Parallel and Adaptive 3D Flow Solution Using Unstructured Grids

Abstract

This chapter develops a parallel adaptive Euler flow solution algorithm for 3D applications on distributed memory computers. It is aimed at developing and implementation of a parallel grid adaptation scheme together with an explicit cell-vertex based finite volume 3D flow solver on unstructured tetrahedral grids. The focus is on bringing a more practical approach to parallel grid adaptive solutions with an affordable cost. A good way to achieve this is to separate all individual tools required for parallel adaptive flow solution and then to couple them in the adaptation cycle. In such an approach, four individual tools are needed; flow solver, grid adapter, grid generator, and grid partitioner. The adaptive flow solutions have better agreement with experiments even better than that of uniformly distributed fine meshes around high gradient regions. One of the important aspects of the grid adaptivity is to choose adequate flow features for the gradient calculation. The gradient of pressure seems to be good enough for very localized mesh refinements in the Euler flow analysis. It is proved that parallel adaptation is very effective for obtaining accurate and efficient flow solutions.