In order to analyze aeroacoustic noise generation processes, researchers from the Institute for Aerodynamics and Gas Dynamics in Stuttgart have performed high fidelity, large-scale flow and acoustic computations using the discontinuous Galerkin spectral element method (DGSEM) on the HPC system Cray XC40 Hornet at the High Performance Computing Center Stuttgart (HLRS). The aim of this investigation is to gain insight into the tonal noise generation process of a side-view mirror.
• 3288 compute cores
• 10 TB data
• 44 million degrees of freedom per variable
• 12.5 million time steps
The reduction of flow-induced sound is increasingly important in many engineering fields such as aviation and the wind energy sector. Acoustic emissions today also play a major role in the development of vehicles with regard to passengers as well as the surrounding environment. Beside the engine and tires, sound generated by the flow around the car is among the most important noise sources. With a growing number of electric vehicles, the significance of aeroacoustics in that industry is expected to rise even further.
Despite advanced experimental methods and fast developing simulation techniques, some of the underlying physical mechanisms are not understood completely. Deeper knowledge of the principle effects can contribute to a more efficient design process and to the development of next-generation quiet vehicles itself.
The aerodynamic simulation technology used in industry today is not sufficient for investigations of such phenomena. Intrinsic, physical details of the flow and the acoustic field and the coupling between both have to be captured, which requires an unsteady and highly accurate representation and thereby a massively parallel simulation approach.
In this project, high fidelity compressible large eddy simulations using the discontinuous Galerkin spectral element method were conducted on the Cray XC40 Hornet at HLRS to shed light on the generation of tonal noise on a side-view mirror. Tonal noise is characterized by narrowband peaks in the frequency spectrum. It is perceived as a whistling sound and can appear particularly annoying even at low sound pressure levels.
The simulation setup is chosen to match wind tunnel experiments, excellent agreement with various measured hydrodynamic statistic quantities validates the simulation. Tonal noise generation is detected at the same positions as in the experiments.
Figure 1 shows a visualization of the turbulent vortex structures behind the mirror placed in a flow at 100km/h. The interaction of the highly organized flow structures shown in Figure 2 with the trailing edge in a region on the inner side of the mirror is found to be the principle tonal noise source. The acoustic field visualization of the tonal component in Figure 3 proves the location of the acoustic source. A detailed analysis of the simulation data reveals the underlying physical mechanism. The emitted sound is tonal due to a resonance effect, in which upstream running acoustic waves excite instabilities in the flow close to the mirror surface. These instabilities, as they are convected towards the trailing edge evolve into the organized vortex structures seen in Figure 2 on the side of the mirror.
Beside the identification of the tonal noise generation mechanism which enables tailored design improvements that prevent this effect, the investigation reveals that the feedback of acoustic waves on the flow field can be important. This mechanism thus should be included in engineering prediction tools to predict tonal noise generation effects like the one at hand.
Hannes Frank, firstname.lastname@example.org
Claus-Dieter Munz, email@example.com