چكيده به لاتين
Abstract:
Despite the detailed information of geological models of hydrocarbon reservoirs, they are not suitable for dynamic flow simulations. The reason is that as the number of simulation grids increases, the number of of equations increases which results in computational cost. Hence, there is always need for a model in which the dynamic flow simulation is cost effective and is a good representation of detailed geological model. Therefore, the main problem in this approach would be the selection of simulation grids (in this model coarse grid) and their properties in order to show the same results as fine grid models while being computationally cost effective.
In this thesis, utilizing the concept of multi resolution analysis, an algorithm for upscaling reservoirs based on wavelet analysis has been proposed. The basic idea is to have an adaptive multilevel upscaling algorithm in which each coarse grid can have different size in a predefined manner. More over, the effect of the upscaling ratio, different thresholds for wavelet as well as the porosity distribution on the results is considered.
In two phase upscaling, an optimization approach based on fluid flow data and B-Spline method is proposed to estimate the Pseudo-relative permeability curves. The effect of endpoints and convex curves are considered. The results show that the optimized coarse model decreases the CPU time up to 90 percent while being precise enough.
Keywords: Upscaling, Wavelet transform, Genetic algorithm, B-Spline, Simulation
