Fouling control of submerged hollow fibre membrane bioreactor with transverse vibration

Journal of Membrane Science 505 (2016) 216–224

Contents lists available at ScienceDirect

Journal of Membrane Science journal homepage: www.elsevier.com/locate/memsci

Fouling control of submerged hollow fibre membrane bioreactor with transverse vibration Tian Li a,b,n, Adrian Wing-Keung Law a,b, Yishuai Jiang a, Agnes Kurniawati Harijanto a, Anthony G. Fane b,c a DHI-NTU Centre, Nanyang Environment and Water Research Institute, Nanyang Technological University, 1 Cleantech Loop, CleanTech One, 637141, Singapore b School of Civil and Environmental Engineering, Nanyang Technological University, 50 Nanyang Avenue, 639798, Singapore c Singapore Membrane Technology Centre, Nanyang Environment and Water Research Institute, Nanyang Technological University, 1 Cleantech Loop, CleanTech One, 637141, Singapore

art ic l e i nf o

a b s t r a c t

Article history: Received 12 November 2015 Received in revised form 30 December 2015 Accepted 2 January 2016 Available online 12 January 2016

The present study further examines the effectiveness of transverse vibration for the fouling control of hollow fibre membranes in a submerged membrane bioreactor (SMBR) with real mixed liquor. The scope of the investigation included both short and long duration tests. The short duration tests incorporated both continuous and intermittent vibration in different concentrations of mixed liquor, and compared among longitudinal vibration, transverse vibration and combined transverse vibration and low aeration. The results indicated that continuous transverse vibration with low frequencies was very effective towards the fouling control, and was more so than longitudinal vibration. Intermittent vibration was somewhat less favourable, but it provided significant energy savings. With intermittent vibration, a short non-vibration time interval of less than 120 s was also found to be necessary to prevent irreversible fouling during this time interval when the permeated extraction continued. For the first time we found that transverse vibration can be more effective for fouling control when combined with low aeration due to the turbulence enhancement by the moving bubbles. The long duration tests confirmed that transverse vibration could be effectively used in the SMBR to control the membrane fouling and prolong the membrane filtration period without chemical cleaning, with high removal efficiencies of both organics and nutrients in the reactor. Higher frequencies of transverse vibration produced larger shear stresses on the membrane surface and enhanced the membrane filtration performance, at the same time the risk of fibre breakage also heightened with the increase in periodic stress cycles on the hollow fibres. The characteristics of the biosolids in the reactor were monitored during the long duration tests. The results showed that both soluble microbial products (SMP) and extracellular polymeric substances (EPS) affected the membrane filtration performance under transverse vibration, but SMP was found to be key membrane fouling contributor. Overall, the study demonstrated that transverse membrane vibration is practical and beneficial for fouling control in the SMBR. & 2016 Elsevier B.V. All rights reserved.

Keywords: Submerged membrane bioreactor Transverse vibration Aeration Fouling control Mixed liquor

1. Introduction A submerged membrane bioreactor (SMBR) combines the conventional bioreactor and a submerged membrane filtration unit together as a treatment facility. It represents an advanced technology for wastewater treatment that can save operating space and also reduce the energy consumption of the process. The bioreactor can completely retain the bacteria and biomass [2,3] n Corresponding author at: DHI-NTU Centre, Nanyang Environment and Water Research Institute, Nanyang Technological University, 1 Cleantech Loop, CleanTech One, 637141, Singapore. E-mail addresses: [email protected] (T. Li), [email protected] (A.-K. Law).

http://dx.doi.org/10.1016/j.memsci.2016.01.003 0376-7388/& 2016 Elsevier B.V. All rights reserved.

and, together with the membrane module separation, also produce a superior quality effluent. The submerged membrane filtration unit typically utilizes either ultra- or micro-filtration (UF/ MF) membranes with a pore size of less than 0.2 μm. With the above advantages, the SMBR has been widely adopted and studied over the past two decades. However, the main challenge in SMBR applications continues to be membrane fouling which can cause a significant reduction in membrane permeability as well as more frequent membrane cleaning and replacement, thus increasing the operational costs. Several approaches have been adopted to reduce membrane fouling in SMBRs, including: (i) pretreatment of the feed water; (ii) improvement of the membrane properties; and (iii) optimization of the hydrodynamic conditions by air sparging [4,5].

