Selecting an Appropriate Upscaled Reservoir Model Based on Connectivity Analysis

Abstract

Reservoir engineers aim to build reservoir models to investigate fluid flows within hydrocarbon reservoirs. These models consist of three-dimensional grids populated by petrophysical properties. In this paper, we focus on permeability that is known to significantly influence fluid flow. Reservoir models usually encompass a very large number of fine grid blocks to better represent heterogeneities. However, performing fluid flow simulations for such fine models is extensively CPU-time consuming. A common practice consists in converting the fine models into coarse models with less grid blocks: this is the upscaling process. Many upscaling methods have been proposed in the literature that all lead to distinct coarse models. The problem is how to choose the appropriate upscaling method. Various criteria have been established to evaluate the information loss due to upscaling, but none of them investigate connectivity. In this paper, we propose to first perform a connectivity analysis for the fine and candidate coarse models. This makes it possible to identify shortest paths connecting wells. Then, we introduce two indicators to quantify the length and trajectory mismatch between the paths for the fine and the coarse models. The upscaling technique to be recommended is the one that provides the coarse model for which the shortest paths are the closest to the shortest paths determined for the fine model, both in terms of length and trajectory. Last, the potential of this methodology is investigated from two test cases. We show that the two indicators help select suitable upscaling techniques as long as gravity is not a prominent factor that drives fluid flows.