Development and Application of Upscaling Techniques for Modeling Near-Well Flow in Heterogeneous Reservoirs

Abstract

Listen

Translate article

Abstract Upscaling is often applied in reservoir simulation to coarsen highly detailed geological descriptions. Flow in petroleum reservoirs is mainly driven by wells, thus upscaling of near-well flow is important in coarse-scale flow simulation. In this paper, we present an upscaling methodology, which involves a recently developed coarse-scale well model, combined with transmissibility upscaling for inter-well regions. The coarse-scale well model, referred to as near-well arithmetic averaging, directly uses fine-scale permeabilities along well trajectories to compute coarse-scale well index. Compared to flow-based near-well upscaling techniques, this method does not require solving any fine-scale flow problem, but can approximately capture the effects of fine-scale heterogeneity in near-well regions. Therefore it is straightforward for use and does not incur any computational overhead. This model is implemented in a general reservoir modeling framework, which allows for the handling of realistic grid geometry and well completion from real field data. In inter-well regions, we apply transmissibility upscaling, which directly computes the upscaled transmissibility (defined at block interfaces) rather than permeability upscaling. Standard upscaling methods (local permeability and border region permeability upscaling) are also considered to illustrate their potential errors when not being applied appropriately (e.g., without near-well treatment). We apply the proposed upscaling methodology (near-well arithmetic averaging for near-well flow and transmissibility upscaling for inter-well regions) to sector models from real field cases. All the cases involve general multiphase flow simulations, including primary production, water flooding to oil-water and three-phase flow models. It is shown that the proposed methodology consistently improves the accuracy of standard upscaling methods that are commonly used in the industry and for some cases, provides accurate flow predictions, demonstrating its practical applicability for real field cases.