Principal Investigator: Claus-Dieter Munz, HPC Platform: Hazel Hen of HLRS
The reduction of aeroacoustic noise emissions from technical systems like wind turbines or airplanes is a today’s key research goal. Understanding the intricate interactions between noise generation and flow field requires the full resolution of all flow features in time and space. Thus, very highly resolved unsteady simulations producing large amounts of data are required to get insight into local noise generation mechanisms and to investigate ideas for reduction. These mechanisms have been investigated by researchers at the University of Stuttgart through direct numerical simulation (DNS) of a generic airfoil configuration on Hazel Hen at HLRS. The results help to understand noise generation from turbulent flows and also serve as a benchmark for future studies.
The acoustic footprint has become an increasingly important criterion for the economic success for technical systems. The noise sources may even determine the range of application. Typical examples are jet engine noise or wind turbine noise: Jet engine noise influences the social acceptance of air traffic in general, and wind turbine acoustic emissions limit the places of their locations. In both examples, the main source of noise is the air flow passing along and through an object. Thus, understanding the generating mechanisms of acoustic emission from fluid flow, i.e. aeroacoustics, is an important step towards designing more environmentally friendly systems.
Numerical simulations provide an indispensable tool in aeroacoustic research across the entire spectrum from basic research to product design. It can assist in discovering noise generation mechanisms at a detailed level, e.g. acoustic feedback, and can give noise level estimates during the design process.
In the current project, researchers of the Institute of Aerodynamics and Gas Dynamics at the University of Stuttgart investigate the generation of aeroacoustic noise on an airfoil configuration through so-called Direct Numerical Simulation (DNS). This numerical technique makes no a priori modeling assumption and resolves the full flow and acoustic fields with all the corresponding interactions. The multiscale nature of these problems requires an unsteady approach with highly accurate and efficient numerical schemes in a massively parallel simulation approach. The scientists used the FLEXI framework based on a Discontinuous Galerkin spectral element discretization of the spatial domain around a NACA 0012 airfoil. The high order polynomial approximation on a mesh with 71,500 elements leads to a total number of 200 million degrees of freedom. The Reynolds number based on the chord length was 100,000 at 2 degrees angle of attack and a Mach number of 0.3. All computations were run on the Cray HPC System Hazel Hen on 20,000 cores.
Figure 1 shows the instantaneous vortical structures of the flow, colored by velocity magnitude. Downstream of a boundary layer trip, transition to turbulence occurs instantaneously on the upper side, while the flow re-laminarizes on the lower side due to the positive angle of attack. At the trailing edge, the laminar rollers and the detached turbulent boundary layer interact, leading to an acoustic spectrum with dominant tonal peaks and broad band noise.
Figure 2 shows the density contours on a slice. Here, the acoustic waves propagating into the far field can clearly be seen. These results will be analyzed further to gain an understanding of the noise generation mechanism and serve as well as a benchmark for Large Eddy Simulations, in which the smallest flow structures are modelled to reduce the computational effort.
Prof. Dr. Claus-Dieter Munz Institute of Aerodynamics and Gas Dynamics University of Stuttgart Pfaffenwaldring 21, D-70569 Stuttgart (Germany) E-mail: email@example.com