Adaptive Mesh Refinement for Spectral-Element Simulations of Compressible Turbulent Flows
Please login to view abstract download link
In the current landscape emphasizing emissions and energy efficiency in transportation, aerodynamics significantly shapes vehicle dynamics. Addressing and minimizing aerodynamic forces require reliable tools, with Computational Fluid Dynamics (CFD) playing a crucial role, especially when accelerated by Graphics Processing Units (GPUs). To handle complex flows, a combined approach of Spectral Finite Elements, operator splittings, and the Entropy Viscosity stabilization model is employed in the GPU-accelerated SOD2D code. Tackling turbulent, multi-scale flows demands meticulous discretization. Utilizing full-hexahedral meshes for Spectral Element Method (SEM) performance, spatial adaptation occurs based on problem solutions where an octree-based structure adapts the coarse mesh, introducing non-conformity across element faces. Preliminary results showcase minimal mass and energy loss in the Taylor-Green vortex case, particularly with increasing polynomial orders. Ongoing research explores more intricate scenarios like High Lift Prediction and AutoCFD geometries, offering a comprehensive evaluation of the method's effectiveness.
