Accuracy of noise predictions for wind farms (English)

Predictions of environmental noise from wind farms are important component of the planning stage for new facilities to be located near to noise sensitive receivers. The noise limits defined by regulatory authorities will often constrain the layout and number of turbines that can be sited within the wind farm, which will in turn constrain the potential power generation capacity. This paper presents the findings of studies conducted by the authors into the accuracy of noise predictions for wind farms using the ISO 9613-2 methodology. A study into the accuracy of predictions at 13 sites across six wind farms found that: The ISO 9613-2 methodology is unlikely to result in significant under-predictions of wind turbine noise at receiver locations with the use of an appropriate ground absorption factor to adjust for apparent topographical effects. In the case of a concave slope from turbine to receiver a ground absorption factor of 0 should be used for predictions. The current correction factor suggested by the Institute of Acoustics should also be considered where it is warranted by the topography. For flat sites, a ground absorption factor of 0.5 is likely to be more appropriate. Further work is required to reduce the chance of considerable over-predictions at sites with steady downward slopes. Under-prediction does not appear to be a risk at these sites with current prediction methods. A preliminary analysis of noise propagation under different meteorological conditions was also undertaken. It appears that periods of high wind shear and temperature inversion do not result in increases in noise propagation for wind farms, despite their known effects on noise sources located close to the ground. Overall, the accuracy of current wind farm noise models is good, especially considering the uncertainty that exists in the prediction of other noise sources such as road traffic noise and industrial noise. Based on investigations undertaken by the authors, there is not believed to be a risk of significant under-prediction of wind farm noise at receiver locations when good practice noise modelling procedures are employed.