T. Li et al. / Journal of Membrane Science 505 (2016) 216–224

Recently, the use of membrane vibration in SMBRs has been proposed as an alternative for fouling control of the membranes. Genkin et al. investigated the vibration of longitudinal membranes with transverse vanes and coagulant addition to improve the filtration performance of submerged hollow fibre membranes [6]. They observed that by using a longitudinal membrane vibration frequency of 1.7 Hz, the critical flux increased from 17 to 46 L/(m2 h) with the coagulant addition of 34 mg/L aluminium chlorhydrate. With the additional transverse vibration of the vanes next to the membranes, a five-fold enhancement in critical flux to 86 L/(m2 h) was also achieved. They attributed the effect of coagulation to the aggregation of fine particles and evacuation of aggregates away from the membrane surface due to inertial and gravitational forces, and the effect of vanes due to the intensified turbulence. Kola et al. found that the transverse vibration of submerged hollow fibre membranes induced less membrane fouling in treating secondary effluent from an anaerobic bioreactor (AnMBR) in comparison to gas sparging and crossflow [7]. They attributed the effectiveness to the fact that the vibrations produced enhanced shear stresses near the membrane surface that reduced the foulant deposition from the bulk feed. Mezohegyi et al. compared the usage of aerated and vibrated membrane filtration systems for the concentration of sewage from a municipal wastewater plant [8]. They concluded that the vibrated membrane filtration had a clear advantage over the aerated system in terms of both fouling control and energy consumption. In contrast, De Vrieze et al. found that membrane vibration in an AnMBR led to a sharp increase in trans-membrane pressure (TMP) and cake layer formation, and thus concluded that vibration alone without mixing was not as effective as the biogas recirculation in their anaerobic system [9]. However, our previous study confirmed that the use of transverse vibration with model feeds of Bentonite and yeast suspensions could effectively reduce membrane cake fouling and lead to an order of magnitude improvement in the fouling reduction [10]. We found that in addition to the generation of high shear stresses introduced by vibrating fibres, transverse vibration can also induce circulation in the reactor, which could further reduce membrane fouling. It should be noted that all of the above studies were conducted in short duration tests (generally less than one day). There is thus still a need to investigate over longer durations comparable to typical commercial SMBR operations. The present study further examined the fouling control of SMBR by transverse vibration for wastewater treatment. The study included both short and long duration tests in laboratory reactors with real mixed liquor. The short duration tests were intended to examine the effects of intermittent vibration and feed concentrations under the action of vibrating hollow fibre membranes. The test duration was typically less than one day so that more combinations could be examined within a reasonable time frame. A comparison study between longitudinal vibration, transverse vibration, and combined transverse vibration and low aeration in mixed liquor was conducted with the short duration tests. Subsequently, experiments with long operating duration were performed to investigate the effectiveness of the vibration approach with a time scale of weeks which is more typical of commercial SMBR operations. The long duration tests included examining the impact of microbial parameters on the membrane filtration performance under vibrations. In particular, the influence of mixed liquor suspended solids (MLSS), mixed liquor volatile suspended solids (MLVSS), as well as SMP and EPS [11–13] in the reactor was closely monitored under transverse vibrations. In the following, the experimental setup and procedures are first described. The results will then be presented.

217

2. Materials and methods 2.1. Hollow fibre membrane module Polyacrylonitrile (PAN) hollow fibres made by Ultrapure Pte. Ltd. in Singapore were used in the present study. The nominal pore size of the membranes was 0.1 μm. Two different fibre sizes were used: inner/outer diameters of 1/1.7 mm and 1/2 mm. The performance of these two types of hollow fibres under longitudinal vibration has been reported in our earlier study [14]. Two types of membrane modules (Modules I and II) were used in the experiments: Module I for the short duration tests, and Module II for both the short and long duration tests. Fig. 1 shows the pictures of the two modules. Module I had a 5  5 membrane fibre bundle with a length of 18 cm and a fibre spacing of 5 mm. Module II can hold up to 200 fibres (arranged in 5 rows with 40 fibres each). The row spacing was 20 mm, and the fibre spacing within a row was 6.5 mm. The 1/1.7 mm and 1/2 mm fibres were utilized in Modules I and II, respectively. For both modules, the fibres were mounted with 1% looseness as we had found previously that a small fibre looseness could significantly improve the membrane filtration performance during vibrations [10,14]. 2.2. Feed In this study, the feeds included both real mixed liquor and synthetic wastewater. The real mixed liquor was collected from the Ulu Pandan Water Reclamation Plant (UPWRP) in Singapore. Two types of mixed liquor were obtained. The first type was collected from the aeration tank of the UPWRP, with MLSS concentrations of 4 71 g/L, and average particle (floc) size of 52.5 μm. The second type was from the membrane tank of the UPWRP, with MLSS concentrations of 8 71 g/L, and average particle size of 67.6 μm. Both represented typical mixed liquor produced in the water reclamation plants across Singapore. For the synthetic

Fig. 1. Membrane modules.

218

T. Li et al. / Journal of Membrane Science 505 (2016) 216–224

Pressure transducer

Permeate pump

Valve

P

Motor

Hollow fibre membrane module Digital balance Air pressure pump

Computer

Air flow meter

Diffuser

Tank

(a) Vibration setup I Feed pump

Pressure sensor

Motor

Mixer

Water level sensor

Hollow fibre membrane module

Liquid flow meter

Permeate tank

Feed tank Aeration pump

Permeate pump

Reactor tank

Air flow meter

Data logging

(b) Vibration setup II Fig. 2. Schematic diagrams of experimental setups.

wastewater, the feed contained 469 mg/L glucose (C6H12O6) as carbon source, 191 mg/L NH4Cl and 28 mg/L K2HPO4 as nutrients, which resulted in a total COD of 500 mg/L and a COD:N:P ratio of 100:10:1. Other compounds (5 mg/L MgCl2
6H2O, 6.45 mg/L CaCl2
2H2O, 4.5 mg/L FeCl3, 1.91 mg/L CoCl2
6H2O, 1.23 mg/L Na2MoO4
2H2O, 0.39 mg/L CuSO4
5H2O, 0.16 mg/L MnSO4
H2O, 0.44 mg/L ZnSO4
7H2O, 0.5 mg/L H3BO3, and 0.1 mg/L KI) were also included according to previous studies [12,15,16] as being representative of the combination of trace elements in the wastewater. 2.3. Experimental Setup Our study utilized two experimental setups (Setups I and II) shown in Fig. 2. Both setups included the reactors, the vibration mechanism and the permeate measurement equipment. Setup I used Module I, the details of which have been described in our previous study [10] and thus are not repeated here. This setup allowed both transverse and longitudinal vibrations, and it was

primarily used for the short duration tests. The long duration tests were conducted in Setup II with Module II using transverse vibrations. Setup II consisted of two Persplex vessels, the feed tank and the reactor tank, and the effluent was connected to the drainage. The feed tank had dimensions of 600 mm (L)  600 mm (W)  800 mm (H), and about 150 L influent was fed into the feed tank every day. The reactor tank was 600 mm (L)  300 mm (W)  800 mm (H), and half filled with 80 L mixed liquor. A mixer with impeller was installed in the feed tank, to keep the feed homogeneously suspended at all times. A feed pump was activated automatically to increase the volume in the reactor tank when the water level inside dropped below the target value, so that a constant volume of mixed liquor suspension could be maintained. A DC motor was located on top of the reactor tank to oscillate the submerged membrane module in a sinusoidal manner with a crank moving mechanism. The hollow fibre membranes were horizontally mounted in the membrane module, so that a transverse vibration of the membrane was induced with the oscillatory movement. The length of the driving rod connected

