Reactor performance and membrane fouling of a novel submerged aerobic granular sludge membrane bioreactor during long-term operation

Membrane bioreactor (MBR) has become more and more popular in the treatment and reuse of industrial and municipal wastewater because of the high effluent quality, small footprint, and low excess sludge production. However, membrane fouling is still an unavoidable barrier in the application of MBRs because of the increase of operational costs (Meng et al. 2017). In MBRs, fouling of membranes is caused by complex physical and chemical interactions between the various fouling constituents in the mixed liquor and the membrane surface (Li et al. 2008a, 2013). Control of membrane fouling through optimization of mixed liquor is one of the hot topics in recent years (Ding et al. 2014; Meng et al. 2017).

The aerobic granular sludge (AGS) process is a technology based on the successful cultivation of aerobic granules in bioreactors. It has excellent settling ability, high biomass retention, ability for simultaneous nitrification-denitrification and removal of toxic substances compared with conventional activated sludge systems (Pronk et al. 2015). The granules are densely packed microbial aggregates with rich and strong microbial structure, which are formed by large varieties of aerobic and anaerobic respiring microbes that colonize in different granule layers. Aerobic sludge granules have been successfully integrated with mesh filters or membranes for development of submerged aerobic granular sludge membrane bioreactors (AGS-MBRs). Aerobic granules played a promising role in membrane fouling reduction in AGS-MBRs as they can form a biocake with more porosity than floc biocake (Tay et al. 2007; Li et al. 2012a, 2012b). The AGS-MBR and flocculent sludge MBR have been compared and the extracellular polymeric substance (EPS) concentration and composition of the foulant was also investigated (Wang et al. 2013). However, most studies have focused on comparison of granular sludge MBR and flocculent sludge MBR in short-term tests using synthetic wastewater. The investigation on stability of the AGS-MBR for municipal sewage treatment is very limited, and the possible granule disintegration and irreversible fouling of the membrane during long-term operation have not been studied.

In this study, two identical AGS-MBRs were developed and operated over 120 days for treating municipal sewage and synthetic wastewater, respectively. The reactor performance and membrane fouling of AGS-MBRs in long-term operation were investigated systemically. Granule size, fractal dimension (D_F), EPS, soluble microbial product (SMP), total organic carbon (TOC), ammonia nitrogen (NH₃-N) and total nitrogen (TN) were monitored continuously. The membrane filtration resistance analysis and excitation-emission matrix (EEM) fluorescence spectroscopy of extracted membrane foulant were also conducted to understand the membrane fouling mechanisms. It can be expected that this study would provide useful information for further development of AGS-MBR technology.

Figure 1 shows a schematic of the laboratory-scale submerged AGS-MBR setup. The reactor is a column-like sequential batch reactor (SBR) with an inner diameter of 20 cm and a height of 140 cm (30 L working volume). The system was operated in a 4 h cycle pattern, with 5 min influent feeding, 215–228 min aeration, 2–15 min settling and 5 min discharge. Membrane filtration was started after 65 min until the end of aeration. The feed pump, air pump, discharge pump and permeate pump were pre-set according to the cycle pattern by a timer. Fine bubbles for aeration were supplied (20 L min⁻¹) from the bottom of the reactor with an air diffuser. Two flat-sheet membrane modules were submerged into the SBR after aerobic granules became mature. The hydrophilic membrane is made of polyvinylidene fluoride (PVDF) with a nominal pore size of 0.1 μm (Sinap Corp., Shanghai, China). Each membrane has an effective surface area of 0.0288 m². A permeate flux of 20 L m⁻²h⁻¹ was maintained by regulating the flowrate of the permeate pump (9 min ON and 1 min OFF, controlled by a timer). The transmembrane pressure (TMP) of the membrane module was monitored by a pressure transducer, which was connected to a personal computer equipped with a data logging system. All the experiments were conducted at room temperature of 26 ± 2 °C.

