Comparative study of different nanofiltration and reverse osmosis membranes for dairy effluent treatment by dead-end filtration

The process waters of dairy industry issued from the starting, the equilibrating, the interrupting and the rinsing steps of the different plant units contribute, besides the cleaning in place, to the effluent production. Their treatment by membrane aimed to concentrate waste organic matter and to use permeate as disposable water for reuse, lowering both the load and volume effluent and the total water consumption of dairy plants. The present work was focused on the concentration of 1/3 diluted skimmed milk (chemical oxygen demand, COD approximate to\ 36 g O-2/L) to about 1/1 milk (volume reduction factor, VRF 3), with nine nanotiltration and reverse osmosis membranes by dead-end filtration. COD was the selected criterion for permeate quality, i.e. rejection of organic milk components assigned to lactose. High COD rejection (>99%) was achieved whatever the membrane and the feed concentration. Rejections of divalent cations >90% were too high for being in accordance with negative rejection of chloride at VRF 3 using NF membranes. The negatively charged proteins at pH 6.6 were likely entrapped in a soft gel which was observed at the end of the run of dead-end filtration. This gel was reversibly removed by a flush with tap water. Dead-end filtration appears as an useful tool to show the relative content of permeate and the occurrence of a limiting flux upon concentration involving a gel formation. At the end of run (VRF 3) with an initial highly charged feed, COD of permeate was always far away the quality of water for human consumption (total organic carbon, TOC < 2 mg/L) but RO permeate can be released as waste. Water quality close to vapour condensates, issued from milk and whey drying steps, is needed for reuse in boiler feed; it should be likely reached with an RO + RO cascade and possibly with a single RO with a low charged feed. (C) 2004 Elsevier B.V. All rights reserved.