The parallel mesh deformation of the DLR TAU-code

Abstract

Mesh deformation for unstructured grids is usually solved employing algorithms based on a spring analogy or on a linear elasticity analogy. These methods require solving discrete equations for the given grid using iterative methods. The computational effort for these methods is not negligible compared to the effort required by the flow solver. In order to optimise the efficiency of the computational chain for coupled simulations of fluid structure interaction an alternative algebraic method was developed for the DLR TAU-Code. The TAU-Code is parallelized by domain decomposition using the message passing concept based on MPI such that each process works on one partition of the grid only. In order to employ the grid deformation on the grid partitions distributed over the processes, a parallel method is required. This paper describes the parallel grid deformation method and shows its efficiency and robustness for the DLR F6 geometry. Further applications are shown in order to demonstrate the grid deformation for dynamic aeroelastic simulations. © 2007 Springer-Verlag Berlin Heidelberg.