Methodology for accelerated pre-selection of UF type of membranes for large scale applications

Abstract

This paper describes two assessment methods for UF type of membranes for large-scale applications. The combination of those two methods results in quite clear and unambiguous answers to the question what membranes are of interest for long-term testing. With the first method, calld dead-end filtration method, information is generated on the suitability of the membrane and on the combination of the membrane material, the module hydraulics and assembly. With this method the evolution of TMP is monitored upon filtration cycles of 20 minutes with raw water at a flux rate of 120 l/h.m 2, alternated with backwash cycles with permeate of 40 seconds at 1.2 bar negative TMP. The second method, called cross-flow filtration method, gives exclusively information on the suitability of the membrane material. This is being done by the measurement of the absolute value of the so-called “plateau fluxes” in cross-flow mode at 0.2 m/s linear velocity. For this purpose raw water concentrates are being used. Three “open” UF type of membranes, all three in hollow fibre configuration were assessed with these two methods. It was shown that the PSf based membrane (Koch PM100) reached already after 4 filtration cycles a TMP of 1 bar and showed the lowest plateau flux (25 l/h.m 2). This indicated that the membrane suffered from interaction with the raw water. Moreover, it is possible that something was wrong with the hydraulics of this membrane. The two other membranes were PES/PVP based. These membranes showed much less TMP increase over time. The first membrane of this type was X-Flow UFC, the second Stork Friesland Superfil 015-010. It was no problem to operate the first membrane for 18 hours without addition of chemicals for cleaning. The second membrane reached the maximum allowed TMP of 1 bar after 16 hours of operation at the end of the filtration cycle. Moreover, for both membranes a higher plateau flux value (35 l/h.m 2) was found. Both observations indicate that this type of membrane material is much more interesting than PSf. It was also shown the X-Flow membrane gives the lowest absolute TMP values, which is attributed to its higher pure water permeability (740 l/h.m 2.bar) as compared to the Stork Friesland membrane (pure water permeability of 350 l/h.m 2.bar) and the Koch membrane (pure water permeability of 290 l/h.m 2.bar). A last observation was a TMP increase of only 0.1 bar per cycle for the X-Flow membrane, as compared to 0.2 bar for the two others. This observation is in agreement with earlier made FESEM pictures of the inner surfaces. This means that the X-Flow membrane rather acts as a depth filter, whereas the two other membranes act as a surface filter.