Abstract:
Fluid-structure interaction (FSI) problems involving partitioned solvers with non conforming grids present formidable challenges, which are evident, particularly in achieving e cient information exchange between solvers and accurately tracking the uidstructure interface over time and space. These challenges demands special considerations for reliable results. This thesis addresses these issues by introducing a novel approach that leverages arti cial neural networks (ANN) to enhance interface treatment within the framework of classical computational approaches, namely, nite di erence method (FDM), nite element method (FEM), meshfree particle method (MPM) and nite volume method (FVM). This study highlights that MPMs, inherently wellsuited for large deformation problems, e ectively handle moving boundaries and complex geometries, making them ideal for FSI applications where traditional mesh-based methods face limitations. The research primarily focuses on partitioned FSI, exploring innovative formulation strategies, domain discretization techniques, and coupling methodologies to enhance interface treatment.
