Wtn09 Andersson

9
Long distance sound propagation over a sea surface Page 1 of 9 Third International Meeting on Wind Turbine Noise Aalborg Denmark 17 – 19 June 2009 Long distance sound propagation over a sea surface B.L. Andersson 1 , K. Bolin 2 , A. Cederholm 1 & I. Karasalo 1,2 Addresses: 1) FOI - Swedish Defence Research Agency, Stockholm, Sweden 2) KTH - Royal Institute of Technology, Stockholm, Sweden e-mails: [email protected] , [email protected] , [email protected] , Abstract [email protected] Results from measurements of sound propagation over a sea surface to a 10 km distant receiver are compared to modelling with the Green’s Function Parabolic Equation (GFPE) method by Gilbert and Di. The purpose is to assess the accuracy of prediction of atmospheric sound propagation by methods that use detailed knowledge of the local geographical and meteorological conditions. Experimental data were collected during a one-week period in June 2005, and consist of data on the transmission loss (TL) of narrow band signals with frequencies 80, 200 and 400 Hz. Meteorological data were provided from radio sounding and balloon tracking up to 2-4 km in height at the receiver location and from meteorological sensors mounted on a 90 m high mast at the emission point. An atmospheric model including a laminar and a superimposed turbulent wind field was fitted to the meteorological data. Comparisons between the experimentally observed TL and predictions by the GFPE- model are presented. A satisfactory agreement is observed of the model-predicted transmission loss as a function of time to the experimental data. Introduction In the light of global warming the transition to renewable power sources is a crucial challenge to today's society. A power source that will probably play a major role in the future is wind turbine power. Until now most of the wind turbines are land based, however large off-shore farms are under construction or planned all over the world. These will exploit the vast wind resources available in at sea and by 2020 50 GW of installed capacity is planned in Europe [1]. Off-shore wind turbines are often located near a coast due to cost increases with increasing water depth. Such installations have given rise to concerns for noise pollution on shore, often in recreational regions unaffected by community noise. Since atmospheric sound propagation is highly

description

wind turbine noise

Transcript of Wtn09 Andersson

Page 1: Wtn09 Andersson

Long distance sound propagation over a sea surface Page 1 of 9

Third International Meeting on

Wind Turbine Noise Aalborg Denmark 17 – 19 June 2009

Long distance sound propagation over a sea surface

B.L. Andersson1, K. Bolin2, A. Cederholm1 & I. Karasalo1,2

Addresses: 1)FOI - Swedish Defence Research Agency, Stockholm, Sweden 2)KTH - Royal Institute of Technology, Stockholm, Sweden

e-mails: [email protected], [email protected], [email protected],

Abstract [email protected]

Results from measurements of sound propagation over a sea surface to a 10 km distant receiver are compared to modelling with the Green’s Function Parabolic Equation (GFPE) method by Gilbert and Di. The purpose is to assess the accuracy of prediction of atmospheric sound propagation by methods that use detailed knowledge of the local geographical and meteorological conditions. Experimental data were collected during a one-week period in June 2005, and consist of data on the transmission loss (TL) of narrow band signals with frequencies 80, 200 and 400 Hz. Meteorological data were provided from radio sounding and balloon tracking up to 2-4 km in height at the receiver location and from meteorological sensors mounted on a 90 m high mast at the emission point. An atmospheric model including a laminar and a superimposed turbulent wind field was fitted to the meteorological data. Comparisons between the experimentally observed TL and predictions by the GFPE-model are presented. A satisfactory agreement is observed of the model-predicted transmission loss as a function of time to the experimental data.

Introduction In the light of global warming the transition to renewable power sources is a crucial challenge to today's society. A power source that will probably play a major role in the future is wind turbine power. Until now most of the wind turbines are land based, however large off-shore farms are under construction or planned all over the world. These will exploit the vast wind resources available in at sea and by 2020 50 GW of installed capacity is planned in Europe [1]. Off-shore wind turbines are often located near a coast due to cost increases with increasing water depth. Such installations have given rise to concerns for noise pollution on shore, often in recreational regions unaffected by community noise. Since atmospheric sound propagation is highly

Page 2: Wtn09 Andersson

Long distance sound propagation over a sea surface Page 2 of 9

dependent upon the changing meteorological conditions, the level of such noise varies strongly with time. Measurements of long distance sound propagation over water surfaces concurrently with meteorological observations have been performed by Konishi and Tanioku [3,4]. However, the meteorological data were registered up to a few hundred meters height only, while knowledge of the meteorological conditions (wind velocity, humidity and temperature) further up in the atmosphere is in general needed for accurate predictions of the soundfield. The aim of this paper is to present measurements of sound propagation at 10 km distance [5] and evaluate the reliability of a sound propagation model with detailed knowledge of the meteorological and geographical conditions at the measurement times. This has been conducted by comparing transmission loss (TL) of the predictions to the measurements.

