LAUSR

Abstract Recently, Rokni et al. [1,2] developed entropy-scaling based pseudo-component techniques to predict the viscosity and thermal conductivity of hydrocarbon mixtures and fuels up to high temperature and pressure conditions using only two calculated or measured mixture properties (number average molecular weight and hydrogen-to-carbon ratio). The models are accurate for many hydrocarbon mixtures that do not contain branched compounds (7 and 2% mean absolute percent deviation (MAPD) for viscosity and thermal conductivity, respectively, on average). However, predictions for some hydrocarbon mixtures and fuels containing iso-alkanes are often less accurate (16 and 19% MAPD for viscosity and thermal conductivity, respectively, on average). To improve predictions, it was proposed by Rokni et al. [1,2] to fit one model parameter using an experimental reference viscosity or thermal conductivity data point, which may not be ideal if experimental reference data are not available. In order to make these models more practical, this study fits empirical correlations for the model parameters, so that accurate predictions can be made without fitting model parameters. The correlations enable viscosity and thermal conductivity predictions for a wide range of hydrocarbon mixtures and fuels, including those containing branched alkanes, and no longer require input of any experimental reference viscosity or thermal conductivity data. The correlations are temperature (fit to data from 288 to 550 K) and pressure (fit to data from 1 to 4,400 bar) dependent and are functions of average molecular weight, hydrogen-to-carbon ratio, iso-alkane and two-ring saturate concentrations. Viscosity and thermal conductivity predictions were found to improve to within 5 and 2% average MAPD, respectively, relative to experimental data for the hydrocarbon mixtures and fuels considered in this study.