Speaker
Description
Fractured porous media play a critical role in geologic energy storage, influencing the behavior of subsurface systems. However, explicitly modeling thousands of fractures remains computationally challenging due to the need for a detailed representation of fracture networks. Traditional approaches such as finite difference (FD), finite volume (FVM), or finite element methods (FEM) encounter significant difficulties in meshing, often requiring super fine meshes around fracture intersections or yielding poor mesh quality. Embedding techniques offer a potential alternative by incorporating fractures into coarse grids. Still, they introduce complexities such as increased computational costs, non-physical artifacts, or difficulty capturing complex fracture-matrix interactions.
Upscaling techniques, which approximate the impact of fine-scale fractures on a coarser scale, provide another option, utilizing methods like homogenization, dual-porosity models, or effective medium theories. However, these methods can be limited by assumptions that reduce their accuracy or applicability across diverse geologic conditions.
To address these challenges, we propose a novel approach leveraging equation discovery to uncover upscaling equations directly from data. Unlike black-box models, which hinder generalizability and interoperability, our framework focuses on interpretable and adaptable solutions. This paradigm enables scalable, physics-consistent modeling of fractured porous media while maintaining computational efficiency and broad applicability.
| Country | United States |
|---|---|
| Acceptance of the Terms & Conditions | Click here to agree |
