This work examines heavy oil viscosity modelling during simulation of steam injection processes, such as steam-line-drive and SAGD, and the sensitivity of oil recovery predictions to the uncertainty in the oil viscosity. Analytical models to predict the sensitivity have been developed, confirmed by numerical simulation. Heavy oil compositional component viscosities are modelled with the Free Volume model. The model is extended in this thesis to estimate the viscosities of long-chain n-alkanes

more »

... om C6H14 to C45H92 within an accuracy of 10% in the temperature range 27 to 300 °C (80 to 575 °F) and pressure range 0.1 to 120 MPa (14.5 to 17,400 psi). It estimates viscosities of long-chain n-alkanes up to C64H130 to within 30%. Extrapolated Free Volume molecular characteristic parameters, optimised based on available viscosity measurements for n-alkanes up to C64H130, are provided, and are the recommended values for use in heavy oil simulation. A heavy pseudo-component, representing a combination of asphaltenes and resins, which are the compounds responsible for the high viscosities observed in heavy oil, is characterised in terms of molecular weight, shape and activation energy for viscous flow. A method to predict its viscosity as a function of its physical properties, pressure and temperature, using the Free Volume model, is demonstrated. A density model based on the Tait model is extended, to predict the long-chain heavy oil compositional component densities within an accuracy of 3%, in the same temperature and pressure ranges as above. A grouping procedure is demonstrated to achieve oil recovery results comparable to a 24-component simulation case, using two pseudo-components. Key is the mixing equation used to calculate the oil phase viscosity as components are grouped. The Arrhenius mixing equation is evaluated for accuracy in predicting hydrocarbon mixture viscosities. Guidelines for accurate use are provided, while mixtures with CO2 are shown to require a different method.