Analysis of Adaptive Grid Refinement in Field-scale Simulations of Hybrid Heavy Oil Recovery

Abstract

Hybrid steam-solvent or solvent-based thermal recovery methods for heavy oils become interesting and important, and have recently been field-tested. The intrinsic advantages of this technology include oil production acceleration, higher ultimate recovery factor, reduction of energy and water treatment expenses, better control of GHG emission etc. As a rule the complex production mechanism related to highly improved initial oil mobility occurs during hybrid process. It comprises gravity drainage and locally viscous-dominating flow, phase transitions, diffusion/dispersion liquid component transport, heat conduction and advection within the multicomponent thermodynamics framework. As almost all these phenomena take place in narrow zone close to the gas chamber (GC) contact with initial (cold) oil, it seems evident that an adequate spatial representation of complex physics required special efforts. Typically the reservoir simulation models are built on numerical grid with cell sizes making impossible accurate description of heat and mass transport at the edge of the GC. Along with this outside this zone this typical grid can be acceptable. The adaptive grid refinement (AGR) may offer a compromise solution if the numerical model accepts the necessary degree of refinement. In our current work, we present a methodology of field-scale numerical simulations for solvent-assisted thermal recovery of heavy oil using AGR. Comparison is performed between oil production mechanisms interaction for typical Athabasca bitumen and for different discretization size. In many cases a choice of trigger variable was critical. The numerical performance including the CPU time analysis is presented in some detail.