SHORTER COMMUNICATION LOW-COST MEMBRANES FOR USE IN A SUBMERGED MBR I.-S. CHANG, M. GANDER, B. JEFFERSON and S. J. JUDD School of Water Sciences, Cran®eld University, Bedfordshire, UK C apital costs of a MBR (Membrane Bio-Reactor) process are substantially in¯uenced by the membrane cost itself, which contributes around 50% of the total process capital cost. The feasibility of a submerged MBR ®tted with three pore-sized NWPP (Non- Woven Poly-Propylene) membranes has been studied. The performance was compared with that of a conventional polysulfone (PS) membrane material. Both the PS and NWPP membranes demonstrated a drastic reduction in permeate ¯ux at the start of operation. Although the NWPP membrane showed a greater fouling propensity than the PS membrane, the difference in ¯ux decline between the two membranes was not great. All membranes produced an ef¯uent extremely low in organic matter and with a low permeate turbidity (<1 NTU), and ammonia removal was higher than 60% in all cases. However, whereas the PS membrane achieved a 7-log reduction in total coliforms, the NWPP membranes achieved only a 2 to 4-log reduction. Consequently, low-cost NWPP membranes can be considered suitable for use in MBR processes for municipal wastewater treatment, but are possibly less suited to domestic wastewater reuse where disinfection is a prerequisite. Keywords: activated sludge; low-cost membrane; membrane bioreactor (MBR); non woven polypropylene (NWPP); total coliform. INTRODUCTION Municipal wastewater is usually treated by the activated sludge process (ASP), which utilizes micro-organisms for natural degradation of pollutants such as organic carbon and nitrogen compounds. The ASP not only requires large aeration and sedimentation tanks, but also generates large quantities of excess sludge. In addition, the ASP suffers from solid-liquid separation problems, such as bulking and foaming. A technical improvement to this system is the membrane bioreactor (MBR), which replaces the two stages of the conventional ASPÐclari®cation and settlementÐ with a single, integrated biotreatment and clari®cation step. The advantages offered by the MBR over the conven- tional ASP include a small footprint and reduced sludge production through maintaining a high biomass concentra- tion in the bioreactor 1,2 . The system is also capable of handling wide ¯uctuations in in¯uent quality, and the ef¯uent can be reused directly for non-potable purposes because treatment ef®ciency is such that a high-quality product water is generated 3,4,5 . Furthermore, an increased rate of nitri®cation can be achieved since a large amount of slow-growing nitrifying autotrophs can be retained in an aeration tank 6 . Notwithstanding these advantages, the widespread appli- cation of the MBR process is constrained by the high capital, maintenance and operating costs. In recent reviews covering membrane applications to bioreactors it has been shown that, as with other membrane separation processes, membrane fouling is the most serious problem affecting this system 7,8 . Fouling leads to permeate ¯ux decline, making frequent membrane replacement and cleaning necessary, which then increases maintenance and operating costs. Fouling problems can be greatly ameliorated by operating at low ¯ux, but this then demands a greater membrane area. Capital costs may be substantially reduced using low-cost micro®ltration membrane materials such as extruded and non-woven polymers 9 , both of which are readily produced from low-cost materials by simple continuous processes. Non-woven ®brous polypropylene membranes (NWPP), such as those used in this study, are produced by a process known as `melt-blowing’. In this process the polymer is extruded through a die and the resultant ®bres air-blown at elevated temperatures onto a moving belt, producing discontinuous ®laments 0.5±3 mm in diameter which form amorphous felt. This material can then be calendered (i.e., compressed) to an appropriate thickness, such that the permeability and nominal pore size of the product membrane material are determined by the ®lament diameter and the degree of calendering. This is an inherently fast (up to 4 ms ±1 ) and low-cost process. Raw material costs are less than £1 kg ±1 . Thus to produce a `50 g weight’ material (i.e., 50 gm ±2 ), the material cost is below 5 pm ±2 , and the total processing cost would be in the region of a further 10± 20 pm ±2 , depending on the exact processing requirements. Thus, a total material cost of in the region of 25pm ±2 for a non-woven ®brous polypropylene membrane material would not be unreasonable. The bulk of the membrane costs, therefore, would be in the surface modi®cation, if hydrophilic materials are required and, most signi®cantly, in the module fabrication. For a simple melt-joined 183 0957±5820/01/$10.00+0.00 q Institution of Chemical Engineers Trans IChemE, Vol 79, Part B, May 2001