Accepted Manuscript Review Anaerobic dynamic membrane bioreactor (AnDMBR) for wastewater treatment: A review Yisong Hu, Xiaochang C. Wang, Huu Hao Ngo, Qiyuan Sun, Yuan Yang PII: DOI: Reference:

S0960-8524(17)31666-8 http://dx.doi.org/10.1016/j.biortech.2017.09.101 BITE 18929

To appear in:

Bioresource Technology

Received Date: Revised Date: Accepted Date:

30 June 2017 12 September 2017 15 September 2017

Please cite this article as: Hu, Y., Wang, X.C., Hao Ngo, H., Sun, Q., Yang, Y., Anaerobic dynamic membrane bioreactor (AnDMBR) for wastewater treatment: A review, Bioresource Technology (2017), doi: http://dx.doi.org/ 10.1016/j.biortech.2017.09.101

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Anaerobic dynamic membrane bioreactor (AnDMBR) for wastewater treatment: A review Yisong Hua, Xiaochang C. Wanga,b,c,*, Huu Hao Ngoc,d, Qiyuan Suna, Yuan Yanga a

Key Lab of Northwest Water Resource, Environment and Ecology, MOE, Xi’an University of

Architecture and Technology, Xi’an 710055, P.R. China b

Key Lab of Environmental Engineering, Shaanxi Province, Xi’an 710055, P.R. China

c

International Science & Technology Cooperation Center for Urban Alternative Water Resources

Development, Xi’an 710055, P.R. China d

Centre for Technology in Water and Wastewater, School of Civil and Environmental

Engineering, University of Technology Sydney, Sydney, NSW 2007, Australia * Corresponding author: X.C. Wang (Tel.: +8602982205841; E-mail: [email protected])

Abstract: Recently, an increasing level of attention has focused on the emerging technology of anaerobic dynamic membrane bioreactors (AnDMBRs), owing to its merits such as low membrane module cost, easy control of membrane fouling, low energy consumption and sludge production, as well as biogas production. As research on AnDMBRs is still in the nascent stage, an introduction of bioreactor configurations, dynamic membrane (DM) module, and DM layer formation and cleaning is firstly presented. The process performance of the AnDMBR for wastewater treatment is then reviewed with regard to pollutant removal, DM filterability, biogas production,

and

potential

advantages

over

the

conventional

anaerobic

membrane

bioreactor(AnMBR). In addition, the important parameters affecting process performance are briefly discussed. Lastly, the challenges encountered and perspectives regarding the future development of the AnDMBR process to promote its practical applications are presented. Keywords: Anaerobic dynamic membrane bioreactor; operational parameter; membrane fouling; wastewater treatment; biogas production

1

Contents 1. Introduction ................................................................................................................................... 3 2. Configuration, membrane module, DM formation and cleaning .................................................. 6 2.1. AnDMBR configuration and membrane module ............................................................... 6 2.2. DM formation and cleaning ............................................................................................... 9 2.2.1 DM formation and stable operation.......................................................................... 9 2.2.2 DM cleaning ........................................................................................................... 10 3. Process performance of AnDMBRs ............................................................................................ 12 3.1. Treatment performance .................................................................................................... 12 3.2. DM filtration performance ............................................................................................... 13 3.3. Biogas production ............................................................................................................ 15 3.4. Advantages of AnDMBRs over AnMBRs ....................................................................... 16 4. Factors affecting process performance ........................................................................................ 17 4.1. Membrane and bioreactor configuration .......................................................................... 18 4.2. Membrane properties ....................................................................................................... 19 4.3. Wastewater characteristics ............................................................................................... 20 4.4. Sludge properties .............................................................................................................. 20 4.5. Operation conditions ........................................................................................................ 22 5. Challenges and perspectives........................................................................................................ 24 6. Conclusions ................................................................................................................................. 27 Acknowledgements ......................................................................................................................... 27 References ....................................................................................................................................... 28

