w a t e r r e s e a r c h 5 0 ( 2 0 1 4 ) 1 1 4 e1 2 3

Available online at www.sciencedirect.com

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

Performance of a submerged anaerobic membrane bioreactor with forward osmosis membrane for low-strength wastewater treatment Lin Chen a,b,*, Yangshuo Gu b, Chuqing Cao c, Jun Zhang d, Jing-Wen Ng b, Chuyang Tang b a

Advanced Environmental Biotechnology, Nanyang Technological University, Singapore 637174, Singapore Singapore Membrane Technology Centre, Nanyang Technological University, Singapore 639798, Singapore c Department of Electrical and Computer Engineering, National University of Singapore, 117576, Singapore d School of Municipal and Environmental Engineering, Harbin Institute of Technology, Harbin 150090, China b

article info

abstract

Article history:

A submerged anaerobic membrane bioreactor with forward osmosis membrane (FO-

Received 12 August 2013

AnMBR) was operated at 25
C for the treatment of synthetic wastewater. As the experi-

Received in revised form

ment progressed, the water flux reduced due to the membrane fouling and the increasing

3 December 2013

salinity in the reactor, and achieved at around 3.5 LMH in one cycle. It was worth noting

Accepted 5 December 2013

that the level of salinity in the reactor was not a concern in terms of inhibition or toxic

Available online 14 December 2013

effects on the biological processes. The FO-AnMBR process exhibited greater than 96% removal of organic carbon, nearly 100% of total phosphorus and 62% of ammonia-nitrogen,

Keywords:

respectively, suggesting a better removal efficiency than the conventional anaerobic

Forward osmosis membrane

membrane bioreactor. The methane and carbon dioxide compositions achieved concen-

Anaerobic bioreactor

trations of around 65%e78% and 22%e35%, respectively; and no obvious difference in the

Salinity

biogas composition was observed with the changes of conductivity. With respect to the

Wastewater treatment

methane yield, an average value of 0.21 L CH4 g1 COD was obtained, exhibiting the feasibility of energy recovery by this FO-AnMBR system. Additionally, an increase in the salinity enhanced the accumulation of soluble microbial products, especially for the proteins with 88.9% increment as the conductivity increased from 1.2 to 17.3 ms cm1. In contrast, a relatively stable concentration of extracellular polymer substances (EPS) was observed, indicating that the influence of conductivity on EPS cannot be directly correlated. ª 2013 Elsevier Ltd. All rights reserved.

1.

Introduction

In recent years, increasing pressure on designing anaerobic reactors, such as reducing the footprint, separating hydraulic retention time (HRT) from the solids retention time (SRT) and minimizing environmental impacts, led to the development of

the anaerobic membrane bioreactor (AnMBR) (Stuckey, 2012). The conventional AnMBR with a long SRT and microfiltration membrane (or ultrafiltration membrane) offered numerous advantages, such as improving effluent quality, reducing waste biosolids production and strengthening methane conversion (Huang et al., 2011). However, small molecular weight substances (e.g. natural organic matter) and trace

* Corresponding author. Advanced Environmental Biotechnology, Nanyang Technological University, Singapore 637174, Singapore. E-mail address: [email protected] (L. Chen). 0043-1354/$ e see front matter ª 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.watres.2013.12.009

w a t e r r e s e a r c h 5 0 ( 2 0 1 4 ) 1 1 4 e1 2 3

contaminants might escape into the effluent, which would be the major barrier to the reuse extent of AnMBR permeate (such as in the drinking water). Additionally, high energy consumption due to the high pressure pumps used in the filtration was another drawback for the conventional AnMBR system; therefore, to explore other plausible technologies with lower energy requirement was necessary. Recent achievements in the membrane technology have demonstrated that the emerging forward osmosis (FO) membrane process was a potential and effective alternative to conventional membrane processes in seawater desalination and water reclamation. It was a natural process driven by the osmotic pressure difference that retained solutes but allowed water to permeate through a semi-permeable membrane (Cath et al., 2006). Compared with pressure-driven processes, FO was a relatively low fouling treatment option for the absence of hydraulic pressures, and the foulant compaction might be milder due to the utilization of osmotic pressure to extract water (Achilli et al., 2009). More importantly, FO process demonstrated better water quality because of a double barrier, which exhibited remarkable removal efficiency for salts (e.g., Ca2þ, above 95%), ammonia (74%), nitrate (78%), sulfamethoxazole (90%), carbamazepine (83%), trace organics (w80%) and so on (Alturki et al., 2012; Cath et al., 2009; Heo et al., 2013; Jin et al., 2012). Owing to these advantages, several attempts have been made to the development of forward osmosis membrane bioreactors (FO-MBR) combining the biological and FO processes (Achilli et al., 2009; Cornelissen et al., 2008; Zhang et al., 2012), which have demonstrated acceptable permeate flux and remarkable removal efficiency for organic compounds. The high rejection capacity of the FO membrane can effectively retain small and persistent trace organic contaminants in the reactor, thus significantly prolonging the retention time and subsequently facilitating their biodegradation. However, previous researches only focused on the aspect of aerobic bioreactor, wherein the costs of aeration and sludge handling remained as the major disadvantages. Considering that AnMBR can provide the same benefits as MBR (Smith et al., 2012), it was reasonable to suppose that the anaerobic membrane bioreactor with forward osmosis membrane (FOAnMBR) retained the inherent advantages of FO-MBR but with reduced energy requirements and lower biomass yield. Nevertheless, to the best of our knowledge no studies have investigated the performance of FO-AnMBR. In this study, the performance of a laboratory-scale FOAnMBR system fed with synthetic wastewater and operated at 25
C was evaluated to report on water flux, reverse salt transport, nutrient removal, volatile fatty acids (VFAs) production and gas composition. Simultaneously, the characteristics of soluble microbial products (SMP) and extracellular polymer substances (EPS) with the increase of conductivity were investigated.

