Development of Molecular Models of Real Fluids for Applications in Process-Engineering

A major challenge in thermodynamics is the quantitative description and prediction of thermophysical properties for industrial applications in energy and process engineering. Because comprehensive experimental measurements are often costly and time-consuming, robust theoretical methods are essential. Molecular modeling and simulation addresses this need by describing intermolecular interactions with molecular models and computing macroscopic properties from these interactions. This approach enables a consistent, physically based description of pure substances and mixtures across wide ranges of state conditions. The present work demonstrates both accurate reproduction of experimental fluid behavior and strong predictive performance for several process-engineering examples. A key limitation for broader use is the limited availability of molecular models tailored for quantitative property prediction. To overcome this, a rapid parametrization strategy is developed that leverages quantum-mechanical ab initio calculations to reduce the number of free parameters. Model site locations and electrostatic parameters are transferred directly from quantum calculations, while only a small remaining set (typically two to four parameters) is optimized to experimental pure-substance vapor–liquid equilibrium data. Using this strategy, 13 new molecular models are constructed and fitted to vapor pressure, saturated liquid density, and enthalpy of vaporization, achieving substantial improvements over literature models. The resulting models provide very good predictions for additional pure-substance properties (e.g., second virial coefficients, structural, thermal, caloric, transport properties, and surface tension) and for mixture behavior. Predictive capability for mixtures is illustrated by accurately capturing the special phase behavior of R227ea + ethanol and by developing a molecular-simulation-based method to determine humid-air dew points that agrees closely with recent experimental data. An outlook is also given toward extending molecular simulation to hydrogels to qualitatively describe solvent-dependent swelling and shrinking as a basis for future quantitative work.