There are many water quality characteristics which could influence the filterability of biologically treated effluent from Melbourne's Western Treatment Plant (WTP). Statistical correlation was used to identify the key water characteristics affecting the microfiltration (MF) and ultrafiltration (UF) filterability in terms of permeate volume of the treated effluent. The models developed showed that turbidity, dissolved organic carbon (DOC) and total suspended solids (TSS) were the key factors which influenced the MF and UF filterability. Turbidity was the dominant factor affecting the accuracy of the model for MF filterability while DOC was the major factor affecting the accuracy of the model for UF filterability. A prediction accuracy of 85% was obtained for MF and 86% for UF filterability of the WTP effluent. The characteristics of the organic components of the wastewater were demonstrated by EEM spectra to have seasonal variation which would have reduced the prediction accuracy. As turbidity, DOC and TSS can be determined on-line, the models would be useful for rapid prediction of the filterability of WTP effluent and this may assist the control of low-pressure membrane filtration processes.
Keywords: biologically treated effluent, filterability, low-pressure membrane filtration, statistical correlation