Improved upscaling of reservoir flow using combination of dual mesh method and vorticity-based gridding

Abstract

International audienceA novel technique for upscaling of detailed geological reservoir descriptions is presented. The technique aims at reducing both numerical dispersion and homogenization error generated due to incorporating a coarse computational grid and assigning effective permeability to coarse-grid blocks, respectively. In particular, we consider implicit-pressure explicit-saturation scheme where homogenization error impacts the accuracy of the coarse-grid solution of the pressure equation. To reduce the homogenization error, we employ the new vorticity-based gridding that generates a non-uniform coarse grid with high resolution at high vorticity zones. In addition, to control numerical dispersion, we use dual mesh method (DMM) which uses different grids to solve pressure and saturation equations. The coarse grid generated from vorticity is used for computation of pressure, and the reference fine grid is used for updating saturation explicitly. The strongest point of the method is that dual mesh method has been incorporated onto non-uniform grid structure. This combination removes the need to solve the full fine grid for two-phase flow modeling, resulting in a less computationally demanding and more accurate upscaling technique. To evaluate the method, we run two-phase simulation using different 2D test cases. Our results indicate that the non-uniform DMM is more accurate than the uniform DMM due to using vorticity-based grids. The speedup achieved in the computation is significant depending on the complexity of model and degree of upscaling

A Quantitative and Qualitative Comparison of Coarse Grid Generation Techniques for Numerical Simulation of Flow in Heterogeneous Porous Media

“…Therefore, numerical dispersion issue in VBG should be treated by a separate procedure in principle (e.g. pseuoization or use of higher order numerical schemes or dual mesh method (for instance see Mahani et al 41 and Ashjari et al 42 ), or if this proves to be costly then by adjustment of upscaling level such that a compromise between numerical dispersion and coarsening level is found. In this paper, we avoided numerical dispersion by simply refining the coarse grid to fine grid resolution.…”

Effect of connectivity misrepresentation on accuracy of upscaled models in oil recovery by CO2 injection

“…The non‐uniform coarsening creates smaller grid blocks inside the regions whose pressures or saturations rapidly change. These models may be obtained based on minimizing heterogeneity over fine grids, vorticity of velocity field, or a wavelet transform of a permeability map . The renormalization method estimates the effective permeability by analogy with the concept of equivalent resistor .…”