Acoustic Array Measurements on a Full Scale Wind Turbine

Acoustic array measurements were performed on a three-bladed GAMESA G58 wind turbine with a rotor diameter of 58 m and a tower height of 53.5 m. The goal was to characterize the noise sources on this turbine, and to verify whether aerodynamic noise from the blades is dominant. In order to assess the effect of blade roughness, one blade was cleaned, one blade was tripped, and one blade was left untreated. The acoustic array consisted of 152 microphones mounted on a horizontal wooden platform (15 by 18 m 2 ), which was positioned about 58 m upwind from the rotor. In parallel to the acoustic measurements, a number of turbine parameters were monitored, such as wind speed, power, turbine orientation, RPM, and blade pitch angle. In total more than 100 measurements were taken at wind speeds between 6 and 10 m/s. Two array processing methods were used to characterise the noise from the turbine. First, the noise sources in the rotor plane were localised using conventional beamforming. These results clearly show that, besides a minor source at the rotor hub, practically all noise (radiated to the ground) is produced during the downward movement of the blades. The noise is produced by the outer part of the blades (but not by the very tip), and blade noise levels scale with the 5 th power of the local flow speed. The second processing method employed rotating scan planes to localise the noise sources on the individual blades. It turns out that the tripped blade is significantly noisier than the clean and untreated blades, which is a strong indication of trailing edge noise (rather than inflow turbulence noise). The similar noise levels for the clean and untreated blades suggest that the untreated blade was aerodynamically clean.