Ultrafiltration permeate flux decline prediction for gel layer forming solutes using monotubular ceramic membranes

Experimental results on permeate flux decline with time obtained from ultrafiltration tests were compared with those predicted by means of Song and Elimelech’s model. Feed solutions consisted of aqueous solutions containing 15 g/L of polyethylene glycol 35 kg/mol. All model parameters were theoretically estimated, except the gel layer concentration that was estimated as described by Lee et al. Experimental tests were performed with monotubular ceramic Tami MSKT membranes from Tami Industries (France). TMP (0.2, 0.3, 0.4 and 0.5 MPa) and crossflow velocity (1, 2 and 3 m/s) were varied. Permeate flux slightly decreased with time for all the experimental conditions tested. The best predictions were obtained for long time scales for a crossflow velocity of 1 m/s. For those experimental conditions the probability of gel layer formation on the membrane surface is the highest. For short time scales, model predictions showed a rapid initial permeate flux decline that was not experimentally observed because it may occur instantly before the first permeate was collected.

For low crossflow velocities, as the probability of gel layer formation is the highest, the values of the steady-state permeate flux predicted by the model are consistent with the experimental results. The best concordance between experimental and predicted permeate flux was observed for the lowest TMP for all the crossflow velocities tested. Moreover, the discrepancy between calculated and experimental initial permeate flux was high. This suggests that pore blocking and/or adsorption, which are not considered by the model, may occur instantly after the ultrafiltration experiment begins. The rapid initial flux decline caused by pore blocking and/or adsorption could not be detected in the experimental tests.