Estimation of Wilson parameters by non-linear regression

Abstract

The Wilson parameters for binary systems are usually estimated by one of the general purpose non-linear estimation computer programs through the minimization of the sum of the squares of the deviations between prediced and observed values. In order to speed up the computation and to fit the program to computers of any size, a computer program solely for the estimation of Wilson parameters of a polar mixture is developed through non-linear regression analysis. A Newton's iteration scheme is utilized for estimating the binary constants from isothermal vapor-liquid equilibrium data. A Newton-Raphson two-dimensional search scheme is used for estimating the Wilson parameters from isobaric experimental data. To promote fast convergence, a pair of approximate values of Wilson parameters are calculated from one point values of experimental data.