Speaker
Description
We developed a Multiscale Embedded Discrete Karst Method (MsEDKM), a three-scale framework designed to quantify complex mass-exchange mechanisms between karst conduits and the surrounding porous matrix in carbonate formations. At the microscale, fluid flow is modeled in a porous matrix intersected by karst conduits with highly irregular geometry, characterized by shell waviness and pronounced local variations in cross-sectional area. This microscopic formulation constitutes the high-fidelity reference model for subsequent upscaling steps. The micro-meso upscaling stage consists in homogenizing the complex conduit geometry within each coarse cell, replacing it by an Equivalent Elliptical Cylinder (EEC) derived by a moment-of-inertia-based strategy. At the meso-macro level, conduit–matrix exchange is quantified through a flow-based upscaling procedure. High-fidelity finite-element simulations of transient diffusion problems are performed on representative mesoscopic matrix/conduit configurations to compute non-neighboring exchange transmissibilities under quasi-stationary conditions. The resulting dataset is then employed to train a surrogate model that predicts transmissibilities from a reduced set of geometric and hydraulic descriptors at significantly reduced computational cost. At the macroscopic scale, the predicted transmissibilities are incorporated as non-neighboring connections (NNC) into finite-volume reservoir simulations, providing an accurate embedded representation of conduit–matrix interactions. Geomechanical effects are encompassed through a formulation under small-deformation assumptions, whereby mechanical fields update hydraulic and geometric properties, leading to stress-dependent exchange transmissibilities. The proposed framework combines geometric homogenization, data-driven surrogate modeling trainned via flow-based numerical upscaling, and hydro-mechanical effects into a unified approach. To reduce computational cost, we rely on analytical computation for geomechanics, based on edometric stress path. Numerical results demonstrate the physical consistency of the methodology supporting its applicability to simulations of karstified carbonate reservoirs.
| Country | Brazil |
|---|---|
| Acceptance of the Terms & Conditions | Click here to agree |