The granule sludge taken from a column-like SBR fed with acetate-based media in our laboratory was seeded into the AGS-MBR. The composition of synthetic wastewater was CH₃COONa (48.0 g L⁻¹), NH₄Cl (26.5 g L⁻¹), MgSO₄·7H₂O (0.4 g L⁻¹), K₂HPO₄ (3.5 g L⁻¹), CaCl₂·2H₂O (0.55 g L⁻¹), and FeSO₄·7H₂O (0.02 g L⁻¹). The prepared synthetic wastewater was stored in a fridge at 4 °C and diluted with tap water when used. The municipal sewage was taken from a local municipal wastewater treatment plant. The average values of TOC, NH₃-N and TN of synthetic wastewater and municipal sewage are given in Table 1.

The reactor fed with synthetic wastewater or municipal sewage was operated without membrane module immersion until the aerobic granules stabilized in the reactor (50 days and 70 days, respectively). Then, the bioreactor was tested with membrane filtration for 120 days. The operating conditions are summarized in Table 2.

Mixed liquor suspended solids (MLSS), MLVSS, sludge volumetric index (SVI₅), chemical oxygen demand (COD) and NH₃-N were measured as proposed by Standard Methods (APHA, AWWA, WEF 2012). TOC and TN were measured by the catalytic oxidation method (Shimadzu TOC-V_CPH). The particle size was measured by a laser particle size analysis system (Malvern Mastersizer 2000, UK) with a range of 0.02–2,000 μm. The D (4,3) was used to represent the equivalent volume diameter of the particles. The granule morphology was determined by an image analysis (IA) system (Olympus SZX9, Japan) with Image Pro Plus software (Media Cybernetics, L.P. version 4.0, USA). Fractal dimension (D_F) was used to determine the internal structure of aerobic granules according to the method reported by Li et al. (2008a). A high D_F indicates compact aggregates whereas low values correspond to more ‘loose’ aggregates. The EPS was extracted from the granules following a ‘formaldehyde-NaOH’ method described previously (Li et al. 2008b). In this study, extracted EPS was analyzed by examining polysaccharide concentration (mg L⁻¹) and protein concentration (mg L⁻¹). The protein concentration was determined by binding of Bradford reagent to the protein with bovine serum albumin as a standard and the polysaccharides concentration was determined according to the phenol-sulfuric acid method with glucose as a standard (Li et al. 2012a, 2012b).

A fluorescence spectrophotometer (LS 55, Perkin Elmer Company, USA) was used for obtaining the fluorescence EEM spectra. Emission spectra between the wavelengths of 230 and 550 nm were collected at 5 nm increments by varying the excitation wavelength from 230 to 550 nm at 5 nm intervals. Excitation and emission slits were set at 10 nm with a scanning speed of 1,000 nm min⁻¹. All samples were filtered with a 0.45 μm filter prior to analysis.

Figure 2 shows the MLSS and SVI of the AGS during the experimental periods of 120 days. The AGS in the AGS-MBR for sewage treatment demonstrated a higher MLSS than that of synthetic wastewater, whereas its SVI was apparently lower than synthetic wastewater. The results indicate that the aerobic granules fed with sewage have a higher density than that of synthetic wastewater. In this study, the MLVSS/MLSS ratio for the AGS-MBR treating municipal sewage is about 0.76, and the MLVSS/MLSS ratio of aerobic granules cultivated using synthetic wastewater was 0.85. Therefore, the higher density of the granules cultivated from sewage is likely attributed to the accumulation of inorganic matter in the granules of AGS-MBR when treating sewage (Wang et al. 2009). In addition, the granules in both of the reactors showed no disintegration for the duration of 120 days, as reflected by the equivalent diameter of the granules (Figure 3). This indicates that granules on both occasions may be at their critical sizes, at which a balance between the granule growth and granule breakage presents (Verawaty et al. 2013). It should be noted that the aerobic granules cultivated with synthetic wastewater had a higher average particle size (898 μm) than municipal sewage (788 μm). This is because readily biodegradable organics in the synthetic wastewater are prone to cause overgrowth of filamentous organisms, leading to a bigger particle size of the aerobic granules (Liu & Liu 2006; de Kreuk et al. 2010).