T. Li et al. / Journal of Membrane Science 505 (2016) 216–224

to the motor determined the submergence of the hollow fibres. The reactor tank was equipped with 5 pairs of air stones (each air stone with a distance of 6.5 cm in a pair, and each adjacent pair with a distance of 13 cm) placed at the bottom of the reactor tank to supply the air bubbles. Each pair was connected to an air flow metre (Dwyer) to monitor the air flow rate into the reactor. The permeate from the membrane module was pumped out by a digital permeate peristaltic pump (Cole-Parmer Instrument Company), and the permeate suction pressure was monitored by a digital pressure sensor. In the present setup, a data logging system with SCADA automation software and PLC control system was used to synchronize the motor and the permeate peristaltic pump either continuously or intermittently. The system also controlled the permeate peristaltic pump and liquid flow metre to maintain a constant permeate flux, while monitoring and recording the TMP values at the same time. 2.4. Experimental procedures Before each experiment, a clean water backwash at 20 and 200 mL/min was performed on Modules I and II, respectively, for 20 min. The filtration test was then initiated after the backwash. When the individual test in mixed liquor was completed, the hollow fibre membrane module was taken out from the reactor and flushed with tap water, then soaked in 0.2% sodium hypochlorite and 0.2% citric acid each for 2 h followed by repeating the clean water backwash procedures. The permeability of the membranes was measured after the chemical cleaning. If the permeability recovered to above 95% of the original value, the membranes were reused, otherwise they were replaced with new ones. 2.5. Analytical methods For the long duration tests, the concentrations of MLSS and MLVSS were quantified according to the Standard Methods (AHPA, 1995). COD was determined using the HACH USEPA reactor digestion method (HACH 2125915/2415815). The nitrogen concentration, represented by NO3–N, was measured by the Ion Chromatography (IC) (Dionex). As reported earlier, SMP and EPS have been found to exert significant influence on the membrane filtration performance [11,13]. Carbohydrates and proteins are the main SMP and EPS components, and their concentrations in the long duration tests were monitored to quantify the biological effects under membrane vibration. Samples of the harvested mixed liquor from the reactor were centrifuged at 4000 rpm for 10 min in order to separate the supernatant and sludge. The supernatant, after filtration by 0.45 μm cellulose acetate membranes (CA, Fisher Scientific), was considered as SMP. The sludge was then re-suspended by the same volume amount of Milli-Q water and placed in the water bath at 80 °C for 30 min after which the sample was centrifuged for 10 min at 4000 rpm. The supernatant was filtered through a 0.45 μm CA membrane, and the permeate was considered as EPS [17]. Both SMP and EPS were analyzed for carbohydrate and protein fractions. The carbohydrates and proteins were determined by the phenol-sulphuric acid method [18] using glucose as standard, and by the modified Lowry method [19] using Bovine Serum Albumin (BSA) as standard, respectively.

219

duration in mixed liquor were performed to examine the membrane performance under mechanical vibration. Module I in Setup I was operated at a constant permeate flux of 25 71 L/(m2 h), and vibrated transversely at an amplitude of 16 mm and two frequencies of 1 and 2 Hz in the reactor with the 8 71 g/L mixed liquor from the UPWRP. The results were compared with no vibration (Fig. 3). It can be observed that there was a huge jump in TMP without vibration, indicating immediate severe fouling of the membrane. With 1 Hz continuous vibration, the TMP remained constant for the first 4 h filtration, and then increased slowly from 15 to 22 kPa after 9 h. With 2 Hz continuous vibration, there was only 1 kPa increase of TMP for the 9 h filtration. This indicated that the transverse vibration was very effective for membrane fouling control, which can be attributed to the direct shear enhancement on the membrane surface. Compared with the membrane filtration performance using the same membrane module in 4 g/L Bentonite or yeast suspensions at the same vibration frequency that was presented in Li et al. [10], membrane fouling in the higher concentration of 8 71 g/L real mixed liquor was less severe, which might be attributed to the larger average particle size of the mixed liquor in the present experiments that was over 50 mm (note that with short duration filtration, biofouling from mixed liquor due to biofilm formation, adhesion and attachment was not significant). With such a large particle size, the lateral migration of mixed liquor particles due to inertial lift or shear-induced diffusion would be dominant [20,21]. Since the inertial lift velocity increases with the cube of the particle size [21,22], the large mixed liquor floc would experience much higher inertial lift forces away from the membrane under transverse vibration, and thus the fouling was reduced. 3.1.2. Effect of intermittent vibration In our earlier work, intermittent vibration with sequential foulant deposition and removal on the membrane was investigated in inorganic Bentonite and organic yeast suspensions to reduce the energy consumption and operating cost for the membrane vibration approach [10]. In this study, the intermittent vibration effect was also investigated in mixed liquor using Setup I. Module I was operated at a constant permeate flux of 25 71 L/(m2 h) and vibrated transversely at an amplitude of 16 mm and frequency of 1 Hz with the two types of mixed liquor mentioned in Section 2.2. Vibration and non-vibration were carried out in an alternate manner with a range of on/off sequences. The results of intermittent vibration are compared with continuous vibration and non-vibration in Fig. 4 (the on time means vibration, while the

3. Results and discussion 3.1. Short duration tests 3.1.1. Effect of vibration frequency Experiments with hollow fibre membrane filtration for a short

Fig. 3. TMP history of Module I in short duration tests with different vibration frequencies.