2

1. Introduction In light of the recently increasing interest in wastewater treatment using sustainable

technologies,

energy-intensive

treatment

processes

are

being

reconsidered with the aim to reduce energy consumption and environmental impact, as well as to recover resources and bioenergy. Compared to the mainstream aerobic activated sludge processes, anaerobic technology is regarded as a reliable and promising alternative for wastewater treatment (Liao et al., 2006; Ersahin et al., 2014) owing to lower energy consumption, high organic loading rate, low sludge production, and bioenergy recovery (Shoener et al., 2016; Ersahin et al., 2017). However, the retention of slow-growing anaerobic biomass was the most important challenge in the early development of an appropriate anaerobic bioreactor (Zhang et al., 2011; Ozgun et al., 2013). As such, the combination of an anaerobic treatment process and membrane filtration technology, the so-called anaerobic membrane bioreactor (AnMBR), has been successfully applied at various scales to maintain anaerobic microorganisms (Liao et al., 2006; Smith et al., 2012; Visvanathan and Abeynayaka, 2012). Hydraulic retention time (HRT) and solids retention time (SRT) in AnMBRs can be independently controlled to extend process applicability for treating wastewaters of various strengths from industries (Liao et al., 2010; Dereli et al., 2012) and municipalities (e.g., landfill leachate) (Ho et al., 2007; Kim et al., 2011; Bohdziewicz et al., 2008; Xie et al., 2014; Nie et al., 2017), as well as solid wastes (such as food wastes and surplus activated sludge) (Meabe et al., 2013; Li et al., 2015; Tang et al., 2017). Thus, the HRT could be largely reduced (e.g., 3 h) (Hu and Stuckey, 2006), but SRT could be maintained at a relatively high value (50-700 d) to produce a solids-free effluent with high chemical oxygen demand (COD) removal owing to the retention of slowly degradable organics within the reactor (Stuckey, 2012; Ozgun et al., 2013). Recent studies have comprehensively reviewed the treatment of various wastewaters, process performance and influencing factors, and 3

membrane fouling characterization and control, and developments in AnMBRs (Stuckey, 2012; Skouteris et al., 2012; Smith et al., 2012; Ozgun et al, 2013; Lin et al., 2013). However, despite these important achievements, it was also recognized that the intensive use of AnMBRs was limited by several critical obstacles, such as low flux, membrane fouling, and high capital and operation costs (Ozgun et al, 2013; Smith et al., 2014). It is well-recognized that anaerobic sludge in AnMBRs is characterized by high viscosity and mixed liquor suspended solids (MLSS) concentration. It also contains large amounts of biopolymers and inorganic substances, resulting in higher fouling potential than activated sludge in aerobic MBRs. The attachment and accumulation of solid particles (such as fine sludge particles, biopolymers, and inorganics) on the membrane surface is a common phenomenon in (An)MBRs, causing membrane fouling (Meng et al., 2009; Guo et al., 2012; Meng et al., 2017). Membrane fouling can be divided into two categories, namely cake layer formation on the surface of the membrane and pore clogging (Ersahin et al., 2016a). It is commonly accepted that cake layer formation is the main contributor to fouling in aerobic and anaerobic MBRs, as in most cases the cake layer filtration resistance accounts for over 80% of the total filtration resistance (Meng et al., 2007; Hu et al., 2016a and 2016b). However, the formed cake layer could function as an additional filter (secondary membrane or dynamic membrane) owing to its capability to reject various pollutants and pathogens (Smith et al., 2015; Ersahin et al., 2016a). As such, the rejection properties are more dependent on the cake layer rather than the underlying membrane (Hu et al., 2016a), and thus, a cheap support material (such as meshes, woven, or non-woven filter cloth) enabling the formation of a cake layer could be used instead of microfiltration/ultrafiltration (MF/UF) membranes in AnMBRs (Ersahin et al., 2012). The well-formed dynamic membrane (DM) layer (also called a secondary membrane) can be used as a filter prior to the support material, and it can provide effective retention in both aerobic DM bioreactors (DMBR) and anaerobic

