Source Characteristics of Wind Turbines
Sound-generation Mechanisms
Wind turbines generate both mechanical and aerodynamic noise. Mechanical noise is associated with the gearbox and generator and, typically, most of the sound energy is between 20 Hz and 550 Hz [26]. While it is usual for mechanical noise to be less important than aerodynamic noise from modern large horizontal-axis wind turbines, it might be significant, particularly if there is wear or damage. Aerodynamic noise is generated by the complex interactions between turbulent flow and the turbine blades. Often blade self-noise is the main component when А-weighted sound pressure levels are of concern. Near the trailing edge of a blade, the various contributions include boundary layer-trailing edge noise, boundary layer separation-stall noise and noise from tip vortex formation [27]. Generally, the resulting sound spectra are broadband, although tonal components might be produced as well [26]. Usually, with modern blade designs, trailing edge bluntness-vortex shedding noise (which is tonal) is unimportant. Also, the interaction between the turbulent flow approaching the wind turbine and the leading edge of the blade (so-called ‘inflow noise’) will generate sound. Compared to the trailing edge mechanism, this process generates much lower sound frequencies, strongly dependent on the turbulent character of the wind. Potentially, the interaction between the tower and the blades is an additional noise generation mechanism, but, in the main, it is more relevant to downwind rotors and these are much less popular than upwind turbines nowadays.
Typical Spectra of Large Horizontal Axis Wind Turbines
In predicting sound propagation from wind turbines, the ensemble of sounds from the various physical processes is more important than their individual contributions. Many detailed outdoor sound propagation models and engineering models assume point source radiation. Accordingly, wind turbine sound emission is often represented by a single ‘effective’ or ‘apparent’ point source placed at hub height [28]. The source power spectrum can then be estimated by calculating back from a sound pressure level spectrum measured close to this virtual hub height position. In this way, the energy radiated from the noise sources distributed over the wind turbine rotor plane is assumed to be concentrated at the effective point source position.
Statistical analysis of many acoustical source power level measurements at 2MW+ wind turbines [29], following the apparent point source procedure, shows that a typical sound power spectrum can be found after normalizing for the total А-weighted level of each individual turbine (see Figure 3.19).
Typically, the sound power of modern pitch-regulated wind turbines increases with wind speed until a wind speed (measured at a height of 10 m) of roughly 8 m/s is reached, and this usually corresponds to the rated electrical power of the wind turbine [29]. Further analysis has shown that, even for wind turbines with rated powers of 200 kW, a very similar normalized spectrum is found [29]. On the other hand, in the low-frequency range, other work has shown that, with increasing rated power (and, therefore, with
Figure 3.19 An averaged А-weighted effective point source power level spectrum for large horizontal axis wind turbines with a rated power exceeding 2MW, normalized to the total А-weighted source power level. The error bars show the 95 % confidence intervals on the means in each 1/3 octave band [29].
increasing size of the wind turbine), the amplitude at lower frequencies increases somewhat more strongly than at the higher frequencies [30]. Potentially, therefore, the simple shift of the spectrum that has been suggested [29] could lead to inaccuracies if these lower frequencies are of concern.