2.

Materials and methods

2.1.

FO-AnMBR configuration and operating conditions

A laboratory-scale anaerobic membrane bioreactor with 3.6 L of working volume was run at 25
C (Fig. 1), which was equipped with pH, conductivity, pressure and oxidation-reduction

115

potential (ORP) monitoring units (Mettler-Toledo M200 system). A flat-sheet membrane module made of cellulose triacetate (CTA) membranes (Hydration Technologies Inc.) with 0.025 m2 was submerged in the tank. The membranes were oriented with active side facing the reactor and support sides facing the draw solution. A synthetic wastewater simulating municipal wastewater was used as feed water (see Table S1). The influent pump was controlled by a water level sensor to maintain a constant water level in the reactor. Produced biogas was recycled through gas diffuser both to mix the biomass and scour the membrane surface for fouling control, and the recirculation rate was controlled at 2 L min1. A 0.5 M NaCl solution was used as the draw solution (with the conductivity in a range of 45.0e45.5 ms cm1), which was maintained by conductivity control connected to a 5 M NaCl solution tank. The flow rate of draw solution was kept at 0.4 L min1 to minimize the effect of internal concentration polarization. The permeate flux was derived by mass balance to account for the mass of 5M NaCl dosed into the draw solution tank, and then normalized for the membrane area. During the entire FO-AnMBR operation, the sludge retention time (SRT) was kept at 90 days, and the hydraulic retention time (HRT) was in a range of 15e40 h depending on the membrane flux.

2.2.

Analytical methods

Mixed liquor suspended solids (MLSS), volatile suspended solids (VSS) and total phosphorus (TP) were determined using Standard Methods (APHA., 1998). Chemical oxygen demand (COD), total nitrogen (TN), and ammonia concentration (NHþ 4 eN) were analyzed using HACH USEPA reactor digestion method (HACH 2125815/2415815), persulfate digestion method (HACH 2714100/ 2672245) and salicylate method (HACH 2606945), respectively. The gas production rate was measured by the liquid displacement method; additionally, the composition of gas was determined using a Agilent GC-TCD fitted with J&W 113-4362 column (0.32 mm  60 m, 0 mm) and Agilent 19095P-MS6 column (0.53 mm  30m, 50 mm). VFAs were analyzed by PerkineElmer HPLC system with Hþ cation exchange column (HAMILTON, 305  7.8 mm, 8e10 mm) and UVeVis detection at 210 nm. The sludge from the bulk phase was harvested by centrifugation (4000 rpm, 10 min), washed with water and then resuspended in sterilized deionized water for analysis of EPS using heat treatment (Morgan et al., 1990). The proteins and carbohydrates were determined using the modified Lowry method (Frolund et al., 1995) and the phenol-sulphuric acid method (Dubois et al., 1956), respectively. The threedimensional excitation-emission matrix (EEM) spectroscopy (LS55, PerkineElmer Co.) was applied to characterize the SMP under different conductivities. Particle size distribution of the anaerobic mixed liquor was measured by laser scattering with a detection range of 0.02e2000 mm (Mastersizer 2000, Malvern).

3.

Results and discussion

3.1.

Flux performance and salt accumulation

Changes of membrane flux and conductivity against operation time are illustrated in Fig. 2. In general, the membrane

116

w a t e r r e s e a r c h 5 0 ( 2 0 1 4 ) 1 1 4 e1 2 3

Fig. 1 e Schematic diagram of the FO-AnMBR system.