220

T. Li et al. / Journal of Membrane Science 505 (2016) 216–224

50.4 kPa after 510 min, confirming the undesirable outcome of a lengthy period of membrane filtration without vibrations inside the mixed liquor. The above findings demonstrated that the shear enhancement generated by transverse vibration with a short non-vibration interval can mitigate the membrane fouling in a near continuous manner. In other words, intermittent transverse vibration could be very effective in real applications. For both the 4 71 and 87 1 g/L mixed liquor, the system could be maintained at constant flux operation with a relatively low frequency of 1 Hz and an intermittency with short non-vibration time (60/60 s).

Fig. 4. TMP history of Module I with different intermittent vibration intervals in different concentrations of mixed liquor: (a) 4 71 g/L, and (b) 8 7 1 g/L.

off time means non-vibration). In the 4 71 g/L mixed liquor (Fig. 4 (a)), the TMP increased significantly to 70 kPa within a short filtration time of 125 min without vibration. With the 30/60 s interval of intermittent vibration, the TMP increase became much slower, reaching 53.7 kPa after 510 min. Furthermore, with the 60/ 60 s interval of intermittent vibration, the TMP only increased to 24.4 kPa in the same duration, which indicated that the longer vibration time was more effective for fouling control. However, with the 300/300 s interval of intermittent vibration, there was a larger TMP increase to 45.8 kPa. This implied that the 300 s of nonvibration caused some irreversible fouling of the membrane. Similar observations were also noted in the experiments with the 8 71 g/L mixed liquor (Fig. 4(b)). The TMP jumped significantly to above 70 kPa in 115 min without vibration, which was slightly shorter than the 4 71 g/L mixed liquor despite the doubling in concentrations. With the 60/60 s interval of intermittent vibration, the TMP increased to 26.4 kPa after 510 min filtration. It further increased to only a slightly larger value of 27.7 kPa with the 60/120 s interval of intermittent vibration for the same test duration, which indicated that the longer non-vibration time was not detrimental while significant energy saving could be realized. However, the TMP increased to 39.6 kPa after 510 min with the 120/240 s interval of intermittent vibration. Thus, the longer nonvibration interval exceeding 120 s was not desirable despite the same total combined intermittent vibration time. With 300/300 s interval of intermittent vibration, the TMP increased further to

3.1.3. Comparison between transverse and longitudinal vibrations In this part of the study, the membrane filtration performances with both transverse and longitudinal vibrations were compared using Module I in Setup I with the 8 71 g/L mixed liquor. Both vibrations were conducted in intermittent mode at an amplitude of 16 mm and a frequency of 1 Hz. Two intermittent vibration intervals were operated: 60/120 s and 120/240 s. The TMP profiles are shown in Fig. 5. With the 60/120 s interval of intermittent vibration, the TMP of membrane filtration increased to 58.6 kPa with longitudinal vibration during the 9 h filtration, while the increase was much smaller at 29.3 kPa with transverse vibration. A similar trend was also observed with the 120/240 s interval of intermittent vibration, but the TMP performance was worse than the former despite the same amount of total combined vibration and non-vibration time, which again confirmed that a long non-vibration time exceeding 120 s was not desirable. The better membrane filtration performance by transverse vibration can be attributed to the separating boundary layers induced by the transverse movement of the hollow fibres, being more effective in reducing membrane fouling [10]. In addition, with the periodic reciprocal movement by the transverse vibration, a streaming secondary flow pattern is produced (i.e. circulation around the membrane) that can further increase the shear rate in the neighbourhood of the membrane surface. However, the risk of fibre breakage due to fatigue with periodic stresses also increases with the transverse vibration at higher vibration frequencies or amplitudes due to the more severe lateral movement compared with longitudinal vibration. The potential damage of the hollow fibres induced by the fibre movement was also noted in earlier studies [10,23,24].

Fig. 5. TMP history of Module I with transverse and longitudinal vibration in short duration tests.

T. Li et al. / Journal of Membrane Science 505 (2016) 216–224

221

and low aeration was very effective for fouling control in the SMBR. Possible reasons for the effectiveness of the combination are: (i) turbulence in the reactor is enhanced by the rising air bubbles, and (ii) the bubble-induced liquid cross-flows flush away the particles displaced by the vibrations from the membrane surface. 3.2. Long duration tests

Fig. 6. TMP history of Module II with combined vibration and low aeration in short duration tests.

3.1.4. Comparison between vibration and air sparging process Experiments were performed with short duration tests to determine the synergistic effects of combined vibration and aeration using the larger membrane Module II in the Setup II. A larger membrane module with a larger membrane surface area can increase the permeate productivity without affecting the permeate flux. Here, Module II in Setup II was tested with different combinations of transverse vibration frequencies and air flow rates. The module was submerged in the 4 71 g/L mixed liquor and operated at a vibration amplitude of 30 mm, with different vibration frequencies of 0.5, 0.67, and 0.83 Hz, as well as two air flow rates of 1 and 3 L/min. The relatively low vibration frequencies were adopted due to the smaller energy consumption and lower probability for fibre breakage. The short duration tests were operated at a constant permeate flux of 2571 L/(m2 h) for typically around 20 h (Fig. 6). With a low vibration frequency of 0.5 Hz and no aeration, the TMP remained relatively constant for the initial 2.5 h, followed by a slow increase to 27 kPa in the next 3.5 h, and then jumped quickly to nearly 100 kPa after about 8 h. With higher vibration frequencies of 0.67 and 0.83 Hz, the TMP maintained constant without severe membrane fouling for over 20 h. In other words, beyond a critical threshold in the vibration frequency, continuous transverse vibration can be very effective for membrane fouling control in mixed liquor. Compared to the smaller diameter membrane Module I (Fig. 4), Module II not only increased the productivity, but also induced less severe membrane fouling under transverse vibration in the mixed liquor. This may be attributed to the slightly larger fibre spacing in Module II that reduced the flux competition (although our earlier study [10] has shown that the effect should be small with the relatively large spacing adopted in the two modules), the larger fibre diameter which induced smaller lateral flux and potentially lesser non-uniformity in the flux distribution [25], as well as having a larger Reynolds number under transverse vibration which can enhance the secondary flows and reduce the membrane fouling [26]. These observations were consistent with our previous findings [14] that the larger diameter hollow fibre membranes performed better under longitudinal vibration in Bentonite suspensions by directly comparing the two types of 1/1.7 mm and 1/2 mm hollow fibre membranes with the same test configuration. With an air flow rate of 3 L/min and no vibration, the TMP increased by 14.1 kPa in just 2 h filtration. However, with a combination of 0.5 Hz transverse vibration and 1 L/min air sparging, almost no membrane fouling was observed for 20 h filtration. This showed that the combination of vibration

