(450h) Systematic Development Of New Molecular Models To Study Phase Equilibrium Of Systems Containing Hydrogen Fluoride And Refrigerants

Since the discovery of the depletion of the ozone layer and the subsequent banning of chlorofluorocarbons (CFCs) from use as refrigerants there has been great interest in accurately predicting phase equilibrium of replacement refrigerants and their mixtures with hydrogen fluoride (HF). HF is used in the production of replacement refrigerants by reaction with the original CFCs. Robust models for accurate prediction of phase equilibrium are required for process design and simulation. This is of industrial significance since a large proportion of capital and operating costs are related to separation processes, many of which are driven by phase equilibrium. Hydrogen fluoride and refrigerant systems are challenging to model since hydrogen fluoride has a strong tendency to form hydrogen bonds, refrigerant molecules are non-spherical and their mixtures tend to contain azeotropes and exhibit liquid-liquid immiscibility. In addition, there is limited experimental data available and for mixtures the conditions where liquid-liquid immiscibility occurs are often not clear. In this work the Statistical Associating Fluid Theory for potentials of Variable Range (SAFT-VR) is used to overcome these challenges. Pure component models have been developed for hydrogen fluoride and refrigerants by optimising to vapour pressures and saturated liquid densities together with a quantum mechanical approach to determine the non-sphericity of refrigerant molecules. Hydrogen fluoride is modelled as a three site molecule and the refrigerants are modelled as containing sites to account for their dipolar nature. Binary mixture models are used to further discriminate between pure component models. Two methods for developing mixture models have been established depending on whether the composition of the liquid phase in equilibrium or the overall composition of the mixture is known. For the mixtures HF + water and HF + R134a (1,1,1,2-tetrafluoroethane), where the composition of the liquid phase in equilibrium is known, asymmetric association models have been developed. In order to model mixtures where only the overall composition of the mixture is known a model of experimental set-up has been developed so that SAFT-VR mixture parameters can be obtained directly from raw experimental data as opposed to fitted data. The models obtained can be used predictively outside the range of the fit. The strategies developed in this work to obtain molecular models are general and can be applied to other systems.