Application of CP-PC-SAFT and PC-SAFT for Simultaneous Prediction of LLE and High-Pressure VLE with Universal k₁₂ Values in Systems of Gases and n-Alkanes with N-Methyl-2-pyrrolidone and Some Substituted Aromatic Compounds

This study examines the accuracy of the simplest polarity- and association-neglecting forms of the critical point-based revision of perturbed chain-statistical association fluid theory (CP-PC-SAFT) predicting phase equilibria in the systems of N-methyl-2-pyrrolidone (NMP) with n-alkanes and nonpolar gases using the universal value of the binary parameter k₁₂ = 0.023. Previously, this value was obtained for the propane-phenol system, and its applicability for predicting phase equilibria in systems of various substituted aromatic and cyclic compounds by CP-PC-SAFT is assessed. It is found that this approach accurately predicts high-pressure vapor-liquid ones (VLE) in the systems of NMP with nitrogen, methane, ethane, and carbon dioxide, slightly overestimates the solubility of argon in NMP, and overpredicts immiscibility in the systems of propane and n-butane. Additionally, despite some scattering, CP-PC-SAFT yields reliable predictions of the upper critical solution temperature (UCST) data in the systems of NMP-heavier n-alkanes including n-hexadecane. It is also found that similar to the previously considered systems of substituted aromatic compounds, CP-PC-SAFT accurately estimates the compositions of the NMP-rich LLE phases but overpredicts the content of NMP in the n-alkane-rich ones. These results are compared with the predictions of PC-SAFT with various sets of pure compound parameters for NMP and the previously investigated substituted aromatic compounds. It appears that the lack of a standardized PC-SAFT parametrization scheme represents a significant obstacle to the implementation of this model to systems of different solvents with transferrable universal k₁₂ values. At the same time, PC-SAFT approaches that consider self-association may predict broader LLE phase splits. Incorporating LLE data into the parametrization procedures of such approaches can significantly improve accuracy at the expense of predictive capability.