Discontinuous galerkin for high performance computational fluid dynamics

Abstract

In this report we present selected simulations performed on the HLRS clusters. Our simulation framework is based on the Discontinuous Galerkin method and consists of four different codes, each of which is developed with a distinct focus. All of those codes are written with a special emphasis on (MPI) based high performance computing. In this report, we show results of our newest code, FLEXI, which is tailored solely towards high performance computing of three dimensional advection dominated problems. Currently, we are interested in extending our direct numerical simulation framework to large eddy simulations and present computations and comparisons of our framework for the flow past a cylinder and the flow past a wing. Furthermore, first results on a MPI based multi-physics extension of our framework is presented. All simulations are typically performed on hundreds and thousands of CPU cores.