A novel quantitative structure–property relationship model for predicting the maximum volumetric heating capacity and the corresponding working temperature of heat pump refrigerants

• A QSPR model was developed to predict the cyclic properties of refrigerants. • Using COSMO-RS theory to parameterize the molecular structure of the refrigerants. • The maximum volumetric heating capacity and corresponding temperature are predicted. • The model provides new methods for designing refrigerants with better performance. This paper presents a structure–property relationship model between the microstructure and macroscopic cyclic characteristics of heat pump working fluids. The model is established based on the molecular surface charge density distribution function “ σ -profile” proposed by the COSMO-RS theory to describe the molecular structure information. The σ -profile curves were parameterized by Gaussian function fitting, and new molecular descriptors expressing the type and quantities of molecular groups were proposed. These molecular descriptors were used to characterize the thermophysical properties of the working fluids through stepwise multiple linear regression. Based on the molecular structure, a QSPR model expression is established to predict the maximum volumetric heating capacity and the corresponding temperature. The average absolute relative deviation (AARD) of the predictions was 9.35% and 7.34%, respectively. The proposed model provides a way to design a new type of working fluid with better cyclic characteristics.