Predicting Cooling Fan Noise of Electric Motor Using Compressible Large Eddy Simulation

Abstract

This paper describes prediction of aeroacoustics from a rotating machinery fan using compressible Large Eddy Simulation (LES). The fan is installed semi-opened space located between the fan cover and the body of rotating machinery such as a electric motor. The fan distributes air from the fan cover intake onto the cooling fins. The Reynolds number of the rotating fan is 9 × 105; its Mach number is approximately 0.1. Under the low Mach number regime, hybrid computational aeroacoustics (hybrid CAA) method, which is solved turbulent flow and acoustics separately, is generally used. However, we used a direct CAA method because interaction between pressure fluctuation from the turbulence and sound propagation should be considered. For the direct CAA method approach, compressible Navier–Stokes equations are solved. Density is calculated from the ideal gas law. To compute turbulence phenomena, LES is used as the turbulence model. The Dynamic Smagorinsky Model is used for the subgrid scale. The sound propagation speed is approximately 10 times faster than the flow speed. Therefore, the numerical schemes, time step, and computational grids size were evaluated with line sound source in the two-dimensional domain as a fundamental study to determine the numerical schemes. Subsequently we evaluated the sound pressure level with the electric motor fan, which is an experimental structure. Through verification of the direct CAA model, we obtained the following results. (1) The predicted pressure fluctuation spectra show good agreement with the experimentally obtained spectra. Specifically, the blade passing frequency (BPF) and trend of the pressure fluctuation decay in the inertial turbulence subrange were predicted. (2) The predicted sound pressure spectra also show good agreement with BPF. Specifically, the acoustic mode and broadband turbulence noise level were predicted.