A Method for Static-based Up-gridding

A method for up-gridding geostatistical reservoir models is proposed. The method generates well-adapted coarse-grid grids for a given realization and target coarse-scale dimensions. While most current grid generation methods are based on either local or global dynamic (flow rate dependent) measures, in this paper, we show that the inclusion of both local and global static information, can greatly improve computational cost while maintaining accurate flow response. Static property information is analyzed to infer a measure of spatially variability in terms of the relative influence of connected bodies on reservoir flow response. These bodies consist of wellconnected permeability falling within a range of permeability values. Bodies may have a strong impact on local and global flow behavior; knowledge of the relative location of such bodies in a given realization may render an accurate coarse-scale representation of the fine scale models in terms of flow response. An investigation of the method under a set of realistic synthetic 2D and 3D reservoirs is conducted. Sensitivity to strength of geological continuity, and to flow boundary conditions are considered for various scale-up ratios. The results indicate accurate reproductions of the flow simulation response when compared to that of the fine scale realizations. A comparative analysis of the method with well-established static methods is also conducted.