Directional Multigrid using Anisotropic Mesh Adaptation for CFD (AMADeo)

Lead Research Organisation: Queen Mary University of London
Department Name: School of Engineering & Materials Scienc


The size of Computational Fluid Dynamics (CFD) calculations of interest to industry is mostly limited by the turnaround time to obtain a converged solution. Industry would like to use meshes of around 10 Million cells which currently involves turnaround times of the order of weeks. A turnaround time of at most a day is required for a computational tool in the industrial design cycle. The aim of this project is to reduce the turnaround time on hybrid grids for aeronautical design applications by around an order of magnitude by focusing on two key aspects.Firstly, solution adaptive mesh refinement reduces the overall mesh size significantly and therefore reduces turnaround time. When applied to highly anisotropic flows such as boundary layers it Is crucial for efficiency that cells can be refined directionally, not isotropically as currently practiced. Previous research has shown the viability and efficiency of-the directional' adaptation in 2D: The proposed project is aimed at extending the application to realistic industrial 3-D configurations.The second key aspect is generating a hierarchy of grids for multigrid methods which have established themselves as the method of choice to converge the solution. It is often difficult to derive a sequence of coarser meshes for hybrid grids. A key novelty of the proposed project is to use the sequence of refined meshes by the adaptation algorithm. Further reduction of the turnaround time will be provided by a novel combination of the mesh refinement with geometric coarsening methods.


10 25 50