Wind turbine noise often is quantified in terms of time averaged overall sound power levels, whilst annoyance due to noise level fluctuations in mid- to high-range frequencies (‘swish’) are not taken into account. Recent experimental research on wind turbine noise has revealed the major causes of the swishing noise to be due to the directivity of the noise sources and convective amplification effects of the moving turbine blades. The findings have been incorporated in the noise prediction tool SILANT which in addition to sound power levels gives sound pressure level predictions for specified observer positions.
The noise sources that are taken into account are trailing edge, inflow and tip noise, using the models of Brooks, Pope and Marcolini (BPM) and Amiet and Lowson. The blade is divided into a number of independent elements for which effective inflow velocity and angle of attack information is a necessary input. A distinction is made between the various profiles along the blade span by including their boundary layer displacement thicknesses at the trailing edge in a profile database.
The propagation model includes directivity, convective amplification, Doppler shift and atmospheric absorption. The effect of the retarded time is taken into account individually for the separate elements along the blade span using the time dependent rotor azimuth position. A simple empirical model is applied to quantify meteorological effects influencing refraction and ground effects.
Prediction results are compared to SIROCCO project measurements from microphones positioned in a circle around a turbine. The high spatial and temporal resolution of the SILANT simulations gives new insights in the variation of wind turbine inflow and trailing edge noise as a function of observer position, rotor azimuth angle and frequency band. The influence of directivity is illustrated for the dominant noise sources.