flux decreased from 9.5 to 3.5 LMH, while the conductivity in the reactor increased from 1.0 to 20.5 ms cm1 within a time interval of approximately 22 days (one cycle). The supernatant was discharged and replaced with influent at this point as the conductivity dropped back to around 1.0 ms cm1, and then the cycle continued. To exhibit this progression, a period from day 47 to day 69 was described in detail, which would be used to showcase this repeating sequence. The membrane flux decreased at a faster pace from initial 9.21 to 6.54 LMH at a rate of 0.89 LMH per day to day 50 at which point the rate slowed down to 0.12e0.31 LMH per day. It remained at this rate until day 66 when the membrane flux reached 3.67 LMH, and then the flux kept relatively stable until day 69. On the other hand,

the conductivity in the reactor followed an opposite trend. Conductivity of the bulk phase after 3 days of continuous operation was 6.95 ms cm1, corresponding to approximately 3.72 g NaCl L1. Then, the incremental rate of conductivity decreased to 0.4e0.8 ms cm1 per day and continued at this rate when the conductivity reached around 18.8 ms cm1. Compared with the relative stable flux achieved during day 66e69, the conductivity still increased in a slower pace which might be due to the influent salt accumulation. Obviously, the decline of membrane flux was contributed to the membrane fouling and the increasing salinity in the reactor; wherein the latter could be due to phenomena including the retention of feed solutes and the reverse

Fig. 2 e Changes of (a) conductivity in the bulk phase and (b) membrane flux during long-term operation.

w a t e r r e s e a r c h 5 0 ( 2 0 1 4 ) 1 1 4 e1 2 3

diffusion of draw solution. The decrease of conductivity in the reactor would only be through the sludge wasting; thus the SRT regulated its value. The salt concentration superimposed by the accumulation of inflow and reverse diffusion of draw solution was calculated as shown in Fig. S1, which demonstrated that the reverse transport of NaCl from the draw solution played a critical role in initial phase of flux decline. The other reason for the flux decline was the increased hydraulic resistance caused by foulants attached onto the membrane surface. Fig. S2 shows photographs of the clean membrane and the used membrane upon immediate removal from the experimental system, exhibiting a thin layer of foulants on the membrane surface. The cake layer could improve the rejection of feed constituents but also increase the permeation resistance. The synergistic interactions between polymers and ions (e.g., Ca2þ, Mg2þ) reduced the mass transfer coefficient, causing a decrease of membrane flux. More importantly, cake-enhanced osmotic pressure was a major factor to overall flux decline for salt-rejecting membranes (Hoek and Elimelech, 2003). Since the hindrance of the salt back-diffusion by the cake layer, there would be a greater salt build-up near the membrane surface, thereby accelerating cake-enhanced osmotic pressure. The elevated osmotic pressure near the membrane surface led to a substantial drop in the net driving force and, thus, resulted in a significant decline in permeate flux. After one cycle (about 22 days) operation, the membrane was rinsed with tap water and the supernatant was discharged until the conductivity down to 1.0 ms cm1. It can be seen that the flux could be recovered to 94% of the initial flux (day 78: flux ¼ 9.34 LMH; day 100: flux ¼ 8.82 LMH), which was possibly due to the smooth membrane surface and no internal fouling. Such high recovery of flux matched literature study which stated that FO systems were competent in reviving flux level back to more than 90% of its initial flux (Mi and Elimelech, 2010). However, the membrane could only be used for around two cycles, and then the conductivity in the reactor would sharply increase from 1.1 to 20 ms cm1 in 6 days (e.g. day 70e76). It might be contributed to the membrane of cellulosic material that was known to be chemically and microbiologically vulnerable (Lay et al., 2011); therefore, the selection of anti-biodegrading membrane and the alleviation of salt accumulation would be conducted in our further research.

3.2.

Performance of the FO-AnMBR

3.2.1.

COD removal and VFA production

The biomass concentration was kept at around 39004600 mg VSS L1 by means of sludge wastes. Fig. 3(a) demonstrates the changes of COD concentration in the influent, supernatant and permeate, as well as the COD removal efficiency. The permeate COD was lower than 15 mg L1, and the COD removal rate during this period was an average of 96.7% corresponding to the influent COD concentration of 460 mg L1. Obviously, this COD removal rate was much higher than the previously reported permeate COD of approximately 65e100 mg L1 in conventional AnMBR (Sutton et al., 2004), which might be resulted from the differences in the membrane configuration: microfiltration versus FO membrane (non-porous). This result was consistent with

