Genetic Programming (GP)-Based Model for the Viscosity of Pure and Hydrocarbon Gas Mixtures

Abstract

Accurate determination of the viscosity and phase behavior of pure hydrocarbon gases and hydrocarbon gas mixtures is essential for reliable reservoir characterization and simulation and, hence, for optimum usage and exploitation. The variety of possible hydrocarbon gas mixtures at different conditions of interest preclude obtaining the relevant data by experimental means alone; thus, the development of prediction methods is required. Many pure hydrocarbon gas viscosity correlations are available. However, a wide-ranging and accurate viscosity correlation of gas mixtures associated with heavier hydrocarbon components and impurity components, such as carbon dioxide, nitrogen, helium, and hydrogen sulphide, is not available. Therefore, this paper presents a new pure hydrocarbon gas and gas mixture viscosity model over a wide range of temperatures and pressures as a function of gas density, pseudo-reduced temperature, pseudo-reduced pressure, and the molecular weight of pure and hydrocarbon gas mixtures. The new model is designed to be simpler and eliminate the numerous computations involved in any equation of state (EOS) calculation. The developed new model yields a more accurate prediction of the pure gas and gas mixture viscosity compared to the commonly used correlations.