Equation of State Coupled Predictive Viscosity Model for Bitumen Solvent-Thermal Recovery

Abstract

Abstract Exponential and polynomial viscosity correlations have been widely applied to model crude viscosities with temperature. These simple correlations are difficult to be applied to predict the viscosity of different solvent-diluted bitumen systems over a wide range of solvent composition. The Expanded Fluid viscosity model consisting of density as an input parameter can be coupled with an Equation of State in a compositional and thermal reservoir simulator. However, the accurate prediction of density using an EoS is the prerequisite to apply this viscosity model. In this work, the expanded fluid theory was coupled with the simplified PC-SAFT EoS (Perturbed-Chain Statistical Associating Fluid Theory) to predict and correlate the rheology behaviour of bitumen/solvent systems. Athabasca and Peace River Bitumen was characterized using a proposed eight-pseudocomponent characterization method for PC-SAFT, which simply required distillation and molar mass data. The obtained density was then input into the viscosity theory to model viscosity. Viscosity predictions were obtained using zero viscosity binary interaction coefficients, whereas pressure-dependent and temperature-dependent viscosity binary interaction coefficients were adjusted to improve the effectiveness of mixing rules. In the case of Athabasca Bitumen with CH4, C2H6, and CO2, the correlated solubility and density Average Absolute Relative Deviations (AARDs) were within 6.6 % and 2.3 %, respectively. Viscosity AARDs by prediction were within 55.4 %; whereas the AARDs were reduced within 13.5 % using pressure-dependent viscosity binary interaction coefficients. In the case of Peace River Bitumen with C2H6, C3H8, n-C4H10, n-C5H12, the predicted density AARDs were within 0.7 %. Viscosity AARDs obtained by prediction were within 24.9 %, and they were reduced within 8.4 % once using temperature-dependent viscosity binary interaction coefficients.