117

Achilli et al. (2009) that the semi-permeable FO membrane exhibited 98% rejection of TOC due to its non-porous composition compared with microporous membranes with typically 28%e87% retention ability. With respect to the supernatant, the COD increased from 103 to 210 mg L1 in the initial phase of one cycle (day 47e69), which might be contributed to the following reasons: 1) the release of SMP with the increasing conductivity; and 2) the membrane interception. The hydrated radii of Naþ (0.36 nm) and Cl (0.33 nm) were comparable to that of the membrane pore radius (Alturki et al., 2012); hence, the reverse salt flux could hinder the pore forward diffusion of the organic solute, leading to a lower adsorption of organics within the membrane and subsequently higher rejection in the FO mode. Then the COD of supernatant reached relatively stable (w185 mg L1) with the further extension of operation time, which was mostly due to the extension of HRT. Many of the compounds exiting were degradable over time, both aerobically and anaerobically, but the HRT was not long enough for them to be degraded effectively (Schiener et al., 1998). The high rejection of FO membrane and the decrease of membrane flux prolonged the retention time of organics in the reactor, thus the accumulated organics can be partially degraded which might explain the phenomenon that the supernatant COD would be stable or even decrease with the further extension of operation time. Furthermore, the impact of gradual increase of salt concentration on organic removal was small, wherein the removal of COD only decreased by around 17.8% compared with the 59% reduction in Johir’s study (2013) when the conductivity increased from 1.5 to 20 ms cm1. The VFAs in the mixed liquor of the reactor were measured, including acetic (C2), propionic (C3), butyric (C4) and valeric acid (C5). As can be seen in Fig. 3(b), only C2 and C3 can be found in the mixed liquor of the reactor, while other VFAs, such as the butyrate and isovalerate, were hard to detect and therefore not depicted in the figure. Acetic acid represented the main proportion of the VFAs in the mixed liquor of the reactor with a concentration varying from 0 to 29.8 mg L1. An increase in the C2 concentration can be observed in the initial phase of one cycle, with a concentration up to 17.2e29.8 mg L1. Obviously, the increase of C2 was due to the discharge of supernatant in each cycle which definitely affected the acetate up-take rates of the methanogenic bacteria; and then the C2 concentration decreased and kept in a range of 0e11.7 mg L1 with the extension of operation time. However, the C3 concentration in the mixed liquor was relatively stable and remained lower than 7.4 mg L1 all the time. Noticeably, the total VFA concentration always kept in a lower level comparing to other reports and did not exhibit a direct relationship with the conductivity, which was in strong contrast with the previous research that an increase in the VFA concentration came along with the increase of conductivity (Jeison et al., 2008). Besides the biomass acclimatization in the saline environment, this phenomenon was partly due to  the configuration of the FO membrane. Part of the CO 3 =HCO3 ions was intercepted by the FO membrane and the pH in the reactor increased from 6.9 to 7.6 in one cycle operation, thus the alkaline environment was helpful to avoid the acid accumulations in the reactor and maintain the stability of the methanogenic process.

118

w a t e r r e s e a r c h 5 0 ( 2 0 1 4 ) 1 1 4 e1 2 3

Fig. 3 e (a) Evolution of the COD concentrations in the feed, bulk and the permeate, and the COD removal percentage; (b) Changes of volatile fatty acids in the bulk phase.

3.2.2. Changes of NHþ 4 eN, TN and TP in the reactor and permeate Fig. 4(a) exhibits the evolution of NHþ 4 eN and TN concentrations in the influent, bulk phase and permeate. With an inflow 1 and TN of NHþ 4 eN concentration of 16.2  2.1 mg L 1 42.7  4.2 mg L , the accumulation of both could be easily observed in the bulk phase due to the excellent rejection of FO membrane. In one cycle (take day 47e69 as an example), the TN concentration in the reactor firstly increased from 48.7 to 101.8 mg L1 and then gradually decreased to 84.5 mg L1; simultaneously, the NHþ 4 -N concentration exhibited the same trend as the TN with a peak around 96.2 mg L1. The decrease of NHþ 4 -N might be contributed to the formation of ammonia calcium/magnesium phosphate. It was also found that NHþ 4 eN was the major form of TN in the bulk phase and accounted for over 90%, indicating the activity of anaerobes in the hydrolysis phase. With respect to the permeate, the concentrations of TN and NHþ 4 eN overlapped with each other, demonstrating that the NHþ 4 eN was the sole form of the nitrogen source in permeate with a total rejection of organic nitrogen by the FO membrane. Additionally, the concentration 1 of NHþ 4 eN in the permeate was in a range of 6.8e36.3 mg L and about 62.7%e81.2% lower than the supernatant concentration between 36.2 and 98.2 mg L1, corresponding to a membrane rejection rate of above 60%, which was in line with

previous literature that the membrane rejection efficiency for NHþ 4 eN could reach up to 74%e80% (Yap et al., 2012). Considering the removal efficiency of TN and NHþ 4 eN by the FO-AnMBR, the removal rate still compared favorably against conventional AnMBR with nearly zero removal. The changes of total phosphorus against operation time in the influent, bulk phase and permeate are illustrated in Fig. 4(b). With an inflow of 4.1  0.8 mg L1, FO-AnMBR had a nearly complete rejection of phosphorus. Simultaneously, the accumulation of TP can be observed in the reactor. In one cycle (e.g. day 47e69), the concentration of TP increased from 3.3 to 19.9 mg L1 in 7 days and followed by a concentration decline. There was a high likelihood that the decline in concentration was due to precipitation of phosphorus when the concentration was high enough. Due to the high rejection property of the selective membrane, C, P, Ca, Mg and N were the main elements to be observed in the reactor, which might be responsible for the formation of scaling (e.g., Ca3(PO4)2) as observed before. The similar phenomenon can be found in other researches. Trzcinski and Stuckey (2009) investigated the morphology and structure of the inorganic precipitate in an AnMBR, and suggested that the high amount of phosphorus could either be dicalcium phosphate dihydrate (Ca2H2(PO4)2) or amorphous calcium phosphate (Ca3(PO4)2).

w a t e r r e s e a r c h 5 0 ( 2 0 1 4 ) 1 1 4 e1 2 3

119

Fig. 4 e (a) Evolution of NHD 4 eN and TN concentrations in the influent, bulk phase and permeate; (b) Changes of TP against operation time in the influent, bulk phase and permeate.

