Analysis of Parameters for the Acceleration of UnsteadySimulations with the DLR-Tau Code

Abstract

In this CFD study, the variation of the parameters of the dual time-stepping method implemented in the DLR-Tau Code for the computation of unsteady simulations are investigated. Two dimensional and three dimensional cases were calculated to nd the best combination of parameters for the acceleration of the simulations in terms of calculation time. With the 2-D Cases (Pitching oscillation of a NACA 0012 airfoil and laminar ow over a circular cylinder), it could be proven, that by using a variation of the time steps from a quick parameter setting to an accurate setup, a reduction of almost 50% in the calculation time is achievable without observing a major impact on the accuracy of the relevant force coe cients versus the use of a constant time step with a considerable number of inner iterations. Furthermore, advantages of the initialization of the ow through a steady-state simulation can be observed especially for the circular cylinder case. The experiences gained from the 2-D cases were applied to the 3-D cases (Pitching oscillation of a transport aircraft con guration DLR-F12 and a rotating installed AGARD propeller). The results obtained for the 3-D Cases are in agreement with those of the 2-D Cases, as a reduction of the calculation time by almost 50% could be achieved. Analogous to the cylinder case, the installed propeller case shows the advantages of the initialization of the ow with a steady-state simulation, which leads to a notable acceleration of the convergence.