Multiple Solutions When Fitting Excess Gibbs Energy Models and Implications for Process Simulation

Excess Gibbs energy models are often fitted to binary experimental data of phase equilibria. Even in the case of two binary parameters fitted to two binary infinite‐dilution activity coefficients, multiple solutions for the parameters may occur. The present work shows that this is also true, if more data – like vapor‐liquid equilibrium data sets – is used. Chemical systems and conditions are identified that lead to multiple solutions. It is shown that the multiple solutions may lead to significantly different results when the models are used in simulation of separation processes. The occurrence and implications of multiple solutions when fitting parameters of excess Gibbs energy models are studied along the NRTL model as an example. Recommendations for avoidance of multiple solutions are given.