3.2.3.

Gas production

Independent on the buildup of salinity, the stable removal efficiency of COD by the reactor confirmed that the FO-AnMBR system was biologically active throughout the experiment. Additionally, the production of biogas and its potential use as source of energy was one of benefits for the anaerobic treatment. Fig. 5(a) describes the biogas composition and methane yield rate (volume of methane produced per gram COD removed). No significant difference in the biogas composition was observed with the changes of conductivity; and the methane and carbon dioxide compositions achieved concentrations of around 65%e78% and 22%e35%, respectively. With respect to the methane yield, an average value of 0.21 L CH4 g1 COD was obtained, representing 58% of the maximal theoretical value. Except the FO-AnMBR system was operated at 25
C, the difference could be partly explained by the solubility of biogas during the gas recirculation, which was not calculated in this study due to the small volume of waste sludge and the loss of dissolved methane in the effluent. Methane losses in the range from 28% to 39% under ambient temperature conditions were also reported (Singh and Viraraghavan, 1998). Additionally, the lower production of methane compared with the theoretical value also indicated that the methanogenesis was partially inhibited under high salinity conditions. Especially in the later phase of one cycle, insignificant decrease of methane production rate at high salinity could be observed, which might be due to the

consumption of substrate by anaerobic biomass to generate compatible solutes and extracellular polysaccharides to survive under high osmotic conditions. However, the methane yield rate found in this study was still comparable to the values reported in previous literature, indicating the feasibility of energy recovery by this FO-AnMBR system. For instance, methane yield from 0.23 to 0.27 L CH4 g1 COD was observed in AnMBR reactor treating municipal wastewater at 20e35
C (Martinez-Sosa et al., 2011). Methane yield in a range of 0.16e0.2 L CH4 g1 COD was estimated in an upflow anaerobic sludge blanket (UASB) reactor treating municipal wastewater in a temperature range from 11 to 32
C (Singh and Viraraghavan, 2003). These results showed that the gas production in FO-AnMBR system was obtainable while optimization of gas production rate required further investigation.

3.3.

Characterization of SMP and EPS

3.3.1.

Changes of SMP with salinity

The variations of carbohydrates and protein contents in SMP against the operation time are illustrated in Fig. 6(a). The carbohydrates content seemed to be less affected by the conductivity ranging from 2.82 to 6.16 mg L1. In contrast, the presence of higher conductivity caused a greater accumulation of proteins, from 26.1 mg L1 for 1.2 ms cm1 (day 47) to 49.3 mg L1 for 17.3 ms cm1 (day 62). As the conductivity further increased, the accumulation rate of proteins slowed

120

w a t e r r e s e a r c h 5 0 ( 2 0 1 4 ) 1 1 4 e1 2 3

Fig. 5 e Changes of biogas composition and methane yield rate during long-term operation.

down or the concentration even kept relatively stable, corresponding to the changes of COD in the bulk phase. While many SMP produced during treatment were degraded, there was a small fraction which due to their chemical structure, e.g. substituted ring compounds, cross-linked cell wall fragments, was difficult to degrade and intercepted by the FO membrane, thereby increasing the SMP concentration. Additionally, it was found that biomass exposed to high salinity produced higher molecular weight (MW) compounds (Vyrides and Stuckey, 2009), which were more slowly degraded resulting in an increase of SMP in the reactor. EEM fluorescence spectra (Fig. 7(aeh)) of SMP were also analyzed in order to clarify the compounds in the SMP and

their behaviors with the changes of conductivity. They were similar in the peak locations at various conductivities, but had different fluorescence intensities. The first peak was located at the excitation/emission wavelength (Ex/Em) of 330/422 nm (Peak A) and the second peak (Peak B) was at Ex/ Em of 247/409 nm, which were attributed to the humic acidlike and fulvic acid-like substances, respectively. The other two peaks were located at Ex/Em of 243/380 nm (Peak C) and 289/380 nm (Peak D), which were associated with the tryptophan and tryptophan-like proteins respectively. The comparison of the EEM at different conductivities revealed that, although the fluorescence peaks locations did not change significantly with the salinity, the peak intensities

Fig. 6 e (a) Variations of carbohydrate and protein contents in SMP against the operation time; (b) Variations of carbohydrate and protein contents in EPS against the operation time.

