Considering the dispersive interactions in the COSMO-SAC model for more accurate predictions of fluid phase behavior
Journal
Fluid Phase Equilibria
Journal Volume
367
Pages
109-116
Date Issued
2014
Author(s)
Abstract
A term to consider the contribution of the dispersive interactions to the non-ideality of mixtures is introduced into the COSMO-SAC model on the basis of molecular simulation data from classical model force fields. This dispersion term is a one-constant Margules equation, where the constant is determined from the molecular dispersion parameter of the components. Furthermore, an atomic contribution method is proposed to calculate the dispersion parameter for a given molecule. For binary systems containing molecules consisting of C, H, N, O, F and Cl atoms, a total of 13 global parameters is introduced with the COSMO-SAC-dsp model. These parameters are obtained from regression to a large training set of binary vapor-liquid equilibrium (VLE) data from experiment. The overall deviations for VLE calculations on this training set are reduced by 25% in terms of the vapor pressure and 12% in terms of the vapor phase mole fraction. This dispersion term can provide a significant improvement for infinite dilution activity coefficient predictions, where the accuracy was increased by around 33%. © 2014 Elsevier B.V.
A term to consider the contribution of the dispersive interactions to the non-ideality of mixtures is introduced into the COSMO-SAC model on the basis of molecular simulation data from classical model force fields. This dispersion term is a one-constant Margules equation, where the constant is determined from the molecular dispersion parameter of the components. Furthermore, an atomic contribution method is proposed to calculate the dispersion parameter for a given molecule. For binary systems containing molecules consisting of C, H, N, O, F and Cl atoms, a total of 13 global parameters is introduced with the COSMO-SAC-dsp model. These parameters are obtained from regression to a large training set of binary vapor-liquid equilibrium (VLE) data from experiment. The overall deviations for VLE calculations on this training set are reduced by 25% in terms of the vapor pressure and 12% in terms of the vapor phase mole fraction. This dispersion term can provide a significant improvement for infinite dilution activity coefficient predictions, where the accuracy was increased by around 33%. © 2014 Elsevier B.V.
Type
journal article