New upgridding and upscaling method to capture the dynamic performance of the fine scale heterogeneous reservoir

Due to advances in computer technology, geo-scientists are increasingly able to build reservoir models which can incorporate significant geological details. It is not unusual to find geological models which contain millions of cells. Flow simulating such models is still a challenge as running a typical simulation model takes two orders of magnitude more time than generating a geological model. Upgridding the geological model to reduce the number of grid blocks while maintaining the essential heterogeneity details to preserve the dynamic performance is the only solution to this problem.;In the literature, many solutions are proposed to upgrid the geological model. The simple static solutions involve combining layers with similar static properties so that the heterogeneous details are preserved. In this research, we take the method one step further. It has been observed in the past that upgridding should be dependent on the flow process. In this research we focus on single and multiphase processes separately. We suggest different upgridding algorithms for both primary depletion and water injection processes.;For primary depletion analytical procedures are proposed which account for vertical communication across layers. This is the first study which addresses the importance of crossflow on the layering scheme. Through analytical development, we demonstrate the importance of crossflow on homogenization of layers. After finding the optimum layering scheme (upgridding), we use a simple depletion drive process and demonstrate that upscaled properties are not constant. Instead, if the goal is to match the performance of the fine scale model, the upscaled permeability changes with time. We provide an analytical solution to determine the upscaled permeability and present the value of upscaled permeability under limiting conditions.;For two phase flow, we propose a method of upgridding which can be applied for either secondary or tertiary recovery processes. Our method is analytical and is based on the idea that fractional flow of a fine scale model has to be preserved on a coarse scale. We use fractional flow on a fine scale model and define our error as the difference between fractional flow of a coarse scale versus fine scale. We combine layers which minimize this difference. Using our procedure, we demonstrate that for favorable mobility ratios, it is better that we isolate low permeability layers; whereas, for adverse mobility ratios, we need to isolate high permeability layers.;We further extend our analysis to incorporate the effect of crossflow in upgridding. We observe that for favorable mobility ratios and crossflow conditions, we can be much more flexible in combining many fine scale layers without impacting the dynamic performance. In contrast, unfavorable mobility ratio and the absence of crossflow make it difficult to combine the fine scale layers without significantly increasing the dynamic error between fine and coarse scale models.