w a t e r r e s e a r c h 5 0 ( 2 0 1 4 ) 1 1 4 e1 2 3

were substantially influenced by conductivity level, especially the Peak A. In order to further exhibit the composition changes of SMP with the salinity, the fluorescence regional integration (FRI) method was employed to analyze the EEM spectra (Chen et al., 2003; Li et al., 2013) and the FRI distribution at various conductivity levels is illustrated in Fig. 7(i). The Regions I, II, III and IV represented the tyrosine, tyrosine-like proteins, tryptophan and tryptophan-like proteins, respectively. The Region V and VI were attributed to the fulvic acid-like and humic acid-like substances. It could be seen that the SMP were dominated by fluorescence in Regions III, IV, V and VI. The sum of Regions III and IV took up for more than 49.2%, whereas the combination of Regions I and II only accounted for less than 9.3%, suggesting that the majority of proteins in the SMP was existed in the form of tryptophan and tryptophan-like proteins, other than the tyrosine and tyrosine-like proteins. Additionally, each region exhibited different changing trends with the conductivity, wherein no obvious changes were observed for Regions I and II. With respect to Region III and IV, a significant

121

increase in tryptophan-like proteins from 23.1% to 27.1% was observed when the conductivity increased from 3.76 to 18.0 ms cm1, while a decrease in tryptophan from 27.9% to 22.1% was found. Region V and VI also exhibited the opposite changing trends with the increasing salinity, wherein the fulvic acid-like substances reduced from 25.0% to 21.7% and the humic acid-like substances increased from 14.6% to 21.7%. The fluorescence characteristics of SMP were different from other research that contained more protein-like organics but tiny amount of humic/fulvic-acid-like substances under a high salinity shock (5% NaCl) (Wang and Zhang, 2010), which might be due to the differences in the reactor operation. However, these findings revealed that more protein-like matters emerged with the increasing salinity level.

3.3.2.

Changes of EPS with salinity

Fig. 6(b) compares the EPS composition at the different conductivities during the reactor operation, wherein the concentrations of proteins and carbohydrates are 43.1  3.6 mg g1 VSS and 11.5  2.1 mg g1 VSS, respectively.

Fig. 7 e EEM fluorescence spectra of SMP (a) conductivity [ 3.76 ms cmL1; (b) conductivity [ 7.05 ms cmL1; (c) conductivity [ 8.90 ms cmL1; (d) conductivity [ 10.6 ms cmL1; (e) conductivity [ 13.2 ms cmL1; (f) conductivity [ 14.3 ms cmL1; (g) conductivity [ 15.9 ms cmL1; (h) conductivity [ 18.0 ms cmL1 m and (i) FRI distribution at various conductivity levels.

122

w a t e r r e s e a r c h 5 0 ( 2 0 1 4 ) 1 1 4 e1 2 3

Surprisingly, no positive or negative relationship can be observed between the EPS and conductivity, although the concentrations of proteins and carbohydrates exhibited some fluctuations resulting from the release and synthesis of EPS with the changes of conductivity. The salinity caused the release of EPS due to the increase of osmotic pressure during metabolism, enhancement of cell lysis, or stimulation of efflux mechanism. On the other hand, microorganisms utilized the substrates to generate EPS, acting as a diffusion barrier between the cell wall and extreme environments under sodium toxicity. Results also showed that no correlation between the SMP and EPS fraction was observed, suggesting that the salinity does not appear to have any significant effect on the distribution of bound and unbound proteins/carbohydrates. In contrast, most previous literature stated that the high sodium concentrations in wastewater induced salt stress to microbial species, resulting in a palpable impact on EPS. Ismail et al. (2010) found that the production of EPS carbohydrates at 10 g Naþ L1 was double that at 20 g Naþ L1 concentrations in the UASB reactors. The difference might be due to the fact that the previous literature focused on the effects of shock/continuous salt loadings on the batch or continuous flow activated sludge systems and concluded that high salinity shocks caused a drastic increase in the EPS production. However, in the FO-AnMBR system, the salinity gradually increased to 20 ms cm1 within 22 days, allowing an initial lag period for the microbial population to adapt to the saline environment, thus exhibiting minor impact on the EPS production. The production of EPS in the FO-AnMBR system was comparable to the values associated with the anaerobic reactor, even under the fluctuating salinity condition. Huang et al. (2011) investigated the EPS production in a submerged AnMBR for low-strength wastewater treatment, and reported that the concentrations of carbohydrates and proteins were in the range of 10e15 mg g1 VSS and 28e50 mg g1 VSS at the SRT of 30 days to infinite days and HRT of 8e12 h. Additionally, as the published literature generally indicated that microorganisms responded to a salt shock by aggregation of the individual cells (Reid et al., 2006) and EPS composition closely affected particle flocculation (Huang et al., 2011), the particle size distribution was analyzed as demonstrated in Fig. S3. A quite similar distribution of particle size distribution was observed with median particle size of 47.9 mm, indicating that the influence of conductivity on particle size cannot be correlated. These results further demonstrated that the EPS concentration in the FO-AnMBR system was not significantly affected by the conductivity changes.

3.4.