4

DM bioreactors (AnDMBRs) combined with several unique merits (such as low membrane cost, high flux, and easy cleaning). However, to date, according to limited reviews of DM technology (Ersahin et al., 2012; Zhang et al., 2014b), most research has mainly focused on applications of aerobic DMBRs from 2000 onwards, and research on AnDMBRs picked up the pace only after 2010. The literature shows that the application of AnDMBRs is still in its early stages, although the technology offers cost-effective wastewater treatment (Zhang et al., 2010; Ersahin et al., 2017). The current research areas regarding the AnDMBR process focus on its feasibility and performance in treating various wastewaters, influencing factors, optimization of membrane module and anaerobic bioreactor, and characterization of bulk sludge and DM layer properties (An et al., 2009; Xie et al., 2014; Alibardi et al., 2014; Ersahin et al., 2017). Limited attention is paid to modeling the DM filtration process, DM formation mechanism, DM layer characterization, and biogas production and collection (Ma et al., 2013a; Ersahin et al., 2016a; Saleem et al., 2017). In addition, a systematic review of the development of the AnDMBR process is lacking. Thus, more efforts are required to address the various aforementioned aspects, in order to promote the application of AnDMBRs. Thus, in this review, we introduce the emerging AnDMBR process, particularly bioreactor configuration, DM membrane module, DM filtration process, and its potential merits. We then discuss several important factors that can affect process performance from various aspects. Finally, the challenges and perspectives of the AnDMBR process for wastewater treatment are outlined to advance its practical application.

2. Configuration, membrane module, DM formation and cleaning 2.1. AnDMBR configuration and membrane module The types of anaerobic bioreactors include completely stirred tank reactor (CSTR), upflow anaerobic sludge blanket reactor (UASB), expanded granular sludge bed 5

reactor (EGSB), fluidized bed reactor (FBR), and others. It is worth noting that, so far, most anaerobic bioreactor types have been successfully coupled with MBRs to create various AnMBRs (Ozgun et al., 2013; Chen et al., 2016; Shoener et al., 2016). However, to date, only CSTRs and UASBs have been integrated with DM filtration technology to create AnDMBRs (Walker et al., 2009; Ma et al., 2013; Ersahin et al., 2014; Quek et al., 2017). As for the membrane modules, three typical module types could be chosen: hollow fiber, flat-sheet, and tubular. However, except for the flat-sheet and tubular DM modules, coarse pore support material is not reportedly used to assemble hollow fiber DM modules (Loderer et al., 2012); this is likely because of the low intensity of DM support material and the complicated fabrication procedures. Based on the relative location between the anaerobic bioreactor and membrane module, membrane configurations could be divided into the submerged (including internal and external submerged) and side-stream types (Liao et al., 2006; Smith et al., 2012; Shoener et al., 2016). The main difference is that in the submerged configuration, the membrane module is directly installed into the bioreactor (internally or externally), with the membrane being operated under vacuum (Ersahin et al., 2014; Quek et al., 2017), whereas for the side-stream configuration, the membrane module is located outside the bioreactor in an additional membrane tank and the membrane is operated under pressure (Alibardi et al., 2016; Ersahin et al., 2016b). Conventional AnMBR studies have reported that gas sparging energy and cost of a submerged membrane are approximately three times lower compared to the side-stream AnDMBR for a given flux (Jeison and van Lier, 2008; Ersahin et al., 2017). Moreover, the energy demand per permeate flow volume for submerged configurations was much lower than that for pumped side-stream configurations in AnMBRs (Martin-Garcia et al., 2011). Irrespective of the type of reactor (AnMBR or AnDMBR), for external configurations, high cross-flow velocity (CFV) could cause two different effects,

