Prediction of vapor–liquid equilibrium for binary mixtures containing R1234yf or R1234ze (E)

Abstract

The use of pure low global warming potential (GWP) refrigerants can be replaced with mixture of carbon dioxide or other refrigerants with low GWP refrigerants due to their better features such as higher cycle efficiency. Hence, developing simple and accurate models for estimation of vapor–liquid equilibrium (VLE) of these mixtures is of great importance. In this work, an intelligent model named CSA-LSSVM was implemented for prediction of VLE data in the binary mixture of CO2 or a hydrofluorocarbon (HFC) or hydrocarbon (HC) refrigerant with two low GWP refrigerants (R1234yf and R1234ze (E)). The precision and reliability of the developed model was evaluated by using various approaches. Results show that the implemented model is accurate and effective for prediction of experimental VLE data. To further validate the precision and effectiveness of the implemented model, its predictions were compared with predictions of PC-SAFT and Peng Robinson (PR) equation of states (EoSs). The PR EoS was combined with van der Waals type mixing rules. The outcomes of comparison also show that the developed CSA-LSSVM model provide better predictions with lower errors compared to PC-SAFT model and PR EoS.