Automatic simulation of soil-rock mixture seepage using image model directly

Abstract

Meso-structures in soil-rock mixtures (SRM) exhibit complex geometries that significantly impact their seepage characteristics. The simulation of these structures through conventional methods, such as the Finite Element Method (FEM), necessitates the development of CAD (computer-aided design) models and the generation of conforming meshes. This process is often labor-intensive and can be cumbersome.

In this work, a seepage analysis that employs image models and structured grids, thus owning high automatic level, is first developed for SRM. Compared to traditional boundary (2D) or surface models (3D), image models can sincerely and efficiently capture the meso-structure of SRM. Using structured grids significantly alleviates the preprocessing workload associated with generating conforming meshes. The element size of the structured grid is completely independent of the pixel size of the image model, facilitating a flexible trade-off between computational accuracy and time efficiency. Weak discontinuities in the seepage field at the soil-rock interfaces are accurately represented using continuous-discontinuous (C-D) trial functions using the Numerical Manifold Method (NMM), enforced through penalty methods. Numerical examples investigate the computational accuracy and efficacy of the proposed method. More promisingly, the direct use of image models and structured grids in simulating SRM greatly reduces manual intervention, demonstrating the new method’s outstanding automation advantage and providing a methodological foundation for the development of related unmanned hardware devices.