Analysis of separation and fouling behaviour during nanofiltration of dairy ultrafiltration permeates

Dairy ultrafiltration permeate (UFP) has been filtered using polyamide nanofiltration (NF) membranes in order to separate potentially valuable calcium and lactose from unwanted monovalent salts. An investigation into both the fouling and separation behaviour has been carried out to gather useful information for the optimal selection of operating conditions that maximise efficiency and yield. Supersaturated calcium phosphate in the feed resulted in severe flux decline during filtration at high temperatures and pH due to salt precipitation. Investigation into the ion separation behaviour using sub-saturated model solutions containing KCl and CaCl2 showed that the addition of the trivalent anions citrate and phosphate changed the pH dependent separation of Kþ and Ca2þ. For the multicomponent system, maximum calcium rejection and minimum potassium rejection occurred at high pH, whereas the opposite was true for calcium in the simpler systems. A possible explanation for this separation behaviour has been proposed which incorporates the membrane charge, represented by the zeta potential, and the speciation of calcium with these anions.

It was observed that calcium plays an important role in both the fouling and separation behaviour during NF of UFP. In particular, calcium phosphate contributes to severe flux decline at high temperature and high pH operating conditions. This is due to the precipitation of supersaturated calcium phosphate and formation of a cake layer. These fouling problems may be overcome by altering the pH of the feed, however changes in the separation profile of the membrane must then be considered. In the multicomponent system containing the common dairy multivalent ions, the maximum rejection of calcium and phosphate was observed at pH values close to the natural pH of UFP. In addition, minimum rejection of potassium also occurred at this higher value. The behaviour of calcium and potassium during NF of model solutions was explained by the surface charge behaviour and by the speciation of calcium with citrate and phosphate. It is likely that the pore charge would also have had a significant effect on the separation behaviour, accounting for the discrepancy between the Kþ rejection minimum and surface IEP.