No abstract

Microfiltration (MF) of milk with a ceramic membrane (0·1 μm mean pore diam.) for the separation of casein micelles from soluble proteins was studied. Experiments were performed at constant flux density (J=65 or 75 l h−1 m−2) and wall shear stress (τw=100 or 110 Pa) with milks containing particles and solutes with different sizes and charges, produced by physicochemical change (heat and mechanical treatment, pH, ionic strength, addition of ions). Membrane separation performance was limited by the build up of a cake with characteristics that depended on the size of particles and soluble proteins. Best performance (higher permeability and whey protein transmission) was obtained with milk containing fat on the one hand and calcium phosphate on the other. An optimal value of the particle size was found, close to 0·5–3·0 μm; above this separation performance decreased. In addition, the present study confirmed that the transfer of charged solutes is the consequence of both size and ionic exclusion. Better performance was achieved with higher ionic strength (1 M).

Sorne experimental evidence is accumulating in the literature which shows that the mean cross-flow velocity is not an effective parame ter for the characterisation of the cross-flow transport in membrane separation processes. The aim of this work was to determine the most appropriate parameters (among the wall shear stress, t w ' the wall shear rate, y, the Reynolds number, Re, and the mean cross-flow velocity, v) that should be used to assess the limiting and critical fluxes in turbulent regime. The experiments were applied to skimmed milk microfiitration (MF) using a tubu-Jar ceramic membrane (0.1 IJ.m mean pore diameter) for the separation of casein micelles from the solluble proteins. These experiments were conducted with several membrane geometries (seven or 19 channel s, 40 or 85.6 cm long) and seve rai rigs. Unlike Re and v, t w made it possible to assess the Iimiting and critical permeation fluxes in milk MF (0.1 IJ.m),whatever the membrane geometry. It was consequently concluded to be an effective parameter to account for cake filtration and to study the crossflow filtration scale-up. The wall shear rate, calculated by assuming the turbulence damped at the membrane surface and the flow laminar, was not suitable for the assessment of the Iimiting and critical fluxes in turbulent regime.

No abstract

No abstract