Automation of group-contribution techniques for estimation of thermophysical properties

A computer algorithm is presented which predicts thermophysical properties of mixtures using only two-dimensional chemical structures. It applies a combination of techniques for the analysis of group contributions and other characteristics of a chemical structure in a table-driven format. These techniques are used to examine the error behavior for predictions of critical properties and saturated properties. UNIFAC predictions of activity coefficients are combined with pure-component properties to demonstrate that accurate estimates of azeotrope behavior and K-values can be obtained.