New Procedure for Enhancing SAFT Molecular Parameters Transferability: The Role of Derivative Properties

The design and optimization of industrial processes, involving the correct selection of suitable solvents, the most advantageous unit operations, and an adequate separation sequence, relies upon the accurate prediction of the behavior the system at different thermodynamic conditions. Besides the phase equilibria description, knowledge of derivative properties, such as heats of vaporization, isochoric and isobaric heat capacities, isothermal compressibility and thermal expansion coefficients, are essential to fulfill the conditions of an optimized industrial process or product.

It has been shown that second order derivatives are more sensitive to errors than the first order derivatives of the thermodynamic function. Consequently, a correct description of first order derivatives, such as phase equilibria calculations, do not assure an accurate prediction of second order derivatives of the same thermodynamic function (heat capacities, speed of sound, etc.) for the same compound or mixture. In spite of the importance of derivative properties and their sensitivity to errors, most modeling efforts have been devoted to the description of standard phase equilibria, while the works on derivative properties are more limited.

We will present a new robust procedure, involving vapor-liquid equilibria (VLE) and derivative properties regression, in a step forward obtaining reliable and transferable parameters from molecular-based equations of state. The method deals with the introduction and testing of coupling factors ranging from 0 (only Cp) to 1 (only VLE), to change the weight of one set of properties over the other in the fitting procedure; this allows assessing the role of derivative properties in the robustness of the parameters. The technique is illustrated with the soft-SAFT equation by calculating a large number of thermodynamic properties of different compounds, and in a wide range of thermodynamic conditions, for compounds from the n-alkanes, n-perfluorocarbons, 1-alkanols and glycols families, as representative of different types of molecular interactions. The most relevant thermodynamic properties to be included in the parameters fitting, the weight they should have in the regression procedure and the influence of association will be discussed in detail.

The method presented here is robust, simple, straightforward to implement and not equation dependent; it can be transferred to any other equation, providing robust molecular parameters for global equations.

 Partial financial support for this work has been provided by the Catalan government under project 2014-SGR1582.