Setup II with Module II was used primarily for the long duration tests, during which the effects due to vibration frequencies and biological characteristics on the membrane filtration performance were examined. The reactor was operated at the room temperature of 25 72 °C. The pH was between 6.7 and 7.1. The membrane flux was 257 1 L/(m2 h) and a total permeate flow rate of 210 mL/min was collected from the total 200 hollow fibre membranes during the filtration. However, the permeate extraction was operated at 1 h intervals for alternative filtration/relaxation. The hollow fibre membranes were vibrated transversely at an amplitude of 30 mm, with two frequencies of 0.67 and 1 Hz in the mixture of mixed liquor and synthetic wastewater. In order to reduce the energy consumption by the vibration, the motor was also switched off during the relaxation stage of the membrane filtration. A total air flow rate of 10 L/min or an equivalent of 1 mm/s superfacial air velocity was used for the biomass oxygenation and suspension, which was less than most other studies reported in the literature [27,28] due to the fact that the transverse vibration of the membrane module can also aid in the mixing of the biosolids inside the reactor as well as the fouling control. The TMP from the membrane filtration was monitored over time. When the TMP exceeded 85 kPa, which was indicative of excessive fouling, the experiment was stopped and the membrane module was taken out for chemical cleaning. Each day, a small amount of sample was taken out from the reactor for water quality analysis. A total of 5 tests were performed for the long duration tests. The first three tests were run at the vibration frequency of 0.67 Hz, while the following two tests were at the higher frequency of 1 Hz. The feed contained different ratios of mixed liquor and synthetic wastewater. Table 1 shows the basic operational information as well as the average SMBR performance for the five tests. The TMP histories of the membrane filtration in the five tests are illustrated in Fig. 7 (note that the TMP values during the permeate relaxation are not presented). In general, the TMP values showed an initial slow rise followed by a rapid TMP jump. At the end of each test, when TMP rose significantly above 70 kPa due to severe membrane fouling, high concentrations of SMP and EPS were typically observed in the bulk phase of the reactor, which accumulated on the membrane surface and increased the filtration resistance significantly. In the first three tests, the TMP history developed differently despite the fact that they had the same transverse vibration frequency of 0.67 Hz. In Test 1, with the relatively high concentration of MLSS in the reactor, the TMP gradually increased to 30 kPa after the first week, then jumped quickly to above 70 kPa in the next three days. It was noted that very high average concentrations of SMP and EPS (66.4 and 797 mg/L, respectively) were recorded in the reactor during the 10 days filtration, which likely led to the severe membrane fouling. The membrane filtration period in Test 2 was nearly the same as Test 1 due to similar MLSS concentrations in the reactor. In the first 10 days filtration, the TMP went up relatively slowly to 20 kPa as compared to Test 1 due to smaller concentrations of SMP and EPS. In Day 11, the TMP jumped suddenly to over 70 kPa together with high MLVSS concentrations in the reactor. The high concentration of MLVSS induced an increased viscosity of the biosolids suspension and reduced the membrane filterability [29].

222

T. Li et al. / Journal of Membrane Science 505 (2016) 216–224

Table 1 Vibration SMBR characteristics in the five tests for the long duration tests. Parameters

Test 1

Test 2

Test 3

Test 4

Test 5

Ratios of mixed liquor/synthetic wastewater *MLSS concentration (g/L) *MLVSS concentration (g/L) *Carbohydrate concentration in SMP (mg/L) *Protein concentration in SMP (mg/L) *Carbohydrate concentration in EPS (mg/L) *Protein concentration in EPS (mg/L) Average COD removal (%) Average NO3–N removal (%)

6:5 17.4 7 4.7 9.9 7 6.0 26.7 7 21.0 39.7 7 13.1 1517 73 646 7141 92.0 99.4

8:3 17.2 7 6.4 13.0 7 4.8 12.3 78.2 26.7 7 13.1 178 763 589 7128 93.5 98.5

5:3 18.17 4.1 13.8 7 3.0 13.2 7 5.3 29.4 7 11.0 2017 34 5767 119 87.4 99.8

1:6 5.8 7 6.4 4.8 7 4.6 18.6 7 9.2 30.4 7 11.6 1077 78 582 7131 96.5 91.4

16:27 12.5 74.4 10.0 7 3.5 11.6 7 17.2 20.2 7 6.8 1867 70 593 7118 93.9 96.1

*Mean value7 standard deviation.

Fig. 7. TMP history of Module II with different vibration frequencies in long duration tests.