The morphological characteristics of the mature granules in the AGS-MBRs for synthetic wastewater and municipal sewage are shown in Figure 4. It can be seen that aerobic granules with a round shape and smooth surface dominated on both occasions. As an indicator of granule regularity, aspect ratio can be used as an effective tool to compare granules (Chen et al. 2008). In this study, the aspect ratio of granules for municipal sewage (0.61) is slightly higher than for synthetic wastewater (0.56), indicating more round granules were developed in municipal sewage treatment (Figure 2). It is also illustrated that D_F value in synthetic wastewater treatment (1.67) was smaller than municipal sewage (1.75). D_F reflects the internal structure of fractal aggregates, and it is a more sophisticated method than the visual observation. Higher D_F correlates to compact and dense aggregates while lower D_F represents loose or porous structure (Guan et al. 1998). The difference in D_F indicated that the granule structure cultivated with synthetic wastewater was relatively looser. This can also be observed in the IA that more filamentous bacteria were present in the granules (Figure 4). Therefore, the difference in granule morphology was likely due to the different substrate type, as readily biodegradable organics favor the growth of filamentous organisms. It should be noted that although the aerobic granules cultivated with the synthetic wastewater had a slightly bigger particle size, these granules are looser than that of sewage. Considering the MLVSS/MLSS ratios, it can be concluded that overgrowth of filamentous microorganisms leads to a smaller granule density compared with the granules feeding with sewage. The results support the fact that density contributed more to settling than particle size, which was in accordance with the observation with SVI.

Figure 5 illustrates the TOC removal of the AGS-MBRs for synthetic wastewater and municipal sewage treatment, respectively. The TOC of the sewage was about 10 mg/L higher than that of synthetic wastewater. In addition, the TOC removal efficiency for synthetic wastewater (96.1%) was apparently higher than that for municipal sewage (84.9%). The synthetic wastewater mainly consists of acetate and glucose and these readily biodegradable organics can be utilized by the microorganisms completely. However, the municipal sewage contains a wide range of organic compounds and some of the organics may be refractory or not readily biodegradable substances. Therefore, the relatively high supernatant TOC in municipal sewage treatment (13.3 ± 2.1 mg L⁻¹) was due to the accumulation of non-biodegradable organics either from the sewage or from the microbial excreted substances. It should be noted that some of the organics in the supernatant were retained by the membrane in the AGS-MBRs (about 2.5–3.0% of the TOC). The result indicates that the carbon removal of the AGS-MBRs was mainly through the biodegradation of the AGS.

Figure 6 gives TN and NH₃-N profiles of the influent and effluent of the AGS-MBRs for the duration of 120 days for synthetic wastewater and municipal sewage, respectively. The NH₃-N is the sole nitrogen source of the synthetic wastewater. Therefore, the TN concentration is equal to the NH₃-N concentration. The TN concentration was apparently higher than that of NH₃-N in the municipal sewage, implying other nitrogen types such as organic nitrogen and nitrate were also present. Although nitrogen was not at the same level in the two reactors, both of the reactors showed a comparable NH₃-N (90–95%) and TN (50–55%) removal efficiency over the duration of 120 days. The results indicate that AGS-MBR is effective in organic and nutrients removal during long-term operation. The simultaneous removal of organics and nitrogen was mainly attributed to the coexistence of heterotrophic, nitrifying and denitrifying populations in aerobic sludge granules, where nitrification occurs at the outer surface of granules and denitrification dominates in the inner part of granules (Bassin et al. 2012). The carbon and nitrogen removal performance demonstrated in this study indicated that the AGS-MBR has a stable performance in terms of pollutant removal. In addition, it is indicated that the test results with synthetic wastewater can be used to simulate the AGS-MBR performance with municipal sewage because the difference between these two feeding wastewaters is negligible.

