Aristeu da Silveira Neto

This study presents a three-dimensional sharp-interface Ghost-Cell Immersed Boundary Method (GC-IBM) integrated into a parallel Structured Adaptive Mesh Refinement (SAMR) framework for simulating dynamic, thermal, and fluid-structure interaction (FSI) problems. Implemented in the MFSim software, the solver employs Finite Volume discretization with a Fractional Step Method for pressure-velocity coupling. Linear systems are solved via the PETSc library, using GMRES with BoomerAMG preconditioning. For high Reynolds numbers, a Large Eddy Simulation (LES) approach with a dynamic Lilly-Germano subgrid-scale model resolves turbulent structures.The robustness of the GC-IBM is verified through rigorous benchmarks. Planar Poiseuille flow is first employed for fluid dynamic and thermal validation. The coaxial Poiseuille flow verifies the leak-tightness of the scheme, ensuring hydrodynamic decoupling across zero-thickness regions. Thermal-fluid coupling is evaluated in a differentially heated cavity, featuring both Dirichlet and Neumann boundary conditions. Flow around a sphere is investigated across laminar and turbulent regimes, utilizing SAMR to resolve high gradients in the vicinity and wake. This assessment encompasses quantitative drag coefficient evaluations and qualitative wake topology analyses, employing the dynamic LES model for cases at Reynolds number 1000 and 5000. Finally, the GC-IBM is applied to FSI to validate against the Lindholm benchmark, accurately capturing the natural frequencies for the first five vibration modes. These results confirm that the framework provides high numerical fidelity and physical accuracy for a wide range of complex multi-physics problems.