Aeroacoustic Simulation in the Context of Turbulence

The aerodynamic noise is increasingly recognized as a factor that impairs the humans' health. Therefore, the reduction of the noise has become particularly significant in the engineering field. Within this work, the simulation of the aeroacoustic quantities in turbulent flow will be studied. Hybrid RANS/LES method will be utilized to simulate the flow.

In the LES method, the Navier-Stokes Equation is filtered to reduce the length scale range of the solution. This operation can reduce the computational cost. In LES, the large scale motions are computed explicitly, while the small scale motions are modeled. Therefore, in comparison to DNS, the computational cost is largely reduced. Common sub-grid scale models are Smagoringsky model and Germano model. According to the operation in the near-wall region, the LES method can be classified into LES-NWR (Near Wall Resolution) and LES-NWM (Near Wall Modelling).

LES performs generally better than Reynolds Averaged Navier-Stokes (RANS). However, the computational cost of LES is 10 times to 100 times larger than RANS, since LES requires a finer grid. Therefore, the hybrid LES/RANS concept was proposed to combine LES and RANS so that the LES is only performed where it is needed while RANS is used in regions where it is reliable and efficient.

In order to acquire the acoustic quantities, many different approaches have

been developed. They can be catergorized into direct noise computation and hybrid methods.

The solutions of the Compressible Navier-Stokes equations contain both the flow variables and the acoustic variables. Therefore, the acoustic variables can be obtained by solving the Compressible Navier-Stokes equations directly. This method is called the Direct Noise Computation method (DNC). This method is, however, not attractive for two reasons. On the one hand, the errors of the discretization are not permitted to be greater than theacoustic quantities in any case, which leads to an extreme fine spatial discretization. On the other hand, the acoustics propagates faster than the flow with a low Mach-number, therefore, the time step for DNC should also be sufficiently small. Both lead to a large amount of computation cost.

The hybrid methods separate the solution process into two steps. First,

the acoustic source is obtained by solving the Incompressible Navier-Stokes equations.

Second, the acoustic quantities are calculated using the governing equations for the acoustics. When solving the Incompressible Navier-Stokes equations, there is no need to discretize the spatial domain with extreme small cells and extreme small time steps as in DNC method. When solving the equations for the acoustic quantities, the cells can be even expanded due to the large length scales of acoustic compared with that of the flow. Therefore, the hybrid method consume apparently lower computation cost than the DNC method.

Aeroacoustics, Turbulence, HPC

Template Student

M.Sc./Dipl.-X

Address: | Dolivostraße 15 |

D-64293 Darmstadt | |

Germany | |

Phone: | +49 6151 16 - 24377 |

Fax: | +49 6151 16 - 24404 |

Email: | |