 41.8 cm2) with a stirring speed of 500 rpm and a flat sheet membrane with pore size of 0.22 mm under a constant transmembrane pressure of one bar The raw experimental data (V and t) were used to plot t/V versus V graph to get the slope (s/L2) which represents the MFI of the sample The MFI is defined as the gradient of the linear region found in the cake filtration equation (Eq (1)) Cake resistance (1/m2) was estimated as multiplying by specific cake resistance a (m/kg) and cake mass C (kg/m3) (Boerlage et al., 2002) m$a$C m$Rm t ẳ Vỵ V A$TMP 2$A2 $TMP (1) Rc ¼ aC (2) In addition, the fouling potential of sludge fractions in the MBR, including suspended solids (SS), colloids (CL) and solutes (SL), were quantified The separation of SS, CL and SL were prepared according to Bouhabila et al (2001) The mixed liquor MBR sludge sample contained SS ỵ CL ỵ SL The sample, containing CL þ SL, was achieved by centrifugation at 4500 rpm and C for one minute Finally, the sample containing only SL was centrifuged twice at 4500 rpm and C for 15 Resistance of each fraction was calculated as follows: Rt ẳ Rm ỵ Rf (3) Rt, Rm, Rf are total, clean membrane and fouling resistance, respectively 494 B.X Thanh et al / International Biodeterioration & Biodegradation 85 (2013) 491e498 Results and discussions 3.1 Organic and nitrogen removal in the SBAR Aerobic granules were cultured from conventional activated sludge and with shell media Biomass started covering the surface of shell carriers during the first 10 days Shell granules started forming after 30 days of operation The granule size gradually increased and varied from 0.5 mm to 9.0 mm during next 80 days The average size of matured granule was 4.9 Ỉ 1.0 mm The color of granule changed from light yellow (initial granules) to dark yellow (mature granules) The average settling velocity was 260 Ỉ 124 m/h which was much higher than that of activated sludge (1e2 m/h) Granules were stable during the study period The study on nitrogen removal and fouling behavior below was conducted since granules matured in the reactor (i.e., after 110 days) At the stage of matured granule formed, organic matter was quickly removed by an aerobic granulation system in each batch Fig shows evolution of concentrations of organic matter and nitrogen species with time in a typical batch 91% of TOC removal was achieved within the first 30 and 94% removal reached during the 90 of aeration stage The dissolved oxygen (DO) concentration during the high aeration stage was saturated during first h of the stage of high aeration rate (w6.5 mg/L) and then reduced to mg/L during the next 48 due to lower aeration rate The pH of SBAR slightly fluctuated after h of operation due to simultaneous nitrification (alkalinity consumption) and denitrification (alkalinity production) in the outer and the inner layer of granules, respectively The influent ammonia was converted into nitrite and nitrate where nitrite was dominant due to partial nitrification in the SBAR The complete nitrification did not occur due to the free ammonia concentration in the reactor The free ammonia inhibited the nitratation (Anthonisen et al., 1976) Complete nitrogen removal (converted to nitrogen gas) was observed to occur during first TOC (mg/L) 300 Low aeration rate High aeration rate 250 200 150 TOC 100 DO 240 30 60 90 120 150 Time (min) 180 210 200 Concentration (mg/L) NH4-N NO2-N NO3-N TN 160 120 80 40 0 30 60 90 120 150 Time (min) pH 50 180 210 240 Fig TOC, DO, pH and nitrogen species profile of SBAR in a batch DO (mg/L); pH 350 30 in which the bulk liquid was rich in organic and nitrite This reveals that the denitrification process could be achieved in the granules when organic substrate is available The simultaneous nitrification-denitrification (SND) could take place in both low and high aeration stages depending on the availability of organic substrate due to the structure of the granule The DO in the outer layer of the granule was almost as high as the bulk liquid while that of the inner core was very low due to the limitation of oxygen transfer from the bulk liquid to the core of granules This phenomenon allowed the denitrification process to occur inside the core of the granule As reported by Tijhuis et al (1994), the anoxic/anaerobic condition could be achieved at the depth of 300 mm below the granule surface In this study, the average size of the granules was about (4.7 Ỉ 1.4 mm) whose radius was almost much greater than the diffusion depth of oxygen inside the core of granule This could lead to the anoxic condition in the core Generally, the special spherical structure of granules favors the SND phenomena to occur in the single aerobic granulation reactor even with bulk DO concentration higher than mg/L Fig shows that the ammonia nitrogen was completely oxidized into nitrite nitrogen during the first three hours of operation in which the removal rate was 0.015 mg N/gVSS h (0.18 mg N/ L h) The nitrite production rate was 0.013 mg N/g VSS h (0.16 mg N/ L h) where it was converted from free ammonia with time Conversion of nitrite nitrogen into nitrate nitrogen was not significant in SBAR due to the existence of free ammonia in the reactor Previous research reported that free ammonia inhibition threshold was 0.1e4.0 mg/L for nitrobacter which plays a role in the oxidation of nitrite (Yang et al., 2004) In this process, the dynamic balance of nitrogen species occurred between ammonia consumption and nitrite production The concentration of TN did not change drastically during the last 3.5 h which could conclude that the SND reached its maximum efficiency at this operating condition The SND only occurs during the first duration where the organic substrate is available The removal efficiency of TN becomes insignificant since available organic matter is limited in the SBAR Some other research also found that the nitrite-oxidizing bacteria (NOB) are not favorable in granular sludge Li et al (2013) reported that compared to sludge flocs sludge granulation with selective sludge discharge help halt ammonia oxidation to the level of partial nitrification rather than complete nitrification This is also confirmed by the molecular analysis that aerobic granulation resulted in ammonia-oxidizing bacteria (AOB) enrichment and reduction of nitrite-oxidizing bacteria (NOB) In addition, Shi et al (2011) also postulated that a fairly large proportion of AOB was close to the granule surface but NOB were rarely found The granules had excellent partial nitrification ability due to inhibition of free ammonia (FA) and limited DO diffusion within granules The data set at steady state during 44 days was used for nitrogen balance The result shows that the TN removal is 59% which includes 12% TN removal by biological assimilation and 47% by the SND process in the SBAR The overall denitrification rate is 22.2 mg N/L/h (1.76 mg N/g VSS h) under aerobic condition without external substrate addition in the reactor In practice, complete nitrogen removal could be fully achieved in the availability of electron donor or lower ammonia concentration in the feed The organic matter removal and SND in the SBAR indicates that there was co-existence of heterotrophic, nitrifying and denitrifying population in the structure of aerobic granules Nitrogen removal could take place even in aerobic granulation reactor (DO greater than mg/L) The results indicate that the complicated anaerobic/ anoxic/aerobic system could be integrated in a single aerobic granulation reactor Table summarizes the treatment performance of BG-MBR system most of the organic matter was removed in the SBAR B.X Thanh et al / International Biodeterioration & Biodegradation 85 (2013) 491e498 Table Treatment performance of BG-MBR Parameters SBAR Settlera MBRb DOC removal (%) Ammonia removal (%) Overall TN removal (%) SVI (mL/g) F/M (dÀ1) MLVSS (mg/L) Settling velocity (m/h) Average particle size (% volume) 97.3 99.9 59 25 (Ỉ5) 0.18 (Ỉ0.05) 12,600 (Ỉ240) 260 (Ỉ125) 4.9 (Ỉ0.2) mm e e 19 e e 35 (Ỉ15)