Coagulation pre-treatment for microfiltration with ceramic membranes

Ceramic microfiltration membranes need to operate at high flux, with very little pre-treatmentchemicals, in order to compete with polymeric membranes in surface water treatmentapplications. In-line coagulation pre-treatment prior to MF/UF membrane filtration is reported toreduce membrane fouling therefore, allow higher operational fluxes. The advantages of (in-line)coagulation are attributed to reduced pore penetration due to agglomeration of colloids, and amore porous filter cake. Particular conditions which lead to improvement of filterability, andreduced colloidal fouling, and its alleviation in ceramic membrane systems, are sparselyelaborated in literature.In this study, filtration behaviour employing various coagulation conditions: coagulant dose(ferric chloride), Gt induction, and feed water pH were investigated. As an intermediate step priorto employing ceramic membranes, filtration tests were performed with an Amicon (Un)stirredCell in dead end and constant pressure mode (TMP=0.1 MPa), with prefiltered (122ìm) surfacewater. A hydrophilic PVDF 0.1ìm microfiltration membrane was employed in the (un)stirred cellfiltration tests.The effect of coagulation conditions on filterability were assessed by processing time and weightdata from permeate collected via an electronic balance. Subsequently, the MFI was calculatedfrom the slope of the Ät/ÄV versus V plots. The calculated MFI values were utilized to comparethe fouling behaviour of various feed water after coagulation pre-treatment. The organic carboncontent of the feed and permeate water was analyzed and compared.In the presence of coagulant, cake filtration was the dominant filtration mechanism observed witha Gt ranging from 9,400 to 140,100. A coagulant dose of 1mg/l effective Fe 3+, at pH 8,employing a Gt of 140,100 gave the lowest MFI observed. This dose (1mg/l) was lower and theoptimum pH (pH=8) higher than suggested in the classical ferric chloride solubility diagram forsweep coagulation and charge neutralisation.Increasing and decreasing the coagulant dose from this local minimum for a range of Gt values(9,400 to 140,100) increased the MFI. The variation in MFI was very high (BBB) at a lowcoagulant doses. For example, the MFI decreased by 66% when the Gt (flocculation time) wasincreased from 9,400 to 140,100 for a coagulant dose of 0.5 mg/l, suggesting that the Gtemployed for (in-line) coagulation was very critical at low coagulant doses ( 15mg/l), as the reduction in MFI was only 10% when the Gt was increased from 9,400 to 140,100for a coagulant dose of 15 mg/l.Varying the pH between 7 and 8 had marginal impact on MFI. It is much less influential than thedose and Gt induced. Nevertheless, higher pH consistently gave lower MFI, and vice versa in therabge pH 6-8. The higher MFI values at pH of 6 were attributed to a lower permeability of thefouling layer due to more denser and less porous flocs.Computation of MFI via the slope of the relation (Ät/ÄV)*0.5/ÄV improved the accuracy of theMFI. This was concluded based on R-squared data from regression analysis. The R square valuefrom regression analysis of MFI computation could be regarded as an index of accuracy whencake filtration is present. Accuracy of results in terms of high R square values was found mostlydepending on the method of time series data processing.This study focussed on the impact of (in-line) coagulation on the filterability of surface water.However, this is only one aspect in the optimization of in-line coagulation of ceramic membranefiltration. To assess the overall rate of fouling, the backwashability of the fouling layer and theefficiency of the chemically enhanced backwash and chemical cleaning to remove membranefoulants need to be assessed in a follow up stud