Fouling membrane in an anaerobic membrane bioreactor treating municipal wastewater

In many tropical countries, up-flow anaerobic sludge blanket (UASB) reactors have now been accepted as a reliable technology for treating municipal wastewater at ambient temperature (Chernicharo et al. 2015). However, biochemical oxygen demand (BOD) removal efficiencies are usually lower than 70% and biomass may be lost during peak biogas production or hydraulic loads. In order to improve the effluent quality from anaerobic reactors, the addition of filtering membranes modules have been recently implemented, resulting in higher BOD and suspended solids removal efficiencies, as well as elimination of pathogens, thus complying with water quality standards (Stuckey 2012; Ozgun et al. 2013).

The application of anaerobic membrane bioreactors (AnMBR) for real municipal sewage treatment is scarce if compared to experiments using synthetic laboratory wastewater; in fact, the aerobic version or MBR has been extensible studied, being a commercial, well accepted technology nowadays (Judd 2016). Notwithstanding, in the last decade, the study of AnMBR using submerged membranes for municipal wastewater treatment has received more attention from researchers around the world. Different membrane configurations have been implemented for that purpose: flat sheet (Hu & Stuckey 2006), hollow fibers (Wen et al. 1999; Giménez et al. 2011), and tubular modules (Jeison & van Lier 2006; van Voorthuizen et al. 2008). The membrane modules have been used directly immersed in the bioreactor or in a separate tank (Liao et al. 2006). In addition, Liao et al. (2006) reported that the most studied AnMBR configurations are based on completely stirred tank reactors (67%), anaerobic filters (15%), UASB reactors (10%), fluidized bed reactors (7%) and septic tanks (2%).

Among the many advantages of submerged AnMBR treating municipal wastewater, some drawbacks may be identified, such as low permeate flux, frequent membrane fouling and high capital and operational costs (Ozgun et al. 2013). According to Stuckey (2012), membrane fouling is caused by a combination of factors in the reactor, such as the presence of soluble organics and colloidal particles from the feed, cell lysis, and precipitation of inorganic species; these factors are influenced by the composition of the biological and chemical systems involved, membrane type, hydrodynamic and reactor operating conditions. In particular, organic matter such as soluble microbial products (SMP) and extracellular polymeric substances (EPS) may reduce the membrane porosity and increase filtration resistance. The molecular size and concentration of SMP contribute to membrane fouling, negatively affecting permeate flux (Liang et al. 2007). In the same way, the operational mode of AnMBR has important effects on fouling. Intermittent filtration with backwashing or bubbling during relaxation time can be used in order to decrease membrane fouling (Cerón-Vivas et al. 2012).

The aim of this study was to evaluate the effect of gas sparging during relaxation time on fouling in a pilot scale AnMBR (UASB with a submerged ultrafiltration membrane) treating municipal wastewater at ambient temperature.

Municipal wastewater was fed into the UASB reactor using a peristaltic pump (Masterflex 77410-10, 0.35 hp, Cole-Parmer, USA). The permeate suction was performed by a peristaltic pump (Masterflex 7553-80, USA) and adjusted by means of a speed control. The permeate flux was collected and stored for further analysis. The transmembrane pressure (TMP) was recorded every 30 seconds by a pressure transducer (OMEGA PX319) located in the permeate line. Analog signals were processed by a data acquisition card connected to a computer using a Lab-View application. The system operated under intermittent filtration mode (4 min on/1 min off), with and without nitrogen gas bubbling (0.75 L min⁻¹) during the relaxation time (IF4NP and IF4P, respectively). Tests were conducted until TMP reached 40 kPa.

Samples of raw wastewater, UASB effluent and permeate were taken daily. Total and volatile suspended solids (TSS, VSS), chemical oxygen demand (COD) and pH analysis were made according to Standard Methods (APHA, AWWA & WEF 2012). SMP, EPS and particle size distribution were measured only in the UASB effluent. SMP were determined in the filtrate using a 0.45 μm filter (Nitrocellulose, Millipore, USA). EPS were extracted using a heating method based on the one reported by Zhang et al. (1999) and Ng & Ng (2010). The same filter used for SMP determination was introduced in an erlenmeyer flask with an equivalent sample volume of MilliQ water; it was heated at 80 °C for 10 min and filtered again through another 0.45 μm filter; EPS were determined in the filtrate. Both SMP and EPS were measured as total organic carbon (TOC) using a TOC analyser (Analytic Jena Multi N/C 2100). Particle size distribution was measured with a Mastersizer 2000 (Mastersizer 2000, Malvern, England). All the samples were analyzed in duplicate. Scanning electron microscopy (SEM) was performed on fouled membranes with a JEOL JSM-7600F SEM using samples previously fixed with 3% glutaraldehyde during 48 hours followed by dehydration in an ethanol:water series: 10%, 30%, 50%, 70%, 90% and finally, absolute ethanol. Then, the samples were dried in a critical point drier and sputtered with gold.

Table 1 shows the performance of the pilot-scale AnMBR at different operating conditions. Temperature varied between 18 and 21° C and pH was always slightly above neutrality. The average total COD removal in the AnMBR was 68.6% and 87.9% for IF4P and IF4NP, respectively. COD Removal efficiency for only the UASB reactor was limited to 52.1% and 79.5%, for IF4P and IF4NP respectively. In this process arrangement, the membrane unit removed 32.4% and 40.0% of the COD coming from the UASB reactor for IF4P and IF4NP respectively, due to the retention of suspended and colloidal matter by this unit. Although it was expected that all TSS were retained on the membrane, around 15 mg L⁻¹ were found in the permeate, due to biofilm growth on the permeate pipe walls. In general, these results show agreement with other authors who reported COD removal efficiencies between 80 and 90% in pilot-scale AnMBR treating municipal wastewater (An et al. 2010; Giménez et al. 2011; Salazar-Peláez et al. 2011).

According to the particle size distribution in the UASB effluent, the mean particle size diameter for IF4P and IF4NP were 40 μm and 50 μm respectively, indicating the presence of large aggregates, which are important in the formation and the structure of the cake layer (Lin et al. 2009). The values found in this study were higher than those from another pilot AnMBR study (Martinez-Sosa et al. 2011) but were similar to the mean particle size values found at bench scale (Hu & Stuckey 2006; Jeison & van Lier 2006).

The pilot scale AnMBR achieved high COD and TSS removal efficiencies and may be considered as a suitable technology for municipal wastewater treatment. Gas bubbling during one minute relaxation time promoted mixing in the upper part of the UASB reactor, resulting in an effective measure to minimize membrane fouling and increase COD removal. The comparison of the two operating conditions allowed to identify EPS as major contributors to membrane fouling.

This research was funded by a grant from the Instituto de Ingeniería UNAM (Fondo de Proyectos Internos). The authors thank Roberto Briones, Lauro Santiago and Margarita E. Cisneros for their technical support. The first author appreciates the support given by the Universidad Pontificia Bolivariana (Colombia) and the scholarship provided by the Instituto de Ciencia y Tecnología del Distrito Federal, México.