One of the most reliable alternative solutions to water shortage and scarcity in urban areas is potable reuse of what would otherwise be considered “waste” water. The combination of low and highpressure membrane processes is the favoured technology for direct and indirect potable water recycling due to the very high water quality produced. Despite the fact that membrane technology is well established, membrane fouling remains a major challenge affecting plant operation technically and economically. This thesis aims to contribute to the understanding of membrane fouling in water reuse by investigating the importance of the feed water quality on fouling development and the impact of the fouling on the treated water quality. As yet, no studies have investigated tertiary effluent organic matter fouling using advanced tools with in addition experiments at laboratory and semi-industrial scales conducted in parallel allowing a direct comparison of the membrane performance for a better understanding of the membrane fouling.
In order to achieve these objectives, it was important to develop better knowledge of the impact of feed water quality on the membrane process. Three different secondary treated effluents from municipal wastewater treatment plants in Australia were studied. Effluents from an enhanced biological phosphorus removal plant (EBPR), a sequencing batch reactor (SBR) including chemical phosphorus removal and a conventional activated sludge plant with a short sludge age of 10 days (SSA) were selected to span a variety of different secondary treatment technologies. The organic composition of the effluents was successfully differentiated with the use of advanced analytical tools such as fluorescence excitation emission matrix (EEM) and liquid chromatography-organic carbon detection (LC-OCD).
To evaluate water quality impact on membrane fouling, a membrane pilot unit consisting of ultrafiltration (nominal pore size of 0.04 μm, 4 modules, 27.9 m2 per module, 35 to 40 L/h.m2.bar)and a two stage reverse osmosis train (18 modules, 7.9 m2 per module, 75 to 82% of recovery, 18 L/h.m2.bar) was operated at the EBPR (4 months) and the SBR site (6 months). Different monochloramine dosages (0 to 2 mg/L NH2Cl) were applied and their impact on RO membrane performance was studied in order to provide recommendations for plant design. The results demonstrated the possibility to decrease the monochloramine concentration without drastic membrane performance decline. Membrane autopsies were performed to confirm the nature of the foulants and the absence of major biofouling development. In order to establish the impact of the fouling on the product water quality, micropollutant rejection was studied on the SBR site with virgin and fouled membranes. The presence of organic matter on the surface of the membranes led to a general decrease in rejection. In addition, composition of microbial communities on RO membranes, an emerging topic, was investigated on both sites by an innovative method, the pyrosequencing of the 16S ribosomal RNA gene, showing that each site is specific leading to different microbial communities.
Innovative fouling indices were evaluated at bench and pilot-scales in order to investigate if simpler and less costly bench-scale experimentations can substitute pilot-scale trials. A bench-scale hollow fibre membrane system (52-58 cm2) was used to assess the fouling rate depending on the feed water characteristics while a pilot-plant was running continuously under similar conditions to obtain in parallel fouling indices. Whilst linear correlations between the results were obtained at both bench and pilot-scales for the total fouling index, there was no significant correlation of the hydraulically irreversible fouling index between the bench and pilot-scales. These results demonstrate the difficulties in replicating hydraulic conditions at bench-scale, in particular, with regards to backwashing. Nonetheless, statistical analysis of the fouling indices and water quality data, indicated that biopolymers and humic substances, the larger molecular weight organics, were found to be the major contributors to total fouling. Adsorption of the low molecular weight neutral compounds onto the membrane was attributed to cause hydraulically irreversible fouling.
Finally, membrane rejection of several contaminants was investigated at bench-scale using controlled operating conditions (constant pressure, pH and temperature) with environmental concentrations of organic micropollutants (ng/L). The fouled membranes from the pilot-plant on the SBR site were sacrificed and installed on a bench-scale RO unit (3 flat sheet cells in series, active area of 138 cm2) together with virgin membrane coupons to assess the impact of the membrane foulant on the removal of organic micropollutants. Similar conclusions were reached at both benchscale with synthetic water, and pilot-scale with the secondary treated effluent from the SBR. Indeed, the rejection decrease was more important with the secondary effluent at pilot-scale than with the synthetic water at bench-scale showing the importance of the matrix on the contaminant interactions. These results indicate that RO membranes are not an absolute barrier and that membrane fouling decreases the quality of the produced water.