Measurement of physical properties of refrigerant mixtures. Determination of phase diagrams

Accurate knowledge of high pressure phase equilibrium is of utmost interest in designing chemical processes and separation units operating at high pressures. Moreover, these data are essential to develop optimized models in order to provide continuous representation of high pressure phase equilibrium. The best models are required to permit process simulators to be the most accurate as possible. There are many experimental ways to obtain information about the phase behaviour of fluid mixtures. They can be classified into two main categories: open and closed circuit methods. Closed circuit methods are composed of two main categories: the synthetic methods where compositions are known a priori and analytic methods where samplers are needed. The laboratory has published several data sets concerning refrigerant systems (binaries and ternaries). The more recent are: – binary and ternary systems with 1,1,1,2,3,3,3-heptafluoropropane (R227ea), difluoromethane (R32) (two hydrofluorocarbons) and propane (R290), – binaries of R227ea or R32 with CO 2 (R744) and an old refrigerant SO 2 (R764), – ternary system composed of dimethyl ether (DME), pentafluoroethane (R125) and 1,1,1,2-tetrafluoroethane (R134a), – binary systems involving DME with R134a and R32. In this paper we have focused our attention on two types of models: – the first one is a thermodynamic model implying equations of state and several mixing rules, – the second one is a neural network model. It is not a “thermodynamic” model, but it can be used to calculate thermodynamics properties like enthalpy, entropy from density values. It has been used to represent density data of the R134a + R125 + DME ternary system.