Despite the same vibration frequency of 0.67 Hz in Test 3, the duration of membrane filtration was, however, shorter than the previous two tests, which was probably due to the higher MLSS concentrations. An average MLSS concentration of 18.1 g/L was recorded in the reactor during the 7-day filtration before the TMP went up to above 70 kPa. More importantly, the high MLSS concentrations led to breakage of the transversely vibrating hollow fibre membranes. Clearly fibre breakage due to fatigue of the membrane material upon repetitive cyclic stresses constitutes a serious concern to the membrane vibration approach for fouling mitigation. The breakage typically occurred at the potting of the fibres. In our previous work, we reported higher probability of fibre breakage with transverse vibration than longitudinal vibration [10]. Higher MLSS concentrations usually lead to an increase in viscosity [29] and the larger fluid-induced drag forces accentuate the stresses at the potting of the fibres, which can accelerate fibre breakage. It should be noted that high MLSS concentrations in the reactor are also highly undesirable in general in terms of membrane fouling. For example, Shane Trussell et al. reported a rapid loss of membrane permeability at high solid concentrations above approximately 20 g/L [30]. Itonaga et al. also suggested that high MLSS concentrations were the cause of severe membrane fouling due to the significant increase of the suspension viscosity [31]. The other constraint on high MLSS is the drop in oxygen transfer efficiency [32] that typically limits MLSS to o15 g/L. In the current study, the experiment was stopped when significant membrane fouling occurred in such high solid concentrations. In Test 4, a higher transverse vibration frequency of 1 Hz was utilized as compared to Tests 1–3. Unfortunately, fibre breakage was also accelerated, and it occurred shortly after the first week of filtration despite the fact that only limited membrane fouling was

observed in terms of TMP. With higher transverse vibration frequencies, the number of cyclic periods of repetitive stresses also increased with time. The fibre breakage was then repaired and the test continued. In the following 20 days, the TMP increased slowly from 18 to 60 kPa. The long duration test was deemed satisfactory for a typical SMBR operation. Thus, the results in Test 4 confirmed that transverse vibration can be regarded as an effective alternative for fouling control in SMBRs with real mixed liquor, although the risk for fibre breakage needs to be minimised. Compared with Test 2, higher concentrations of soluble carbohydrate/protein and lower concentrations of bound carbohydrate/protein were recorded in Test 4. Liu and Tay studied the role of hydrodynamic shear force in the formation of biofilms and granular sludge [33]. They pointed out that the high shear force could induce both aerobic granules and biofilms to secrete more carbohydrates resulting in a balanced microbial structure and stability of aerobic biofilms and granules. Shane Trussell et al. studied the influence of mixed liquor properties and aeration intensity on membrane fouling in an SMBR with high suspended solid concentrations [30]. They found that the highest soluble protein concentration was obtained with the highest aeration intensity. They attributed this to the fact that the additional shear force provided by the increased aeration intensity led to protein release from the biological flocs to the soluble environment. In an analogous manner, the higher shear stresses with the higher frequency of transverse vibration could also induce the release of carbohydrate/protein from the biological flocs to the soluble system. Another possible factor was that relatively high concentrations of SMP per unit MLSS concentrations were recorded in Test 4, which was attributed to the higher proportion of synthetic wastewater. This was consistent with the results from Xie et al., who demonstrated that an increase in substrate concentration would lead to an increase in SMP production [34]. In Test 5, the membrane filtration lasted for more than 40 days without chemical cleaning, further confirming that the fouling control by transverse vibration was effective. Compared with Test 4 with the same vibration frequency of 1 Hz, the membrane filtration was much longer despite the much higher MLSS concentrations. The longer duration can be likely attributed to the lower SMP of  30 g/L as well as higher EPS of  800 g/L, which might aggregate the biosolids into larger flocs and reduce the membrane fouling [29,35]. Such relationships between membrane fouling and SMP/EPS concentrations have also been reported in many previous studies. Rosenberger and Kraume found that the carbohydrate concentration in SMP correlated well with the membrane fouling rate for eight different SMBR sludges in a batch cross-flow membrane cell [36]. Rosenberger et al. also pointed out that soluble carbohydrate was the primary cause of increased membrane fouling rate [37]. Mukai et al. [38] and Hernandez Rojas et al. [39] suggested that the soluble protein was the key factor

T. Li et al. / Journal of Membrane Science 505 (2016) 216–224

affecting the membrane filtration performance. Ji and Zhou studied an SMBR with air bubbles [40]. They found that proteins were more prone to be adsorbed on the membrane surface than carbohydrates, which increased the membrane hydrophobicity and caused irreversible membrane fouling. Shane Trussell et al. reported that the total SMP concentration determined the membrane fouling rate due to the exposure to the increased soluble organic content, no matter whether it was carbohydrate or protein in the SMP [41]. Our results were consistent with Shane Trussell et al. [41] showing that SMP was the dominant factor to contribute to membrane fouling. In all five tests, high removal efficiencies of both the organics and nutrients were generally recorded. The COD removal efficiency in Test 3 was relatively lower, which might be due to a relatively low nutrient concentration from the feed and an exceedingly high MLSS concentration in the reactor, both leading to decreased oxygen transfer efficiency and slow microorganism digestion rate. The NO3–N removal efficiency in Test 4 was also relatively low, which was attributed to the very high nutrient concentration from the synthetic feed and very low MLSS concentration in the reactor, resulting in the slow growth rate and inefficiency in converting the nutrients into biosolids. Finally, we would like to reiterate that a low air flow rate (10 L/ min) was used in the 80 L reactor in the present study producing a superficial air velocity of 1 mm/s, which was much lower than the general superficial air velocities of 3–6 mm/s in SMBRs [28,42–44]. Correspondingly, lower energy consumption (113.6 W for the 10 L/ min aeration in this study) was incurred for the air bubbling in this case. The additional transverse vibration only required 33.8 and 35.2 W, which was a small increment of energy consumption due to low vibration frequencies of 0.67 and 1 Hz, respectively. Hence, the combined vibration and low aeration approach for fouling control can be much more energy efficient than air sparging alone. The proposed combination of vibration and low aeration has not been reported in the literature to date, and is being proposed here for the first time as far as we are aware. It should be noted that the biomass could not survive without respiration, thus the air bubbling for oxygenation is essential in the aerobic vibratory SMBRs.