6

resulting in high energy consumption and deteriorating sludge properties (such as reduced particle size and microbial activity). High CFV would also contribute to effective membrane fouling control. No consensus is forthcoming on membrane configurations in AnDMBRs, as until now, most studies have used the submerged configuration. Only one study has been conducted at the lab-scale to compare the impact of bioreactor configurations (submerged versus side-stream) on treatment and filterability performance without further assessment of applicability (Ersahin et al., 2017). However, similar to the experience gained with conventional MBRs and AnMBRs, the submerged configuration is preferred owing to low energy consumption for effluent extraction and low investment cost. The latter results from that additional space is not needed for membrane installation in the internal submerged configuration unlike the side-stream configuration in AnDMBRs (Ersahin et al., 2017). As shown in Fig. 1, all types of anaerobic bioreactors could theoretically be used to develop AnDMBRs, by coupling with submerged and side-stream membrane configurations. Thus, many potential AnDMBR configurations may be developed and verified in the future. The commonly used DM module types are flat-sheet and tubular (as presented in Fig. 1), both of which have been applied in AnDMBRs with different membrane configurations (submerged and side-stream). However, unlike the variety of commercial membrane modules that could be chosen and applied at different scales of AnMBRs (Lin et al., 2013), there are no commercially available DM modules, and modules used in AnDMBR studies are completely self-made. Only a few studies utilized tubular DM modules to develop AnDMBRs (Walker et al., 2009; An et al., 2009; Zhao et al., 2010; Liu et al., 2016), possibly owing to the fact that tubular modules are difficult to assemble compared to flat-sheet modules. Moreover, tubular DM modules show a structure similar to flat-sheet DM modules but different from traditional tubular UF/MF membranes, as they exhibit a double-faced, large inner

7

and outer diameter filter assembled on the outside of the supporting skeleton (Liu et al., 2016). Most commonly used flat-sheet DM modules consist of a frame, and inner and outer layers of support material (Chu et al., 2010; Zhang et al., 2011; Ersahin et al., 2012; Hu et al., 2017a; Quek et al., 2017). The frame is used for holding the inner support layers and outer support layers together, and the material of the frame is similar to that of the common flat-sheet MF/UF membrane, including PVC, stainless steel, and other materials (Fan and Huang, 2002; Chu et al., 2010; Liu et al., 2016). The inner support layers are used to support the outer support layers, and thus, the materials of the former should have high intensity and chemical stability. Thus, stainless steel is mostly used (e.g., large-pore-size stainless steel (10 mm)) (Hu et al., 2017a and 2017b). The outer support layers use the so-called support materials, including meshes, woven, and non-woven fabrics (Ersahin et al., 2012) to support the DM layer for effective solid–liquid separation in both aerobic and anaerobic DMBRs. Previous studies indicate that the flat-sheet module is mainly used in submerged AnDMBRs rather than in side-stream ones (An et al., 2009; Ma et al., 2013; Ersahin et al., 2014; Alibardi et al., 2014; Alibardi et al., 2016). The flat-sheet DM module that could be used in side-stream configurations is quite different from that observed in AnMBRs; however, the reason for this is unclear. The tubular modules are also mostly used in submerged AnDMBRs (Ersahin et al., 2012), largely because they are of a structure similar to the flat-sheet modules (as mentioned above) and are not suitable for use in a side-stream AnDMBR configuration with their present structure. Fig. 1.

2.2. DM formation and cleaning 2.2.1 DM formation and stable operation The commonly used supporting material in AnDMBRs with pore sizes ranging from 10–200 μm cannot effectively and independently retain fine particles and

8

colloids. Therefore, the formation of a stable DM layer determines membrane filterability. Studies on AnDMBRs adopted two operation modes for effluent extraction, namely, the gravity-driven mode (Zhao et al., 2010; Zhao, 2012; Liu et al., 2016; Tang et al., 2017) and the pump-driven mode (An et al., 2009; Xie at al., 2014; Ersahin et al., 2014; Alibardi et al., 2014; Ersahin et al., 2017). This results in filtration under constant pressure (water head) and constant flux, as well as different DM formation processes and filtration behavior (Alibardi et al., 2014; Liu et al., 2016). Several studies have used the gravity-driven mode for DM operation to allow continuous effluent extraction without relaxation (Zhao et al., 2010; Zhao, 2012; Liu et al., 2016; Tang et al., 2017). Zhao et al. (2010) reported that the DM formation time in an anaerobic bioreactor was approximately 40–50 min, during which the effluent suspended solids (SS) rapidly decreased from 345 mg/L to zero. Effluent turbidity was as low as 1.5 NTU and remained stable after DM formation. Moreover, the initial filtration flux ranged from 300–350 L/m2h, stabilized at 15–30 L/m2 h, and could be maintained for approximately 12–15 d. Another study noted that 100 min was needed for DM layer formation, which was accompanied by a decrease in flux from 137 L/m2 h to 6 L/m2h as well as a decline in effluent turbidity from 78 NTU to 10 NTU (Zhao, 2012). The stabilization of the flux and turbidity at 6 L/m2 h and 10 NTU, respectively, after 120 min indicated the formation of the DM layer, and the stable operation period lasted for approximately 7 d. The analysis illustrated that under the gravity-driven mode, the flux and effluent turbidity in AnDMBRs generally presents a multistage variation tendency (rapid initial decline followed by a slow decrease), similar to the phenomena detected in the recently extensively investigated gravity-driven MBR process using MF/UF membranes (Wu et al., 2016). However, the two processes differed in terms of initial flux and effluent quality. Nevertheless, the formation of the DM layer was quick (7 d) (Zhao et al., 2010; Zhao, 2012; Liu et al., 2016;

