Thermal Conductivity Modeling of Refrigerant Mixtures in a Three-Parameter Corresponding States Format

Abstract

A predictive model for the thermal conductivity (TC), in a corresponding states (CS) format, is proposed here for mixtures of homologous fluids such as the halogenated alkanes (HA) and the alkanes (A), most of which are used as refrigerants. The predictive nature of the model originates from a new study carried out for the TC of pure fluids. For the dilute-gas term the model requires an individual correlation for each component, whereas for the excess contribution the model structure makes use of TC dedicated equations (TCDEs) of two reference fluids, which in this work are methane and R134a. The mixture model adopts specific mixing rules for each of the two TC contribution terms: the dilute-gas term λ 0mix is obtained from the Mason and Saxena mixture model, while the excess term Δ E λmix is determined from the Wong et al. mixing rules in the one fluid model approach. Setting the mixing rules interaction coefficients to unity, the resulting model presents a completely predictive character. The model has been tested on both liquid and vapor phases of the following systems: R32/R125, R32/R134a, R125/R134a, R404a, and R32/R134/R125. For a total of 1223 experimental points in the liquid phase, the overall AAD is 5.39%, while for a total of 2358 points in the vapor phase, the AAD is 2.62%. These predictive mode performances can then be regarded as particularly satisfactory and are of a level similar to the claimed experimental uncertainty. An improved version of the model is also proposed for modeling azeotropic mixtures. The results reached in this case for a total of 1989 experimental points give an average AAD of 5.30%. Considering both the predictive nature and the simple computational procedure of the model, it significantly enhances the calculations of the TC of mixtures.