Prediction and correlation of triglyceride–solvent solid–liquid equilibria with activity coefficient models

Abstract

Solid–liquid equilibria (SLE) for solvent–triacylglyceride (TAG) systems are studied with activity coefficient models. Systems consisted of 29 data sets of C10:0 to C18:0 saturated triglycerides in nine solvents, benzene, acetone, carbon disulfide, carbon tetrachloride, chloroform, diethyl ether, ethanol, ethyl acetate, and n-hexane making up 562 data points. Eighteen activity coefficient models containing 0–2 fitting parameters were evaluated including a number of Flory Free Volume (FFV) and Entropic Free Volume (FEFV) models evaluated by Coutinho et al. [J.A.P. Coutinho, S.I. Andersen, E.H. Stenby, Evaluation of activity coefficient models in prediction of alkane solid–liquid equilibria, Fluid Phase Equilibria 103 (1995) 23–39], the UNIFAC model and its variations, the Wilson, UNIQUAC, and van Laar equations and some proposed modifications of the FEFV and UNIFAC models. The modifications were as follows: (i) a Flory type χ parameter was added to the FEFV model (FEFV-X) and (ii) the exponent in the UNIFAC combinatorial term was treated as a fitting parameter (UNIFAC- C). Models were compared in terms of absolute average percent deviation (AAPD) in temperature. The best predictive model for TAG–solvent systems was found to be the Dortmund version of UNIFAC. Addition of a fitting parameter to the UNIFAC models provided significant improvement. For the UNIFAC-1977, UNIFAC (Lyngby) and UNIFAC (Dortmund), these exponents were found to be 0.897, 0.858, and 0.749, respectively. The best single parameter correlative model was found to be the FEFV-X model, which gave an AAPD of 0.305%. The Wilson equation was the best two parameter correlation equation and provided an AAPD of 0.154%.