4. Conclusions Fouling control of the SMBR by transverse vibration in aerobic bioreactors was investigated experimentally in the present study. Both short and long duration tests of membrane filtration in mixed liquor were performed. From the short duration tests, we found that continuous transverse vibration with low frequency was very effective for fouling control, and was more so than longitudinal vibration. Intermittent vibration was generally less favourable for membrane fouling reduction, but provided a saving in energy consumption. With intermittent vibration, a short non-vibration time interval below 120 s was found to be necessary to prevent irreversible fouling during this time interval when the permeate extraction continued. Fouling control by transverse vibration was also found to be even more effective with a combined low aeration due to the turbulence enhancement by the moving bubbles. The long duration tests confirmed that the use of transverse vibration in the SMBR could effectively control membrane fouling for a lengthy period of weeks (comparable to commercial SMBR operations) and minimize the needs for chemical cleaning, with high removal efficiencies of organics and nutrient typically recorded in the reactor. Higher frequencies of transverse vibration induced larger shear stresses on the membrane surface, therefore enhanced the membrane filtration performance and prolonged the filtration time during the long duration tests. However, the risk of membrane fibre breakage also increased at the same time, due to

223

the increase in cyclic stress on the fibres during vibration. Analysis of the biosolids in the reactor during the long duration tests indicated that both SMP and EPS affected the membrane filtration performance under transverse vibration, but SMP was found to be the main membrane fouling contributor. The overall results of the present study showed that the use of transverse membrane vibration in the SMBR is a practical and beneficial technology to be further exploited in the future.

Acknowledgements The authors would like to gratefully acknowledge National Research Foundation Singapore (NRF) for the financial support under EWI Project No. of MEWR C651/06/175 and Proof-Of-Concept (POC) Project No. of NRF2013NRF-POC002-010.

References [1] AHPA, Standard methods for the examination of water and wastewater, 19th edition. American Public Health Association, Washington, DC. 1995. [2] C. Chiemchaisri, Y.K. Wong, T. Urase, K. Yamamoto, Organic stabilisation and nitrogen removal in a membrane separation bioreactor for domestic wastewater treatment, Filtr. Sep. 30 (1993) 247–252 240. [3] K. Krauth, K.F. Staab, Pressurized bioreactor with membrane filtration for wastewater treatment, Water Res. 27 (1993) 405–411. [4] P. Le-Clech, V. Chen, T.A.G. Fane, Review: fouling in membrane bioreactors used in wastewater treatment, J. Membr. Sci. 284 (2006) 17–53. [5] C.C.V. Chan, P.R. Bérubé, E.R. Hall, Shear profiles inside gas sparged submerged hollow fiber membrane modules, J. Membr. Sci. 297 (2007) 104–120. [6] G. Genkin, T.D. Waite, A.G. Fane, S. Chang, The effect of vibration and coagulant addition on the filtration performance of submerged hollow fibre membranes, J. Membr. Sci. 281 (2006) 726–734. [7] A. Kola, Y. Ye, P. Le-Clech, V. Chen, Transverse vibration as novel membrane fouling mitigation strategy in anaerobic membrane bioreactor applications, J. Membr. Sci. 455 (2014) 320–329. [8] G. Mezohegyi, M.R. Bilad, I.F.J. Vankelecom, Direct sewage up-concentration by submerged aerated and vibrated membranes, Bioresour. Technol. 118 (2012) 1–7. [9] J. De Vrieze, T. Hennebel, J. Van den Brande, R.M. Bilad, T.A. Bruton, I.F. J. Vankelecom, W. Verstraete, N. Boon, Anaerobic digestion of molasses by means of a vibrating and non-vibrating submerged anaerobic membrane bioreactor, Biomass Bioenergy 68 (2014) 95–105. [10] T. Li, A.W.K. Law, A.G. Fane, Submerged hollow fibre membrane filtration with transverse and longitudinal vibrations, J. Membr. Sci. 455 (2014) 83–91. [11] D. Okamura, Y. Mori, T. Hashimoto, K. Hori, Effects of microbial degradation of biofoulants on microfiltration membrane performance in a membrane bioreactor, Environ. Sci. Technol. 44 (2010) 8644–8648. [12] Z. Huang, S.L. Ong, H.Y. Ng, Submerged anaerobic membrane bioreactor for low-strength wastewater treatment: effect of HRT and SRT on treatment performance and membrane fouling, Water Res. 45 (2011) 705–713. [13] T.V. Bugge, P. Larsen, A.M. Saunders, C. Kragelund, L. Wybrandt, K. Keiding, M. L. Christensen, P.H. Nielsen, Filtration properties of activated sludge in municipal MBR wastewater treatment plants are related to microbial community structure, Water Res. 47 (2013) 6719–6730. [14] T. Li, A.W.K. Law, M. Cetin, A.G. Fane, Fouling control of submerged hollow fibre membranes by vibrations, J. Membr. Sci. 427 (2013) 230–239. [15] H.Y. Ng, S.W. Hermanowicz, Membrane bioreactor operation at short solids retention times: performance and biomass characteristics, Water Res. 39 (2005) 981–992. [16] J.R. Pan, Y.C. Su, C. Huang, H.C. Lee, Effect of sludge characteristics on membrane fouling in membrane bioreactors, J. Membr. Sci. 349 (2010) 287–294. [17] J.W. Morgan, C.F. Forster, L. Evison, A comparative study of the nature of biopolymers extracted from anaerobic and activated sludges, Water Res. 24 (1990) 743–750. [18] M. DuBois, K.A. Gilles, J.K. Hamilton, P.A. Rebers, F. Smith, Colorimetric method for determination of sugars and related substances, Anal. Chem. 28 (1956) 350–356. [19] B. Frølund, T. Griebe, P.H. Nielsen, Enzymatic activity in the activated-sludge floc matrix, Appl. Microbiol. Biotechnol. 43 (1995) 755–761. [20] S. Chellam, M.R. Wiesner, Particle transport in clean membrane filters in laminar flow, Environ. Sci. Technol. 26 (1992) 1611–1621. [21] R.H. Davis, Modeling of fouling of crossflow microfiltration membranes, Sep. Purif. Methods 21 (1992) 75–126. [22] G. Belfort, R.H. Davis, A.L. Zydney, The behavior of suspensions and macromolecular solutions in cross-flow microfiltration, J. Membr. Sci. 96 (1994) 1–58. [23] F. Wicaksana, A.G. Fane, V. Chen, Fibre movement induced by bubbling using

