Bridging the gap between big data and physics for improved prediction of parent-child interaction and their well performances

摘要

Geologic complexity and heterogeneity have been a big hurdle for accurate reservoir characterization and 3D geomodelling, especially in fractured reservoirs. Due to depositional and tectonic factors, rocks exhibit multiple levels of heterogeneity in stratigraphy, structure, and lithology. This scale-dependent nature is more prominent in fractured reservoirs where larger fractures or fault sets bound and spawn many smaller fracture sets. Neural networks have been successfully applied in capturing these variations in reservoir characterization through AI-driven geomodelling tools. The benefits of these Al tools can be extended to the problem of better representing the complex coupling of multi-physics processes underlying the hydraulic fracturing of a child well in the presence of depletion and rock fabric alteration due to a previously stimulated parent well.