Pure Substance and Mixture Viscosities Based on Entropy Scaling and an Analytic Equation of State

This study proposes a simple model for viscosities, based on entropy scaling, for real substances and mixtures. The residual entropy is calculated with the perturbed chain polar statistical associating fluid theory (PCP-SAFT). The model requires two or three pure component parameters, noting, however, that an entirely predictive group contribution approach as proposed in our previous work [Loetgering-Lin O.; Gross J. Ind. Eng. Chem. Res. 2015, 54, 7942–7952] gives also very good results. Overall, 140 real substances are considered with relative mean deviations from experimental data of about 5% (without excluding “outliers”). We performed molecular simulations for mixtures of simple model fluid in order to determine a suitable mixture model. A completely predictive approach for viscosities of real mixtures is thereby obtained. The model is evaluated for 566 mixtures with about 34,500 experimental data points of various complexity (i.e., nearly ideal systems as well as highly asymmetric mixtures). Mixtures of nonpolar substances and mixtures with at least one polar, but nonhydrogen-bonding component, are predicted very accurately with relative mean deviations of on average 6.2% (173 mixtures considered) and 5.3% (126 mixtures considered), respectively. Limitations of the model are found for mixtures with hydrogen-bonding (associating) components such as amines and alcohols, where deviations are systematically higher. Lastly, we present results of mixture viscosities using the purely predictive group contribution framework and find similar results for the predictive approach.