Sieving variations due to the choice in pore size distribution model

The effect of the choice of the standard probabilistic model to describe the pore size distribution was theoretically studied on predicting membrane performance parameters, area average water flux and area average membrane sieving coefficient. Preliminary discrete pore size distributions were generated from rejection profiles of dextran and PEG for 10,000, 30,000 and 100,000 molecular weight cutoff (MWCO) polysulfone and cellulose acetate membranes. The standard probability distribution functions (PDF), gamma, lognormal, normal. Weibel and Rayleigh were used to fit the resulting pore size distribution data. It was observed that the area averaged sieving coefficients are sensitive to the choice of the PDF. These results implied that an uncertainty in the choice of distribution in describing the membrane morphology could lead to a propagated uncertainty in predicting overall membrane performance. (C) 2002 Elsevier Science B.V. All rights reserved.