Serum protein removal from skim milk with a 3-stage, 3× ceramic Isoflux membrane process at 50°C

Small pore microfiltration (MF) can be used to remove serum proteins (SP) from skim milk. The process's SP removal efficiency directly influences the technology's economic feasibility. Our objective was to quantify the capacity of 0.14μm ceramic Isoflux MF membranes (TAMI, Nyons, France) to remove SP from skim milk. A 3-stage, 3×, feed-and-bleed MF study with diafiltration in the latter 2 stages was conducted at 50°C using Isoflux membranes to determine cumulative SP removal percentages and SP removal rates at each processing stage. The experiment was replicated 3 times starting with 3 separate lots of raw milk. In contrast to 3× MF theoretical cumulative SP removal percentages of 68, 90, and 97% after 1, 2, and 3 stages, respectively, the 3× Isoflux MF process removed only 39.5, 58.4, and 70.2% of SP after 1, 2, and 3 stages, respectively. Previous research has been published that provides the skim milk SP removal capacities of 3-stage, 3× 0.1μm ceramic Membralox (Pall Corp., Cortland, NY) uniform transmembrane pressure (UTP), 0.1μm ceramic Membralox graded permeability (GP), and 0.3μm polymeric polyvinylidene fluoride spiral-wound (PVDF-SW) MF systems (Parker-Hannifin, Process Advanced Filtration Division, Tell City, IN) at 50°C. No difference in cumulative SP removal percentage after 3 stages was detected between the Isoflux and previously published PVDF-SW values (70.3%), but SP removal was lower than published GP (96.5%) and UTP (98.3%) values. To remove 95% of SP from 1,000kg of skim milk in 12h it would take 7, 3, 3, and 7 stages with 6.86, 1.91, 2.82, and 17.98m(2) of membrane surface area for the Isoflux, GP, UTP, and PVDF-SW systems, respectively. The MF systems requiring more stages would produce additional permeate at lower protein concentrations. The ceramic MF systems requiring more surface area would incur higher capital costs. The authors hypothesize that SP removal with the Isoflux membranes was lower than theoretical for the following reasons: a range of membrane pore sizes existed (i.e., some pores were too small to pass SP), the selective layer modification and reverse flow conditions at the membrane outlet combined to reduce the effective membrane surface area, and the geometric shape of the Isoflux flow channels promoted early fouling of the membrane and rejection of SP by the foulant.