Coarse-scale Modeling of Flow in Gas-injection Processes for Enhanced Oil Recovery

Subsurface formations that arise in the simulation of gas-injection processes for enhanced oil recovery may exhibit geometrically complex features with complicated large-scale connectivity. They must be included in simulations of flow and transport because they can fundamentally impact simulation results. However, to reduce computational costs, simulations are generally performed on grids that are coarse compared to the given geocellular grids, so accurate upscaling is required. In this work we are concerned with transmissibility upscaling. In the presence of full-tensor effects, multi-point flux approximations (MPFA) are desirable for the sake of accuracy, as opposed to two-point flux approximations (TPFA). However, MPFA methods add computational costs and may suffer from non-monotonicity (see [NAE06]). In [LGM06], Variable Compact Multi-Point (VCMP) upscaling was introduced. This method constructs a local MPFA that accommodates full-tensor anisotropy, and guarantees a monotone pressure solution (see [GLM06]). While it has been demonstrated that VCMP performs quite well, compared to other upscaling methods, it does not perform as well for highly channelized domains that are likely to arise in the simulation of gas injection processes. In this paper, we consider some modifications to VCMP in order to improve its accuracy for such cases. The general design of the VCMP methods is summarized in section 2. In Section 3, we present strategies for improving the accuracy and/or efficiency of VCMP. We discuss the results and future directions in section 4.