Assessment and development of the viscosity prediction capabilities of entropy scaling method coupled with a modified binary interaction parameter estimation model for refrigerant blends

Low global warming potential (GWP) refrigerant blends have been identified that may play a role according to the gradually reduction of the baseline amount of hydrofluorocarbons (HFCs) that can be replaced on the future market. Entropy scaling theory is a potential way to describing viscosity behavior of refrigerant mixtures, which is crucial for designing and optimizing of refrigeration system. However, an important prerequisite to apply entropy scaling model to mixture is reliable binary interaction parameters (k(ij)). To promote intensive industrialization of entropy scaling theory and avoid expensive and timeconsuming experiments for k(ij) regression simultaneously, a modified Quantitative Structure Property Relationship (QSPR) theory is proposed in this work to provide reasonable estimated kij values with an additional dispersion term. To evaluate the performance of the proposed theory systematically, five different estimation methods are considered when predicting phase behavior and viscosity. Such model shows better performances: for the non-associating mixtures including HFCs, hydrofluoroolefins (HFOs), R744 and hydrocarbons (HCS), vapor liquid equilibrium data are predicted with the deviation of 1.47%, liquid mixture viscosities are exactly reproduced with an error of 3.49%. Besides that, the modified QSPR associated with the PC-SAFT equation of state (EoS) has superior performance of k(ij) estimation than the London's dispersive theory and the original QSPR, especially in describing some highly nonideal systems. As a supplement, the residual entropy scaling model for viscosity by Yang et al. and the commonly used extended corresponding states model are also applied to verify the modified method. (c) 2022 Elsevier B.V. All rights reserved.