Generation and Adaptive Modification of Anisotropic Meshes

The ability to quickly and reliably simulate high-speed flows over a wide range of geometrically complex configurations is critical to many of NASA's missions. Advances in CFD methods and parallel computing have provided NASA the core flow solvers to perform these simulations. However, the ease of use of these flow solvers and the reliability of the results obtained are a strong function of the technologies used to discretize the domain. Many applications involve solutions with highly anisotropic features: boundary layers, shear layers, wakes, shocks etc. Efficient resolution of those features motivates matching the mesh resolution/anisotropy to the solution's anisotropy but, in the more challenging applications, the location and strength of those features is difficult to precisely estimate prior to solution. Currently available meshing tools are not capable of producing and controlling the required initial meshes, nor adapting the mesh to match evolving anisotropic features. This project will combine Simmetrix Inc. expertise in the development of meshing components for flow simulations, and Rensselaer's Scientific Computation Research Center expertise in the development of adaptive mesh control technologies, to provide NASA the mesh generation and adaptation technologies needed. New techniques will be developed to create highly anisotropic semi-structured and unstructured meshes suitable for CFD simulations with high Reynolds number flow features (e.g., boundary layers, bow shocks, free shear layers, wakes, contact surfaces). Techniques to adapt these meshes based on mesh correction indicators will be developed to enable fully automated adaptive simulations. All procedures to be developed will work effectively in parallel on large-scale parallel computers and will support a wide range of flow solvers. The overall capabilities will be demonstrated through execution of fully automated parallel adaptive simulations on problems relevant to NASA.