For all the advances in computational power of modern computers the problem of upscaling the properties of geological reservoir models is an ongoing challenge for reservoir simulation. Unfortunately, the upscaling of parameters like absolute permeability in single phase flow produces errors due to discarding the subgrid complexities. For tackling this problem, reconstruction techniques have been developed to downscale the local dynamic regions of the reservoir. Nonetheless, the upscaling part of these methods still has a significant effect on the final results since a poor upscaling can lead to errors even if coarse scale solution is downscaled or refined. In this work we focus on application of a range of conventional upscaling methods including pressure-solver, Renormalization, renormalization based modifications and wavelet transmissibility upscaling.

We simulate incompressible flooding scenarios in two and three dimensions. The results are subsequently compared to a modified local-global upscaling method coupled with modified nested-gridding downscaling. CPU runtime, error in breakthrough time and production values are the parameters considered in the comparative analysis.

We show improvements by iteration in providing better coarse scale solutions compared to conventional upscaling methods. Finally, we hope that this study can present a reliable ranking of renormalization and wavelet methods compared to iterative reconstruction method in terms of the trade-off between precision and speed.

You can access this article if you purchase or spend a download.