224

T. Li et al. / Journal of Membrane Science 505 (2016) 216–224

submerged hollow fibre membranes, J. Membr. Sci. 271 (2006) 186–195. [24] A.P.S. Yeo, A.W.K. Law, A.G. Fane, The relationship between performance of submerged hollow fibers and bubble-induced phenomena examined by particle image velocimetry, J. Membr. Sci. 304 (2007) 125–137. [25] S. Chang, A.G. Fane, The effect of fibre diameter on filtration and flux distribution-relevance to submerged hollow fibre modules, J. Membr. Sci. 184 (2001) 221–231. [26] V. Geraldes, V. Semião, M.N. de Pinho, Flow management in nanofiltration spiral wound modules with ladder-type spacers, J. Membr. Sci. 203 (2002) 87–102. [27] K. Kimura, T. Naruse, Y. Watanabe, Changes in characteristics of soluble microbial products in membrane bioreactors associated with different solid retention times: relation to membrane fouling, Water Res. 43 (2009) 1033–1039. [28] A.M. Al Taweel, A.O. Idhbeaa, A. Ghanem, Effect of electrolytes on interphase mass transfer in microbubble-sparged airlift reactors, Chem. Eng. Sci. 100 (2013) 474–485. [29] L.D. Chabaliná, J.B. Ruiz, M.R. Pastor, D.P. Rico, Influence of EPS and MLSS concentrations on mixed liquor physical parameters of two membrane bioreactors, Desalination Water Treat. 46 (2012) 46–59. [30] R. Shane Trussell, R.P. Merlo, S.W. Hermanowicz, D. Jenkins, Influence of mixed liquor properties and aeration intensity on membrane fouling in a submerged membrane bioreactor at high mixed liquor suspended solids concentrations, Water Res. 41 (2007) 947–958. [31] T. Itonaga, K. Kimura, Y. Watanabe, Influence of suspension viscosity and colloidal particles on permeability of membrane used in membrane bioreactor (MBR), Water Sci. Technol. 50 (2004) 301–309. [32] J. Henkel, M. Lemac, M. Wagner, P. Cornel, Oxygen transfer in membrane bioreactors treating synthetic greywater, Water Res. 43 (2009) 1711–1719. [33] Y. Liu, J.H. Tay, The essential role of hydrodynamic shear force in the formation of biofilm and granular sludge, Water Res. 36 (2002) 1653–1665. [34] W.M. Xie, B.J. Ni, T. Seviour, H.Q. Yu, Evaluating the impact of operational

[35] [36] [37]

[38]

[39]

[40]

[41]

[42]

[43] [44]

parameters on the formation of soluble microbial products (SMP) by activated sludge, Water Res. 47 (2013) 1073–1079. L.H. Mikkelsen, K. Keiding, Physico-chemical characteristics of full scale sewage sludges with implications to dewatering, Water Res. 36 (2002) 2451–2462. S. Rosenberger, M. Kraume, Filterability of activated sludge in membrane bioreactors, Desalination 151 (2003) 195–200. S. Rosenberger, C. Laabs, B. Lesjean, R. Gnirss, G. Amy, M. Jekel, J.C. Schrotter, Impact of colloidal and soluble organic material on membrane performance in membrane bioreactors for municipal wastewater treatment, Water Res. 40 (2006) 710–720. T. Mukai, K. Takimoto, T. Kohno, M. Okada, Ultrafiltration behaviour of extracellular and metabolic products in activated sludge system with UF separation process, Water Res. 34 (2000) 902–908. M.E. Hernandez Rojas, R. Van Kaam, S. Schetrite, C. Albasi, Role and variations of supernatant compounds in submerged membrane bioreactor fouling, Desalination 179 (2005) 95–107. L. Ji, J. Zhou, Influence of aeration on microbial polymers and membrane fouling in submerged membrane bioreactors, J. Membr. Sci. 276 (2006) 168–177. R. Shane Trussell, R.P. Merlo, S.W. Hermanowicz, D. Jenkins, The effect of organic loading on process performance and membrane fouling in a submerged membrane bioreactor treating municipal wastewater, Water Res. 40 (2006) 2675–2683. C.R.G. Tavares, G.L. Santanna, B. Capdeville, The effect of air superficial velocity on biofilm accumulation in a 3-phase fluidized-bed reactor, Water Res. 29 (1995) 2293–2298. S. Brinke-Seiferth, J. Behrendt, I. Sekoulov, The biofilm filter sequencing batch reactor (BFSBR), Water Sci. Technol. 39 (1999) 77–83. S. Villaverde, F. Fdz-Polanco, M.L. Lacalle, P.A. Garcia, Influence of the suspended and attached biomass on the nitrification in a two submerged biofilters in series system, Water Sci. Technol. 41 (2000) 169–176.