Effects of Grid Block Dimensions and Heterogeneity Scale on Flow Simulation Results

This two fold study addresses the effects of the scale of spatial heterogeneities versus the size of the grid blocks when exploring the uncertainty on the results of oil reservoir flow simulations. In practice, three popular and simple approaches are frequently used to achieve the tradeoff between flow details and speed. Upscaling the grid, upscaling the well data, or directly using the point scale well data on a coarse grid are all methods that are currently in used. We have designed an experiment to compare these different approaches. A fine grid has been defined as the reference case. The fine grid has been coarsened up three times. A series of 5 different variograms were used to simulate different spatial correlations on each grid. Ten geostatistical simulations were produced for each combination of grid, variogram, and upscaling method. It can be shown that upscaling the grid results in a systematic bias. Upscaling the well do not suffer systematic bias but have a tendency to overestimate the uncertainty. Using the point scale data without upscaling leads to larger bias and larger overestimation in the uncertainty.