Department of Mathematics
University of Bergen
In-situ combustion (ISC) is the process of injecting air into oil reservoirs to oxidize part of the crude oil. The viscosity of the remaining crude oil is reduced by the significant heat generated from combustion reactions, that contributes to enhanced oil recovery.
ISC is a strongly non-linear multi-physics multi-scale process. The chemical reactions are order of magnitudes faster than other physical subprocesses, such as advection and conduction. In addition, the combustion front is typically very narrow and grid block sizes need to be on the centimeter scale for accurate resolution. At the field level, affordable grid block sizes at the field level are orders of magnitudes larger, however, necessitating an effective upscaling technique.
The chemical reactions in ISC simulations are traditionally modeled via source terms following an Arrhenius kinetic law, and lead to very stiff systems. Solving these systems accurately in time on the field scale is truly computationally demanding. In addition, the Arrhenius reaction models are sensitive to the size of the grid blocks in the simulation.
This sensitivity makes Arrhenius models notoriously difficult to upscale to field scale, since changes in grid size typically requires a recalibration of the kinetic model parameters. Even more importantly, robustness of the predictability decreases as a result of the recalibration process.
We propose a non-Arrhenius kinetic upscaling approach, where Arrhenius reaction terms are replaced with equivalent source terms. The source terms are determined by a workflow integrating both lab experiments and high-fidelity numerical simulations. The experimental-numerical workflow will be presented and we demonstrate how the new formulation alleviates both stiffness and grid dependencies of the traditional Arrhenius approach.