Colloidal organic matter fouling of UF membranes: role of NOM composition & size

The aim of this study was to fractionate pre-filtered surface water using a 3.5 and a 10 kDa dialysis membrane, and to compare the rate of fouling and the fouling reversibility/irreversibility of the NOM fractions. Trial dialyses (3.5 and 10 kDa) were carried out for 6 and 21 days with pre-filtered surface water using synthetic surface water as dialysate. The aim of the trials was to optimize the dialysis process for NOM fractionation. DOC, Ca2+, Mg2+, soluble silica and bacteria were monitored at intervals during the dialysis process. Thereafter, the various NOM fractions (with low and high Ca2+) were fed to a miniature UF system operated at a constant flux of 138.5 L/m2 h, filtration cycle times of 31.5 min and backwash duration of 1.75 min. A PES/PSV hollow fiber UF membrane (MWCO 100 kDa) with a surface area of 0.0125 m2 was employed for the filtration tests (X-Flow). Transmembrane pressure (TMP) and UF feed and permeate (LC-OCD) were monitored at regular intervals. For a dialysate recirculation of 95 L/h, sample to dialysate ratio of 5.2:80 L and a dialysate change frequency of 3 times per 24 h, the shortest duration of dialysis was about 6–7 days for both 3.5 and 10 kDa dialyses membranes. The removal of organic carbon (OC) increased with dialysis duration and MWCO of the bags. The biopolymer fraction increased from 120% to 240% when the duration of dialysis was increased from 6 days (1.1 mg DOC/L, 151 mg Ca/L) to 21 days (0.82 mg DOC/L, 133 mg Ca/L) with the 10 kDa dialysis membrane. The increased biopolymer fraction in the NOM sample that was dialyzed for 21 days resulted in a doubling of the fouling rate from 3.5 to 6.6 mbar/min per mg DOC/L. The other NOM fractions (humics and building blocks) and the Ca/DOC ratio was more or less the same in both NOM samples suggesting that biopolymers were the major cause of UF fouling.