Phase equilibrium calculations may be made by either semiempirical or theoretical methods. In the semiempirical approach one uses an empirical equation of state for one or more of the phases involved; for the liquid phase one of the empirical equations for the excess Gibbs energy (Wilson, van Laar, etc.) is usually used. The semiempirical approach gives good results provided that one has a significant amount of experimental data available for the mixture.However, such methods are better suited to interpolation of existing data than to extrapolation or prediction. Theoretical methods are based in statistical thermodynamics, require less mixture data, and should be more reliable for prediction. Theoretically-based methods that have found extensive use by chemical engineers include regular solution theory (1), corresponding states methods (conformal solution theory) (2-4), and perturbation expansions based on a hard sphere fluid as reference system (3,_4) • F o r mixtures of simple nonpolar molecules these methods give good results, especially the corresponding states and perturbation expansion theories (3). However, all three theories are based on the assumption that the molecules are spherical, with intermolecular forces that are a function only of the intermolecular separation. This assumption is strictly valid only for mixtures of the inert gases (Ar, Kr, Xe) and for certain fused salts and liquid metals. In spite of this restriction the corresponding states and perturbation methods have been applied with success to mixtures in which the intermolecular forces depend on the molecular orientations, e.g., mixtures containing 02, N 2 , light hydrocarbons, etc. (see ref. 4 for review of applications up to 1973). The extension to weakly nonspherical molecules can be accomplished, for example, by introducing shape factors as suggested by Leland and his colleagues (5). These methods have been extensively exploited for both thermodynamic (2) and transport (6) properties. They are predictive only if equations are available for the composition dependence of the shape factors. The existing methods for doing this work well for relatively simple mixtures, but break down when constituents with 344 Downloaded by GEORGE MASON UNIV on December 20, 2014 |