Insights into the Limitations of Parameter Transferability in Heteronuclear SAFT-type Equations of State

Abstract

The use of heteronuclear models are often viewed as ways to improve the predictive ability and parameter transferability of advanced association models, such as those derived from the Statistical Associating Fluid Theory (SAFT). Indeed, several results in the literature have suggested that this approach can be useful to accurately describe a given family/series of homologous compounds and their mixtures, with accuracies competitive (and in some cases better) than those obtained using the more traditional SAFT variants. However, the parameter transferability of the different groups, i.e. between different families of compounds, without the introduction of new groups or refitting existing ones, is seldom reported, and often overlooked, making an accurate evaluation of the heteronuclear models difficult. This work analyzes whether the increased complexity of a heteronuclear treatment of a SAFT-type EoS, namely the SAFT-{\gamma}-Mie EoS, results in a significant increase on both the predictive ability and parameter transferability of the model, across different families of compounds. This is done by using a case study involving some different (yet related) families of compounds, containing a small number of common functional groups. The results obtained show that the transferability of group parameters, across different families of compounds, in a heteronuclear SAFT-type EoS does not allow an adequate description of the phase equilibria of these systems. Therefore, to achieve a reasonable accuracy in the description of these systems, a specific refitting of group parameters is required for a given family, or even for a particular system, destroying the predictive capability of these models. Moreover, this increases the number of adjustable parameters to numbers similar to those used in homonuclear approaches, further reducing the advantages of using heteronuclear models.