9

Tang et al., 2017). In addition, many AnDMBR studies adopted the pump-driven mode to intermittently produce permeate similar to that in MBRs (Ersahin et al., 2014; Alibardi et al., 2014). Ersahin et al. (2014) claimed that a stable DM layer formed after 10–20 d operation, and flux, turbidity, and COD removal rates could be stabilized after 30 d operation at a constant flux of 2.6 L/m2 h in a submerged bioreactor. In a cross-flow AnDMBR for synthetic wastewater treatment at fluxes of 1.0–7.2 L/m2 h, however, the DM layer formation required 10–25 d but lasted for a longer operational period (approximately 50 d) (Alibardi et al., 2014). Thus, the results indicated that under a constant-flux operation mode, a long time was necessary for DM formation, but the stable operation period could be prolonged. 2.2.2 DM cleaning Xie et al. (2014) investigated the usage of mixed-liquor cycling to induce a CFV along the DM surface for fouling control. Physical cleaning by water flushing was performed to remove the sludge cake layer when severe membrane fouling occurred, as evidenced by the trans-membrane pressure (TMP) reaching 40 kPa. Using these methods, a longer and sustainable operation was obtained owing to effective control of the DM fouling and complete DM layer detachment and reformation. Biogas circulation was also employed for scouring the DM by installing a diffusing device below the membrane module, along with intermittent operation (effluent extraction of 190 s followed by hydraulic backwashing of 35 s), and membrane fouling was thereby substantially mitigated (Ersahin et al., 2014). Biogas scouring and hydraulic backwashing were combined for regeneration of the DM. In another study, AnDMBR was reported to treat municipal wastewater at a high flux of 65 L/m2 h, and water backwashing effectively restored the DM permeability when the TMP exceeded 25 kPa (Zhang et al., 2011). In a cross-flow AnDMBR with an external membrane for sludge filtration, a peristaltic pump could circulate mixed liquor along the mesh surface and provided a sufficient CFV to control the fouling layer thickness 10

and membrane permeability (Alibardi et al., 2014). An et al. (2009) attempted to remove membrane fouling and recover membrane permeability using a chemical cleaning procedure (0.5% (v/w) NaClO solution, 2 h duration). Generally, the chemical cleaning efficiency was satisfactory; however, poor effluent quality was observed at the initial stage after chemical cleaning, subsequently entering the stable stage after 1–2 d of operation. These analyses demonstrated that it is feasible to implement an individual or combined strategy (such as sludge circulation, biogas recycling, and relaxation) for DM fouling control in AnDMBRs. When serious fouling occurs, physical cleaning is sufficient for efficient DM layer removal and permeability recovery in most cases, similar to the cleaning performance observed in aerobic DMBRs (Fan and Huang, 2002; Chu et al., 2010; Hu et al., 2016a). Limited research has reported on chemical cleaning for DM regeneration (An et al., 2009). Thus, considering the easy operating and control as well as nonuse of chemicals, physical cleaning could be the preferred strategy for DM regeneration, whereas chemical cleaning could be an alternative only when serious physically irremovable fouling occurs (An et al., 2009; Xiong et al., 2014). Based on the recent achievements in aerobic DMBRs and AnDMBRs, the basic information in terms of DM layer formation process and mechanism, DM characterization technique, and cleaning strategies are summarized and presented in Table 1. Table 1