Figure 7 illustrates the TMP profiles of the AGS-MBRs at a flux of 20 L m⁻²h⁻¹. The TMP was increased slowly from 0 to 25 kPa over the duration of 120 days without membrane cleaning, implying that membrane fouling of AGS-MBRs was apparently lower compared with flocculent sludge MBRs (Wang et al. 2013). In addition, both of the reactors demonstrate a similar TMP profile. The result indicates that the membrane fouling mechanism is similar regardless of the type of the wastewater. The membrane filtration resistance of the fouled membranes was assessed at the end of operation (Figure 8). It showed that irreversible fouling dominated (>85%) on both occasions, whereas the contribution of reversible fouling to the total filtration resistance was insignificant. In AGS-MBR, aerobic granules are much bigger than the membrane pores. This makes it difficult for them to attach or clog on the surface of the membrane. Therefore, the membrane foulant in AGS-MBRs is likely due to attachment of the colloid or solutes in the mixed liquor, to which SMP and EPS contribute the most (Li et al. 2012a, 2012b).

Figure 9 presents the SMP and EPS of the granular sludge with synthetic wastewater and municipal sewage, respectively. The content of SMP and EPS of municipal sewage was obviously higher than synthetic wastewater. The microorganisms in municipal sewage treatment have to produce more EPS than those of synthetic wastewater to maintain their own survival and growth because of its far more complex composition. SMP can be regarded as dissolved EPS, which is often the main source of components in the supernatant of granular sludge especially when readily biodegradable matter is used as substrate (Tu et al. 2010). The periodical starving condition in the granulation process makes the aerobic granules utilize a greater amount of polysaccharides in the EPS and SMP (Wang et al. 2013). As a result, protein is the dominant substance in the microbial biologically produced substances in AGS-MBRs (Figure 9), which is different from conventional MBRs (Kimura et al. 2015).

In order to further understand the composition of membrane foulant in the AGS-MBR for different wastewaters, the organic component extracted from the fouled membranes at the end of each operation was analyzed by EEM fluorescence spectroscopy (Figure 10). EEM spectra indicate that the main peaks found in the foulants on both occasions could be attributed to tryptophan-like substances originating from microbial by-products (peak I, at Ex/Em = 270 nm/370 nm) and aromatic proteins (peak II, at Ex/Em = 220 nm/350 nm) (Chen et al. 2003; Adav et al. 2008). The peaks dominating in the foulants suggested that tryptophan-like substances and aromatic proteins related to microbial processes were the major source of irreversible fouling resistance regardless of substrate type (Figure 10). It is believed that proteins have a hydrophobic nature (Verawaty et al. 2013). Therefore, the results indicate that protein in the EPS of AGS is more important to membrane fouling of AGS-MBR as it is prone to be attached by the hydrophilic membranes in the reactor.

In this study, two identical AGS-MBRs were developed and operated over 120 days for synthetic wastewater and municipal sewage treatment, respectively. Mature aerobic granules of 750–950 μm were developed and no disintegration was found in the experimental period of 120 days. The AGS-MBR demonstrates a stable carbon and nitrogen removal regardless of the substrate type. Moreover, the reactor was operated at 20 L m⁻² h⁻¹ for 4 months without membrane cleaning (TMP was only increased to 25 kPa), implying the advantage of the AGS for fouling mitigation. The filtration resistance analysis indicates that the main membrane resistance was due to the irreversible fouling in both of the reactors. The EEM analysis indicates that proteins in EPS are more prone to be attached by the membrane of AGS-MBRs because of their hydrophobic nature. This study shows that AGS-MBR is effective and stable for municipal sewage treatment and reuse during long-term operation.

The study was supported by grants from National Natural Science Foundation of China (Grant No. 51408351), and Shanxi Scholarship Council of China (Grant No. 2016-001).