Investigation of Noise Generation by a Model-Propeller in Turbulent Inflow Through Advanced Signal Processing Techniques

The far-field acoustic emissions of an isolated scale-model propeller ingesting turbulent inflow are experimentally investigated in wind-tunnel measurements. Phase-averaging and phase-shifting signal processing techniques are used to separate the deterministic (tonal) and random (broadband) components of the recorded sound signals. It is reported that a combination of the two techniques prevents tones unrelated to the shaft rotational frequency from contaminating the estimate of the broadband part of the signal. Scaling and directivity analyses of the obtained broadband component highlight the presence of two noise generation regimes depending on the considered frequency range. In particular, a quasi-omnidirectional directivity pattern is observed for frequencies for which the Helmholtz number is much lower than unity when turbulence is ingested. On the other hand, a dipole-like pattern with minima close to the propeller’s rotational plane gradually appears for higher frequencies. A time- and frequency-domain analysis through the Continuous Wavelet Transform (CWT) method shows how the increase in broadband noise is due to a large number of short-duration pulses linked to the ingestion of turbulence.