3. Process performance of AnDMBRs 3.1. Treatment performance Based on recently reported results, Table 2 presents the treatment performance of AnDMBR applications. All these studies were conducted at the lab-scale mainly using synthetic wastewater. The major pollutants were COD, ammonia, total nitrogen (TN), total phosphorus (TP), SS, and turbidity. Less attention has been paid 11

to emerging micropollutants likely to cause environmental risks (Luo et al., 2014). When simulated wastewater (COD of 1200 mg/L) was subject to AnDMBR treatment, COD removal ranged from 92%–94%, accompanied with high rejection of low-molecular-weight organics. However, the removal of ammonia was as low as 13000 mg/L) and ammonia (>3000 mg/L) contents in the influent, the average COD removal rate was approximately 62.2%, with no obvious removal of ammonia. Ersahin et al. (2014) used monofilament-woven fabric with pore size of 10 µm to develop a membrane module for AnDMBR operation. COD, turbidity, and SS removal exceeded 99% when treating high-strength synthetic wastewater, but much lower treatment efficiencies for TN and TP (average removal rates of 20% and 13%, respectively) were noted. Similar results were reported regarding COD removal (63%–80%) and SS removal (>90%) (Zhang et al., 2011; Alibardi et al., 2016; Saleem et al., 2016). However, Ma et al. (2013a) noted a satisfactory removal of COD and SS, low removal of ammonia, and high TP removal (60%) using the AnDMBR. This was attributed to the retention of particulate phosphorus, consistent with the good removal of SS (>90%) for treating real municipal wastewater. Thus, the results indicated that AnDMBR can provide stable removal of some pollutants (such as COD and SS), a COD removal rate of 60%–90%, and SS and turbidity removal rates of 90%–100% (Xie et al., 2014; Ersahin et al., 2014; Alibardi et al., 2016a). This was attributed to the fact that a large amount of COD in wastewater existed in particulate form, which could be easily removed by the adsorption and biodegradation by microorganisms, as well as by the retention effect of the stable DM layer. In addition, the treatment performance of TN, TP, and ammonia were generally poor (all removal rates were lower than 30%) (Ersahin et al., 2014). As simulated wastewater was commonly used in the AnDMBR studies,

12

nitrogen and phosphorus contained mostly soluble substances. Coupled with a low biological nutrient removal (BNR) effect in the single-anaerobic environment, this resulted in poor nutrient removal performance. The unexpected high removal of TP (60%) might be due to the fact that real municipal wastewater containing plentiful particulate phosphorus was used rather than synthetic wastewater (Ma et al., 2013a; Ersahin et al., 2014). Table 2

3.2. DM filtration performance Table 3 shows the DM filtration performance based on the results from the literature. The information presented includes support material and module properties, membrane flux, TMP, and CFV, and so on. As mentioned in Section 2.3.1, two modes were adopted for the effluent extraction, namely, the gravity-driven mode and pump-driven mode. Generally, the different effluent production modes affected DM filtration behavior with regard to changes in flux, effluent quality (such as turbidity and SS), and duration of the operation cycle. Relatively few investigations have adopted the gravity-driven mode for effluent extraction. Zhao (2012) studied the filtration behavior of the DM layer using a water head of 2.5 cm for permeate production during the stable operation period. The results indicated that the flux showed a decreasing tendency, falling from 300–350 L/m2h to 15–30 L/m2h, for an operation cycle of approximately 15 d. During six successive operation cycles, similar variation patterns of the flux were noted. However, a continuous reduction in initial fluxes after physical cleaning was detected (from 350 L/m2 h of the first cycle to 300 L/m2 h of the last cycle), which was attributed to the irremovable fouling caused by residuals retained on the support mesh. On the contrary, the physical cleaning showed no obvious effects on DM layer formation and stable operation processes. Several recent works used the gravity-driven mode for waste digestion (such as waste sludge and food wastes) to produce valuable products such as volatile fatty acids (VFAs) and lactic acid (Liu et 13

al., 2016; Tang et al., 2017). For high-concentration solid waste treatments, the HRT was very long accompanied by a low flux (