Integrated multi-objective membrane systems application of reverse osmosis at the Amsterdam Water Supply

Abstract

Within the scope of a project funded by AWWARF and USEPA three very promising IMS's were identified for surface water treatment. This paper will cover some highlights of the research carried out by Amsterdam Water Supply (AWS) and Kiwa on the combination of biological activated carbon filtration, slow sand filtration and (ultra low pressure) reverse osmosis. Primary objectives of this IMS study are restriction of (bio)fouling, productivity control, disinfection and removal of DBP-precursors. To prevent fouling two pretreatment schemes were tested. The reference scheme consisting of coagulation, sedimentation, rapid sand filtration and slow sand filtration lowered the MFI to 2 s/l 2. The alternative scheme consisting of the reference scheme extended with ozonation and biological activated carbon filtration before slow sand filtration lowered the MFI to 1 s/l 2. Biofouling was characterized by AOC and biofilm formation rate. Average AOC values were low (<10 μg/l) and the biofilm formation rates were lower than the standard for biologically stable drinking water. After 9 months of operation no significant membrane fouling was observed. For both pretreatments only a slight flux decline was observed during a period of nine months of operation. No significant increase in pressure drop was observed. After an initial flux decline during the first two months for both pretreatments only little flux decline occurred. The flux of the RO pretreated with only slow sand filtration (reference scheme) still showed a (small) decline after the nine months test period. After the initial flux decline, the RO with extended pretreatment consisting of an additional ozonation and biological activated carbon filtration (alternative scheme) operated at a nearly constant MTC. Disinfection and integrity were tested by MS2 phage spiking resulting in a removal capacity of 3.4 log units. Because of their larger size the removal of other micro-organisms should be at least equal. However, with direct cell counts and heterotrophic plate counts a lower efficiency was found. This might be due to regrowth in the RO installation on the permeate side. The DBP-precursor content after both pretreatments was low. The Cl 2 demand for the reference scheme was twice as high as the Cl 2 demand for the alternative scheme. The DBP-precursor content in the permeate of both treatments was below the detection limit. It is concluded that the IMS at Leiduin is reliable. Fouling is manageable and productivity is nearly constant. The dosing of MS2 phages shows an acceptable removal. DBP-precursors are removed to below the detection limits by the ultra low pressure membranes.