Multiobjective Optimization of PCP-SAFT Parameters for Water and Alcohols Using Surface Tension Data

Abstract

With predictive methods, such as classical density functional theory and predictive density gradient theory (pDGT), it is possible to model bulk phase properties and interfacial tensions using the same model. For nonassociating fluids, these models can be used to predict interfacial properties for systems that lack experimental data. For associating components, however, predictions often show large deviations to experiments, which is at least partially rooted in highly correlated pure component parameters. Therefore, we use interfacial properties for discriminating pure component parameters by amending the PCP-SAFT parameter estimation for water and alcohols by including surface tension data in the objective function. To obtain a comprehensive comparison between different association models, a multiobjective optimization is performed. By analyzing the resulting pareto fronts, it is shown that including a fitted dipole moment improves the results for water but not for alcohols. The result of the multiobjective optimization is inconclusive about the optimal choice of association scheme for water as the preferred model changes along the pareto front. For small alcohols, in contrast to chemical intuition, the 4C association scheme gives the best results. For longer alcohols, the pareto analysis shows the limits of the homosegmented modeling approach.

Modeling Subsurface Hydrogen Storage With Transport Properties From Entropy Scaling Using the PC‐SAFT Equation of State

“…In contrast to hydrogen, the accurate prediction of thermodynamic properties of water is a challenging subject, even for modern EoS. The parameter set for water used in this work was proposed by Rehner and Gross (2020) and we chose the non‐polar water model in combination with a 2B association scheme. They included surface tensions from predictive density gradient theory in addition to vapor pressures and liquid densities in a multi‐objective optimization to ensure that the resulting parameter set is suitable for the calculation of interfacial properties.…”

PCP-SAFT Parameters of Pure Substances Using Large Experimental Databases

“…We do not include dipolar interactions for associating molecules, because of a model shortcoming; namely, dipolar and associative Helmholtz energy contributions are treated as independent perturbations, although the perturbation approaches should actually be developed in a defined sequence. Even though including both association and dipolar interactions can lead to better predictions, e.g., for water, 64 for the majority of substances considered in this study we decided to take a conservative approach by treating all associating fluids as nonpolar.…”

Confidence-Interval and Uncertainty-Propagation Analysis of SAFT-type Equations of State

“…In the literature, several strategies have been suggested to try to mitigate the problem of parameter degeneracy. When several local minima are present, instead of local solvers, solvers that search a larger parameter space have been used. ,, To explore a larger area of the parameter space and obtain a compromise between several properties, multiobjective optimization ,,, has also been applied. Others have tried exploring the impact of adding more pure-component thermodynamic data such as vaporization enthalpies, heat capacities, speeds of sound, and surface tensions to improve the quality of the fit. ,− Another approach used also includes adding mixture data to correlate the pure-component parameters and discriminate between model parameter sets leading to models that predict pure-component and mixture phase equilibria with high accuracy. ,, For more information on what has been tried in the past to combat parameter degeneracy, we invite the reader to examine the analysis by Cripwell et al…”