Challenges and perspectives

Compared to conventional AnMBR systems, the FO-AnMBR system was able to produce higher quality permeate and exhibited comparative economical advantage in the wastewater treatment. Additionally, a sudden removal of salinity from biomass which had been exposed to high saline conditions did not seem to cause any severe effects on the reactor operation. However, as a high rejection system, there would be challenges associated with the use of FO-AnMBR systems, especially in the aspects of the membrane, biology and the

supernatant disposal. As stated before, the CTA membrane had a low tolerance to both the high temperature solution and the biological attachment, hence the uncertainty of the stability of membrane made it difficult to sustain the operation for a long time. Additionally, considerable physiological and physical-chemical-biological challenges would be expected with the increasing salinity, which might reduce the effective driving force, decrease the microbial kinetics, yield unfavorable parameters for salt precipitation or solute interaction, and aggravate the membrane fouling (Lay et al., 2010). With respect to the inorganic-rich (e.g. nitrogen, phosphorous and metal) supernatant discharging from the reactor, the high concentration of metal might limit the post-treatment technology or the reuse extent such as in the agriculture irrigation. Therefore, future work should focus on the following aspects: - Selection and development of an optimized FO membrane for the FO-AnMBR system. - Understanding and alleviation of salinity effects on the physciobiological performance of FO-AnMBR system. - Exploiting a suitable process to dispose or reuse the inorganic-rich supernatant.

4.

Conclusions

The FO-AnMBR was successfully operated at 25
C to treat synthetic wastewater, and the following conclusions are drawn on the basis of the experimental results: (1) Within a time interval of approximately 22 days (one cycle), the membrane flux decreased from 9.5 to 3.5 LMH; simultaneously, the conductivity in the bulk phase increased from 1.0 to 20.5 ms cm1. The reduction of membrane flux was contributed to the membrane fouling and the increasing salinity in the reactor. (2) The increment of salinity in the reactor did not exhibit toxic effects on the wastewater treatment, wherein more than 96% removal of organic carbon and averaged 0.21 L CH4 g1 COD were obtained for the FO-AnMBR system. Additionally, the removal rates for NHþ 4 eN and TP were up to 60% and 100%, respectively, due to the excellent interception of FO membrane. (3) Results from SMP revealed that an increase in the salinity enhanced the accumulation of SMP, especially for the tryptophan and tryptophan-like proteins. However, no positive or negative relationship can be observed between the EPS concentration and the conductivity, wherein the concentrations of proteins and carbohydrates were 43.1  3.6 mg g1 VSS and 11.5  2.1 mg g1 VSS, respectively.

Acknowledgements The authors acknowledge the financial support of the Environment and Water Industry Programme Office, Singapore (IRIS-08-06) for this project.

w a t e r r e s e a r c h 5 0 ( 2 0 1 4 ) 1 1 4 e1 2 3

Appendix A. Supplementary data Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.watres.2013.12.009

references

Achilli, A., Cath, T.Y., Marchand, E.A., Childress, A.E., 2009. The forward osmosis membrane bioreactor: a low fouling alternative to MBR processes. Desalination 239 (1), 10e21. Alturki, A.A., McDonald, J.A., Khan, S.J., Price, W.E., Nghiem, L.D., Elimelech, M., 2012. Removal of trace organic contaminants by the forward osmosis process. Sep. Purif. Technol. 103, 258e266. APHA, 1998. Standard Methods for the Examination of Water and Wastewater, twentieth ed. American Public Health Association, Washington, DC. Cath, T.Y., Childress, A.E., Elimelech, M., 2006. Forward osmosis: principles, applications, and recent developments. J. Membrane Sci. 281 (1), 70e87. Cath, T.Y., Drewes, J.E., Lundin, C.D., 2009. A Novel Hybrid Forward Osmosis Process for Drinking Water Augmentation Using Impaired Water and Saline Water Sources. final report (Project #4150). Water Research Foundation, Denver, CO. Chen, W., Westerhoff, P., Leenheer, J.A., Booksh, K., 2003. Fluorescence excitation-emission matrix regional integration to quantify spectra for dissolved organic matter. Environ. Sci. Technol. 37 (24), 5701e5710. Cornelissen, E., Harmsen, D., De Korte, K., Ruiken, C., Qin, J.-J., Oo, H., Wessels, L., 2008. Membrane fouling and process performance of forward osmosis membranes on activated sludge. J. Membrane Sci. 319 (1), 158e168. Dubois, M., Gilles, K.A., Hamilton, J.K., Rebers, P.t., Smith, F., 1956. Colorimetric method for determination of sugars and related substances. Anal. Chem. 28 (3), 350e356. Frolund, B., Griebe, T., Nielsen, P.H., 1995. Enzymatic activity in the activated-sludge floc matrix. Appl. Microbiol. Biotechnol. 43 (4), 755e761. Heo, J., Boateng, L.K., Flora, J.R., Lee, H., Her, N., Park, Y.-G., Yoon, Y., 2013. Comparison of flux behavior and synthetic organic compound removal by forward osmosis and reverse osmosis membranes. J. Membrane Sci. 443, 69e82. Hoek, E.M., Elimelech, M., 2003. Cake-enhanced concentration polarization: a new fouling mechanism for salt-rejecting membranes. Environ. Sci. Technol. 37 (24), 5581e5588. Huang, Z., Ong, S.L., Ng, H.Y., 2011. Submerged anaerobic membrane bioreactor for low-strength wastewater treatment: effect of HRT and SRT on treatment performance and membrane fouling. Water Res. 45 (2), 705e713. Ismail, S., de La Parra, C., Temmink, H., Van Lier, J., 2010. Extracellular polymeric substances (EPS) in upflow anaerobic sludge blanket (UASB) reactors operated under high salinity conditions. Water Res. 44 (6), 1909e1917. Jeison, D., Del Rio, A., Van Lier, J., 2008. Impact of high saline wastewaters on anaerobic granular sludge functionalities. Water Sci. Technol. 57 (6), 815e819. Jin, X., She, Q., Ang, X., Tang, C.Y., 2012. Removal of boron and arsenic by forward osmosis membrane: Influence of membrane orientation and organic fouling. J. Membrane Sci. 389, 182e187. Johir, M.A.H., Vigneswaran, S., Kandasamy, J., BenAim, R., Grasmick, A., 2013. Effect of salt concentration on membrane

