Titel: Long-range sound propagation over sea with application to wind turbine noise
Författare: Mathieu Boué
Utgivare: Kungliga Tekniska Högskolan, Vindforsk
Årtal: 2007
Ämnesord/Subject: Ljud
Sökord/Keywords: Atmospheric Absorption, Kalman filtering, Wind and Temperature Gradients, Meteorology, Öland.
Rapport (Pdf)
Sammanfattning/Abstract: The classical theory of spherical wave propagation is not valid at large distances from a sound source due to the influence of wind and temperature gradients that refract, i.e., bend the sound waves. This will in the downwind direction lead to a cylindrical type of wave spreading for large distances ( > 1 km). Cylindrical spreading will give a smaller damping with distance as compared to spherical spreading (3 dB/distance doubling instead of 6 dB). But over areas with soft ground, i.e., grass land, the effect of ground reflections will increase the damping so that, if the effect of atmospheric damping is removed, a behavior close to a free field spherical spreading often is observed.
This is the standard assumption used in most national recommendations for predicting outdoor sound propagation, e.g., noise from wind turbines. Over areas with hard surfaces, e.g., desserts or the sea, the effect of ground damping is small and therefore cylindrical propagation could be expected in the downwind direction. This observation backed by a limited number of measurements is the background for the Swedish recommendation (Swedish Environmental Protection Agency report no. 6241), which suggests that cylindrical wave spreading should be assumed for distances larger than 200 m for sea based wind turbines.
The purpose of this work was to develop measurement procedures for long range sound transmission and to apply this to investigate the occurrence of cylindrical wave spreading in the Baltic sea. This work has been successfully finished and is described in this report. Another ambition was to develop models for long range sound transmission based on the parabolic equation. Here the work is not finished but must be continued in another project. Long term measurements were performed in the Kalmar strait, Sweden, located between the mainland and Öland, during 2005 and 2006. Two different directive sound sources placed on a lighthouse in the middle of the strait produced low frequency tones at 80, 200 and 400 Hz.
At the reception point on Öland, an array of 8 microphones created an acoustical antenna directed towards the sound sources. Wind and temperature data was measured at the source location and during one measurement period (June 2005), wind and temperature profiles were also mapped in the reception area. In order to increase the signal to noise ratio different signal enhancement methods were tested including a Kalman Filter technique and periodic time-averaging. The most accurate results were obtained by combining the Kalman Filter model with a Fast Fourier Transform (FFT). Sound pressure levels as low as a few dB could be detected by using this algorithm.
The final results expressed as a transmission loss (“damping in sound pressure level corrected for the atmospheric damping”) between the source and the receiver, have been compared to simultaneously measured wind and temperature profiles. The transmission loss data have also been expressed as statistical distributions from which e.g. the average value can be obtained. This average, based on data for the summer period June 2005/2006, has been compared with the Swedish Environmental Protection Agency recommendation7. It is found that the breaking point for cylindrical propagation is close to 700 m instead of the 200 m assumed in the recommendation.
This is a significant difference and it shows that probably the Swedish recommendation uses a too small value for the expected breaking point. Of course in general the value of the breaking point can depend on the location and for which part of the year one takes the average. How large the variation can be due to such factors is today still unknown. Here only more measurements and perhaps simulations combined with the wind data base available in Sweden can provide an answer.