Investigating the impact of production conditions on membrane properties for MBR applications

Hollow fibres used in membrane filtration units submerged into the biological stage of a municipal wastewater treatment plant, so-called membrane bioreactors (MBR), are investigated in terms of pore structure, membrane morphology, and membrane material characteristics. In order to reduce production costs and to optimise the filtration performance, membrane samples manufactured under different production conditions are analysed by means of pure water filtration tests, molecular weight cut-off (MWCO) measurements, scanning electron microscopy (SEM), atomic force microscopy (AFM), contact angle measurements, and X-ray photoelectron spectroscopy (XPS). These analytical tools are evaluated regarding their significance to predict the impact of production conditions on membrane properties and by that on the susceptibility to fouling. In this paper, initial results of membrane characterisation are discussed and the next steps to develop a membrane characterisation protocol are outlined.

The overall objective of these studies is to investigate the impact of production conditions on membrane properties and thereby on the filtration performance of submerged hollow fibres in activated sludge filtration. Therefore, virgin hollow-fibre membranes provided by KMS and manufactured under different production conditions have been investigated by means of pore structure, membrane morphology, and membrane material characteristics. The membrane samples investigated have shown different filtration and separation performances in terms of pure water permeability and MWCO measurements. The trends of these experimental results are qualitatively consistent among each other and with respect to SEM images. Thus, both pure water filtration tests using a single fibre test rig and rejection tests using a cross-flow test cell are suitable techniques to estimate the comparative performance of the membrane samples. Nevertheless, the mean deviation from mean permeability is amounting to 11–32% and the mean deviation from mean MWCO to 6– 19% of the respective arithmetical average. Thus, efforts will be made to improve the reproducibility of experiments by pre-treating the virgin membrane samples with wetting agents. In addition to dextran rejection tests, rejection tests using suspensions of uniform latex particles will be conducted. Moreover, SEM measurements will be continuously performed to observe the impact of production conditions on pore structure characteristics. To conclude, the production time in combination with the chemical key concentration shows a significant influence on the pore structure characteristics of the membrane. While reducing the production time to 1.4% and the key concentra- tion to 10% compared to F2 leads to a more open pore structure and a lower dextran rejection, a reduction to 0.2% respectively 5% leads to denser membranes with a higher dextran rejection. Hence, optimal production conditions with respect to membranes with high pure water permeabilities and high MWCO values arise in-between the production parameters of sample F2 and F5. This type of membrane is favoured as long as the demands to fulfil legal requirements regarding the effluent quality are not too high, since operational costs are increasing with an increasing filtration resistance of the membrane. In addition, costs for membrane production are decreasing with a decreasing production time and chemical key concentration, which is another advantage of membrane F3 and F5 in comparison to membrane F2. Nevertheless, pure water filtration tests only are not capable of predicting the performance in activated sludge filtration. Thus, complementary pilot-scale long-term filtration tests under real process conditions were conducted within this research project [37,38]. In accordance with the pure water filtration tests, membrane sample F3 shows continuously the highest permeability during activated sludge filtration for 15 months. AFM might be a useful tool to estimate the average surface roughness of virgin membrane samples, which is a criterion to evaluate the affinity to fouling. Initial experiments have shown that a relatively huge number of tests are required to gain an average surface roughness since the margin of deviation is apparently high. In addition to that, drops of water on the surface of the membrane cause measurement errors which lead to very high peaks in the AFM image. Thus, it is difficult to distinguish drops of water among peaks of the membrane material. Up to now, these limitations of AFM are too significant to interpret the images shown in Fig. 4. As a result out of these initial investigations, future experiments will be carried out while the membrane sample is submerged into pure water, which in turn increases the complexity and the expenses of these tests.