Skim milk was concentrated at 10 °C using forward osmosis (FO), reverse osmosis (RO) and pressure-assisted forward osmosis (PAFO). A pressure of 40 bar, in the form of draw solution osmotic pressure (FO and PAFO modes) or transmembrane hydraulic pressure (RO mode) was applied; an additional hydraulic pressure of 2 bar was applied in the PAFO mode. More severe protein fouling was observed in RO, followed by PAFO and then FO. This was credited to the difference in the initial permeate flux, induced by the different effective driving pressures, with RO having a greater deviation of the initial flux from the critical flux value. The critical flux was determined for the FO and RO modes using a step-wise increase of draw solution osmotic pressure or hydraulic pressure, at a constant milk solids content. The critical flux was between 5.4 L/m2h (1.5 × 10−6 m3/m2s) and 7.2 L/m2h (2 × 10−6 m3/m2s) for both the FO and RO modes at a cross flow velocity of 0.2 m/s. The similarities in the critical flux for FO and RO suggests that the critical flux does not depend on the nature of pressure applied on the system (hydraulic or osmotic). Therefore, when operated at the same flux and crossflow velocity, FO would not fundamentally provide a lower fouling environment compared to RO. An increase of the solids content from 8.7% to 17.3% caused a reduction in the critical flux from 5.4 L/m2h to 3.1 L/m2h (8.5 × 10−7 m3/m2s).
The concentration of skim milk and whey was investigated at a pilot scale using forward osmosis membranes with an installed membrane area of 24 m2. The pilot plant was operated in batch mode using a draw solution (48–57 g/L of NaCl) that mimics the potential brine streams available in a dairy processing plant. This approach avoids or limits the need for the regeneration of a synthetic draw solution. A concentration factor of ∼2.5 was achieved for both the skim milk and fresh whey, resulting in a total solids concentration of ∼21 wt% and 15 wt%, respectively. Increasing the transmembrane pressure was found to be effective in improving the water flux, whereas a much greater increase in the draw solution osmotic pressure would be needed to achieve the same enhancement of flux. This study also showed that small organic molecules, such as lactose, were not fully rejected by the forward osmosis membranes. A cleaning protocol was established for recovering the membrane performance after milk and whey concentration. The specific energy required for milk and whey concentration using only the forward osmosis step (5–10 kWh/t water removed) is much lower than that required by reverse osmosis. Forward osmosis is an energy efficient and effective process for dairy applications if unlimited access to a brine stream can be made available within or in the proximity of dairy processing plants.
Cheese whey was concentrated to a concentration factor of 2.7 in a pilot scale forward osmosis filtration system, using a commercial cellulose triacetate membrane in a spiral-wound configuration. The whey was concentrated in a batch mode, using sodium chloride as the draw solution at initial osmotic pressures of 53–75 bar. During the process, flux was shown to reduce due to the simultaneous decrease in the bulk osmotic pressure of the draw solution, increase in the bulk osmotic pressure of the whey and the effect of concentration polarisation on both sides of the membrane. The flux is known to be driven by the effective osmotic pressures of whey and the draw solution on the surface of the membrane active layer. A short-cut approach that requires minimal information in advance about the osmotic pressure of whey and the geometry of the filtration system was implemented, enabling the determination of these effective osmotic pressures. The results obtained were shown to be in agreement with the fundamental forward osmosis flux model. The short-cut approach can be utilised for estimating effective osmotic pressures of other liquid food streams to be concentrated by forward osmosis, without the need of external measurements.