Author:

Abstract:

In the last decade, membrane bioreactors (MBRs) have become a mature technology that can compete with conventional wastewater treatment processes. An MBR comprises an activated sludge bioreactor to biodegrade the polluting components and a membrane module which separates the biomass from the effluent by means of micro- or ultrafiltration. The latter replaces the sedimentation tanks of conventional activated sludge processes. While a reduced plant footprint and a better effluent quality, completely free of suspended solids, are among the many advantages, membrane fouling remains the major bottleneck. Despite extensive research efforts, membrane fouling in MBRs is still far from being fully understood, mainly due to the inherent interactions between the activated sludge, the influent and applied process conditions. Due to the lack of clear causal relationships between activated sludge mixed liquor characteristics and membrane fouling, in practice most, if not all, MBRs are operated in such a way that effects of membrane fouling are reduced. It is clear that the remediation of fouling increases the overall cost of an MBR and is, unfortunately, not always very effective.In order to be able to prevent membrane fouling, the focus in this dissertation has been on investigating the impact of the activated sludge mixed liquor characteristics on membrane fouling in MBRs. From lab-scale, pilot-scale and full-scale MBRs experiments the following results are obtained.Better bioflocculation ... less foulingThe influence of activated sludge deflocculation and reflocculation was studied in a lab-scale MBR. Activated sludge could be deflocculated and reflocculated by changing the ratio of monovalent over polyvalent cations ([M/P]) in the influent. As a result of a high [M/P] in the influent, severe sludge deflocculation occurred and worsened filtration characteristics were observed. Increasing fragment surface as a consequence of activated sludge deflocculation coincided with a reduced activated sludge filterability. Once the influent was switched to a low [M/P], the activated sludge reflocculated and filtration characteristics improved significantly within three weeks after the switch. With this experiment, it is evidenced that bioflocculation strongly influences membrane fouling in MBRs.Filtration characteristics deteriorate dramatically when activated sludge deflocculates whereas a reflocculated sludge improves the filtration performance significantly.Higher SRT ...better bioflocculation... less foulingA pilot-scale MBR was operated at different solids retention times (SRT) to study the influence of SRT on effluent quality, bioflocculation and membrane fouling in MBRs.Effluent quality was proven to be unaffected at an SRT of 50 days and 30 days, while at an SRT of 10 days nitrification decreased from 99% to 81%. Also COD removal rates were slightly lower compared to SRTs of 30 days and 50 days. Apart from effluent quality, also activated sludge bioflocculation deteriorated at an SRT of 10days. The deflocculation could be monitored by means of automated analysis of microscopic activated sludge images. An increasing relative fragment surface in the images, reflecting a more deflocculated state of the activated sludge, contributes to severe membrane fouling.Conclusively, a higher SRT induces a better bioflocculation which improves the filterability of the activated sludge and reduces the membrane fouling.It is all about bioflocculation ... or not?Ten full-scale MBRs were sampled in both winter and summer. Each of these activated sludge samples was subjected to the Delft Filtration Characterization method (DFCm) to determine the activated sludge filterability while a large set of parameters was measured. Although in literature most of these parameters have been indicated as major membrane foulants, no clear correlation between any single parameter and filterability could be established. This suggests that every parameter alone is a poor indicator of the activated sludge fouling propensity.The results from this sampling campaign reveal that a combination of sludge morphology and relative hydrophobicity (RH) allows for a first classification of activated sludge filterability. It was found that activated sludge deflocculation and low activated sludge RH have a negative impact on activated sludge filterability and vice versa. For sludge samples that have poor to good activated sludge filterability, accurate estimations of sludge filterability can be calculated on the basis of floc-bound proteins (eEPS PN), soluble polysaccharides (SMP PS), sludge morphology, the dissociation constant and RH.Finally, the impact of the activated sludge characteristics on membrane fouling was studied using a reverse engineering approach. A commercially available membrane performance enhancer (MPE50, Nalco) was added to activated sludge samples originating from five different membrane bioreactors treating municipal or industrial wastewaters. Before and after the addition of MPE50, a large set of activated sludge parameters was measured and filtration tests were performed in a lab-scale MBR. As a consequence of the MPE50 addition, activated sludge bioflocculation improved and the concentration of SMP in the bulk liquid strongly reduced. Also the concentration of eEPS reduced, though to a lesser extent. For sludges with a low RH, the addition of MPE50 increased the RH.These experiments confirm that a rough classification of activated sludge filterability can be made from RH and image analysis data alone. Based on activated sludge morphology, eEPS PN and RH, the activated sludge filterability can be accurately estimated for sludge samples with moderate to good filterability.