Analysis of Upscaling Algorithms in Heterogeneous Reservoirs with Different Recovery Processes

Abstract

Abstract Even though computer power has increased dramatically over the years it is still essential to upscale detailed geological models before reservoir simulation, to decrease simulation time and cost. This paper presents the upscaling of a high resolution geological model constructed for a reservoir in the Anadarko basin. The Dykstra-Parsons coefficient computed indicates a highly heterogeneous and complex reservoir. This work compares different upscaling algorithms for both primary and secondary reservoir depletion processes. Also, the effects of different boundary conditions were tested. Numerous upscaling ratios were studied to ascertain ideal upscaled grid sizes. Prior to property upscaling, quality check techniques which include cell volume, cell angles and cell inside out computations were applied. This ensured consistency with the fine-scale model and reduced simulation errors. Additional quality check techniques were applied to the coarse models after the property scale-up. Simulation results and volume computations from the fine model were used as a benchmark for the study. Several coarse simulation models were developed and compared to the fine-scale model to study the effect of upscaling techniques on the quality of simulation results. The deviation of coarse simulation results from the benchmark were statistically quantified and ranked according to the respective upscaling algorithm. The closest coarse model to the fine scale was used to construct a history match. The results showed clearly that different depletion processes had significant effects on the upscaling techniques employed. Nearly all of the algorithms worked well for primary depletion processes, for example, with less favorable outcomes for the waterflood utilizing the same models. Simulation results were very sensitive to both the type of boundary condition employed and the upscaling ratio. The quality checks ensured good orthogonal grid geometry, which eventually reduced overall simulation errors and saved simulation time. The methodology presented in this paper establishes a road map to follow when upscaling geological models to ensure accuracy in reservoir modeling.