Coarsening Criteria for Reservoir Upscaling

he hydrocarbon reservoirs are characterized by the distribution of their petrophysical properties which are obtained from well data and seismic information. The study of an oil field begins generally with the construction of a geological model with a fine grid. This model respects the architecture and the heterogeneity distribution of the reservoir. To correctly consider all the scales of heterogeneities, the geologists build a model with very small cells compared to the reservoir simulation size. At this detailed scale of the model, the numerical fluid flow simulations may require a prohibitively large computation time and are generally impossible to achieve. To reduce the simulation cost, reservoir engineers generally upscale the geological model to an upscaled mesh. This upscaled mesh has lesser number of gridblocks. The choice of the number of gridblocks is generally the first and critical problem to be solved. In this paper, the authors list different numerical criteria, easy to compute, which can help in evaluating the characteristics of information loss in reservoir upscaling to define the best upscaled model, in terms of precision against number of cells.