Performance of a Conductor-Like Screening Model for Real Solvents Model in Comparison to Classical Group Contribution Methods

Group contribution methods, such as modified UNIFAC (Do) are powerful tools for the reliable prediction of phase equilibria and excess properties, which are necessary information for the development, design, and optimization of separation processes and other applications of industrial interest (Gmehling, J. Present Status of Group-Contribution Methods for the Synthesis and Design of Chemical Processes Fluid Phase Equilib. 1998, 144, 37−48). Despite the efficiency and simple applicability, group contribution methods have the disadvantage that they need at least a limited number of experimental data for fitting the required group interaction parameters; however, the necessary data are sometimes missing so that predictions for these systems cannot be performed. The conductor-like screening model for real solvents (COSMO-RS) developed by Klamt et al. (Klamt, A. Conductor-Like Screening Model for Real Solvents: A New Approach to the Quantitative Calculation of Solvation Phenomena. J. Phys. Chem. 1995, 99, 2224−2235. Klamt, A.; Jonas, V.; Bürger, T.; Lohrenz, J. C. W. Refinement and Parametrization of COSMO-RS J. Phys. Chem. A 1998, 102, 5074−5085. Klamt, A.; Eckert, F. COSMO-RS: A Novel and Efficient Method for the A Priori Prediction of Thermophysical Data of Liquids. Fluid Phase Equilib. 2000, 172, 43−72) has attracted much attention in the chemical engineering community. From the results of quantum chemical calculations for a single molecule, the COSMO-RS model is used to compute the required activity coefficients and, therefore, provides a possible route for a priori prediction of phase equilibria. Since for this new method, up to now no comprehensive validation of the predicted results exists, a comprehensive comparison with the results of different group contribution methods was carried out using a large database (Dortmund Data Bank) (Dortmund Data Bank and DDB Software Package, DDBST GmbH, Oldenburg, Germany, 2004 (www.ddbst.de)). Not surprisingly, the comparison showed that much more reliable results were obtained using group contribution methods, in particular, modified UNIFAC. For example, in the case of VLE (database: >1360 thermodynamically consistent data sets), only a 0.35 mole % higher mean absolute deviation (0.94 mole % instead of 0.59 mole %) was observed for the vapor phase composition than for the correlation method UNIQUAC. For COSMO-RS(Ol)*, 5 times higher mean deviations (1.86 mole % (2.45 mole % instead of 0.59 mole %)) are observed.