123

bioreactor (MBR) performances: detailed organic characterization. Desalination 322 (0), 13e20. Lay, W.C., Liu, Y., Fane, A.G., 2010. Impacts of salinity on the performance of high retention membrane bioreactors for water reclamation: a review. Water Res. 44 (1), 21e40. Lay, W.C., Zhang, Q., Zhang, J., McDougald, D., Tang, C., Wang, R., Liu, Y., Fane, A.G., 2011. Study of integration of forward osmosis and biological process: membrane performance under elevated salt environment. Desalination 283, 123e130. Li, Y., Li, A.-M., Xu, J., Li, W.-W., Yu, H.-Q., 2013. Formation of soluble microbial products (SMP) by activated sludge at various salinities. Biodegradation 24 (1), 69e78. Martinez-Sosa, D., Helmreich, B., Netter, T., Paris, S., Bischof, F., Horn, H., 2011. Anaerobic submerged membrane bioreactor (AnSMBR) for municipal wastewater treatment under mesophilic and psychrophilic temperature conditions. Bioresour. Technol. 102 (22), 10377e10385. Mi, B., Elimelech, M., 2010. Gypsum scaling and cleaning in forward osmosis: measurements and mechanisms. Environ. Sci. Technol. 44 (6), 2022e2028. Morgan, J., Forster, C., Evison, L., 1990. A comparative study of the nature of biopolymers extracted from anaerobic and activated sludges. Water Res. 24 (6), 743e750. Reid, E., Liu, X., Judd, S., 2006. Effect of high salinity on activated sludge characteristics and membrane permeability in an immersed membrane bioreactor. J. Memb. Sci. 283 (1), 164e171. Schiener, P., Nachaiyasit, S., Stuckey, D., 1998. Production of soluble microbial products (SMP) in an anaerobic baffled reactor: composition, biodegradability, and the effect of process parameters. Environ. Technol. 19 (4), 391e399. Singh, K.S., Viraraghavan, T., 2003. Impact of temperature on performance, microbiological, and hydrodynamic aspects of UASB reactors treating municipal wastewater. Water Sci. Technol. 48 (6), 211e217. Singh, K.S., Viraraghavan, T., 1998. Start-up and operation of UASB reactors at 20 C for municipal wastewater treatment. J. Ferment. Bioeng. 85 (6), 609e614. Smith, A.L., Stadler, L.B., Love, N.G., Skerlos, S.J., Raskin, L., 2012. Perspectives on anaerobic membrane bioreactor treatment of domestic wastewater: a critical review. Biores. Technol. 122, 149e159. Stuckey, D.C., 2012. Recent developments in anaerobic membrane reactors. Biores. Technol. 122, 137e148. Sutton, P.M., Be´rube´, P., Hall, E.R., 2004. Membrane Bioreactors for Anaerobic Treatment of Wastewaters Task 1. In: Phase 1 Report: Compilation/Review of Existing Literature. Water Environment Research Foundation. Trzcinski, A., Stuckey, D., 2009. Continuous treatment of the organic fraction of municipal solid waste in an anaerobic twostage membrane process with liquid recycle. Water Res. 43 (9), 2449e2462. Vyrides, I., Stuckey, D.C., 2009. Effect of fluctuations in salinity on anaerobic biomass and production of soluble microbial products (SMPs). Biodegradation 20 (2), 165e175. Wang, Z.-P., Zhang, T., 2010. Characterization of soluble microbial products (SMP) under stressful conditions. Water Res. 44 (18), 5499e5509. Yap, W.J., Zhang, J., Lay, W.C., Cao, B., Fane, A.G., Liu, Y., 2012. State of the art of osmotic membrane bioreactors for water reclamation. Biores. Technol. 122, 217e222. Zhang, J., Loong, W.L.C., Chou, S., Tang, C., Wang, R., Fane, A.G., 2012. Membrane biofouling and scaling in forward osmosis membrane bioreactor. J. Membrane Sci. 403e404, 8e14.