For every kilogram of cheese a dairy makes, it is left with roughly nine litres of whey. The liquid is dilute, about 93 to 94 per cent water, but it holds most of the lactose of the milk, the whey proteins and a large share of the minerals (Walstra et al., 2006). Its organic load is high enough that discharge to a watercourse or an undersized treatment works causes serious pollution, and regulators treat it accordingly (Prazeres et al., 2012; Carvalho et al., 2013). The same composition makes whey a raw material. Smithers (2008) traced how whey moved, in his phrase, from gutter to gold as the nutritional and functional value of its proteins was recognised and as the means to recover them became available.

Those means were chiefly membranes. Ultrafiltration made it possible to concentrate whey proteins without heat or a change of phase, and the other pressure-driven processes followed for clarification, demineralisation and water removal (Pouliot, 2008). The plants that adopted them were large, because protein powders require evaporators and spray dryers that are costly and energy-hungry. Small and medium cheesemakers mostly continued to give whey to farmers for pig feed, spread it on land or pay for its treatment.

This review considers what membrane filtration can do for the smaller plant. The question is practical: which processes, in which combination, turn a liability into something saleable or at least cheaper to move, with equipment that a small team can run and clean.

The article is a narrative review of the published literature: textbooks and handbooks of dairy and membrane technology, review articles on whey use and on membrane processes in the dairy industry, and primary studies of whey ultrafiltration, nanofiltration and fouling. It was not conducted as a systematic search and reports no counts of records. "Small and medium" is used loosely for plants that cannot justify their own evaporator and dryer and that lack specialist membrane staff. Sweet whey from rennet cheeses is the main case; acid whey from fresh cheeses and strained yoghurt is noted where its behaviour differs. Electrodialysis, ion exchange and chromatography, which large processors combine with membranes, lie outside the scope.

Whey composition sets the terms for every separation. Lactose is by far the largest component of the solids, followed by minerals and by protein, of which β-lactoglobulin and α-lactalbumin are the principal fractions. Sweet whey also contains fine casein particles and residual fat from the vat, and it carries starter bacteria that continue to produce acid unless the whey is cooled or pasteurised promptly. Acid whey has more calcium and lactic acid and less lactose, which makes it harder to concentrate and dry (Bylund, 2015).

The range of products that can be made from whey is wide: whey powder, protein concentrates and isolates, lactose, permeate powder, fermented beverages, and feedstocks for ethanol, organic acids or biogas (Siso, 1996; Yadav et al., 2015). Most of these routes begin with a membrane step, either to separate protein from lactose or simply to remove water. Fermentation routes for the lactose fraction have been reviewed by Guimarães et al. (2010), who stress that they make economic sense only as part of an integrated scheme in which the protein has already been recovered.

Pressure-driven membrane processes form a series distinguished by pore size and by the pressure needed to drive liquid through the membrane (Cheryan, 1998). Figure 1 shows the series as applied to whey, and Table 1 compares the four processes on the points that matter to a small operator. In each case the feed flows tangentially across the membrane surface, the material that passes through is the permeate, and the material held back is the retentate.

The separations themselves are well understood. Whether a small plant can make them pay depends on a set of operating matters that receive less attention in the literature than membrane selectivity does.

Three configurations emerge from the literature as realistic. The simplest is reverse osmosis or nanofiltration alone, used to pre-concentrate whey before it is collected by a larger processor. It creates no new product, but it cuts transport and chilling costs and returns water to the plant. The second is ultrafiltration to produce a liquid protein concentrate for use on site, in the plant's own yoghurt, ricotta-type cheese, ice cream or beverages, where it replaces purchased milk solids and needs no drying. The third adds a volume-reduction step on the permeate so that the lactose stream too can be transported economically.

Shared facilities are a further possibility. Several small cheesemakers in one district can pre-concentrate their whey individually and send it to a common plant with drying capacity, or can jointly own a mobile or central membrane unit. The technical literature has little to say about such arrangements, although the logic of concentrating before transport is well established (Daufin et al., 2001; Pouliot, 2008).

Published performance data come mostly from laboratory rigs and from large plants. Information on how small installations fare over years of use, including membrane replacement intervals, cleaning costs, labour and downtime, is scarce in the open literature, and that information is what an investment decision needs. Techno-economic studies framed for plants of modest throughput, with realistic assumptions about seasonal milk supply and product prices, would be more useful to this sector than further refinement of separation performance.

Acid whey is a second gap. Its volume has grown with the market for strained yoghurt and fresh cheeses, many of them made by smaller firms, and its high mineral and lactic acid content makes both membrane concentration and drying harder. Work on nanofiltration and diafiltration strategies tailored to acid whey, and on low-cost uses for the resulting streams, would address a need that is increasingly felt.

Membrane filtration offers small and medium dairies more than one way to deal with whey, and the most refined product is seldom the right target. Reverse osmosis or nanofiltration for volume reduction, and ultrafiltration to make a liquid protein concentrate for the plant's own products, are within the technical reach of a modest operation. Success depends less on the membrane than on what surrounds it: clean, cool, clarified whey at the inlet, disciplined cleaning, gentle operation that limits fouling, and a definite outlet for the lactose-rich permeate. Where those conditions can be met, whey ceases to be only a cost. Where they cannot, a shared facility or simple pre-concentration for collection is the more prudent course.

To be declared by the author(s) before publication.

To be declared by the author(s) before publication.