Measurements The measurements were performed between the 15th and the 21st of June 2005 in the Kalmar strait and the island Öland in the Baltic Sea. A motivation for this choice of experimental period was that most annoyance from wind turbine noise could be expected in the summer due to increased recreational outdoor activity.

FIG. 1: Measurement setup.

Acoustical measurements

Source site Two sound sources were placed on Utgrunden lighthouse located 9 km from shore. The sources were mounted at height 30 m on the lighthouse roof, with reference microphones placed 1 m front of respective source for recording the emitted signals.

Page 3: Wtn09 Andersson

Long distance sound propagation over a sea surface Page 3 of 9

The first source was a compressed-air-driven siren (Kockum Sonics Supertyfon AT150/200 with Valve Unit TV 784). It produced a 10-second signal with an average level of 130 dB and average frequency 200 Hz. Both this fundamental frequency and the first harmonic at 400 Hz were used in the analysis. The signal presented variations of the order of 1% in frequency and about 20 dB in sound level within each sound pulse, caused by the decreasing pressure of the compressed air driving the siren. The second source consisted of a sound generator coupled to a loudspeaker and a 1.2 m-long resonator tube. It produced a 1 minute long 80 Hz tone with constant sound pressure level of 113 dB at 1 m distance. Both sound sources were employed simultaneously.

Receiver site

The receiver site was on the island Öland 750 m from shore with ground height 7 m above sea level (see Fig. 1 for the experimental setup). The site was adjacent to the houses closest to the shoreline, in a quiet residential area. The receiver was a horizontal linear array of eight ½-inch microphones oriented parallel to the direction to the source. The array was placed at 1.7 m height accordingly to ISO 1996. The distance between the microphones was set to 40 cm, equal to half the wavelength at 400 Hz, to ensure a beam pattern free from grating lobes at all three frequencies. The signals were transmitted through a preamplifier to an UA100 analyzer and then processed in Matlab as explained in Ref. [5]. Meteorological measurements Source site

The wind speed was measured at 38, 50, 65, 80 and 90 m above sea level on a meteorological mast at the emission point. The wind direction was determined with wind vanes at 38 m and 80 m heights. The temperatures were measured at five heights: 6, 38, 50, 65 and 80 m. The relative humidity was measured at 38 m height. Data from these sensors were registered at 10 minutes intervals, and the average and standard deviations were recorded.

Receiver site

During the measurements performed in June 2005, wind profiles at the receiver site were measured during the day using radio probes and theodolite tracking of free flying balloons [6]. These measurements were performed by staff from the Department of Earth Sciences, Uppsala University. Wind velocity (horizontal components), humidity and temperature were measured up to 3500 m height.

Page 4: Wtn09 Andersson

Long distance sound propagation over a sea surface Page 4 of 9

GFPE Model The GFPE method was developed by Gilbert an Di [7,8] and later slightly improved by Salomons [9]. The method is particularly designed for atmospheric sound propagation and can use considerably longer range-steps than conventional PE methods. Because of its computational efficiency, the GFPE model was used in this study. Model description The method computes a 2D field in the rz-plane where r is the radial distance from the source and z is the vertical axis. From the 3D Helmholtz equation for the sound pressure, p, in cylindrical coordinates combined with a variable substitution φ=exp(-i k0 r) pr0.5 two expressions (1) and (2) can be derived [7,9]

−−∆×Φ+

+−−∆×−Φ+Φ×

×∆

=∆+

∞−∫

zirr

zikrr

r

ekkriri

dkekkkrikrkRkr

kzkrizrr

ββββ

π

δφ

))(exp(),(2

'))'(exp())',()'()',((21

)2

)(exp(),(

22

'22

2

(1)

where Δr is the horizontal step size, k(z)=ω/c(z) is the wave number, kr is a reference wave number (kr=k0=k(0) in this paper [9]), R(k’)=(k' Zg-kr)/(k' Zg+kr)) is the plane-wave reflection coefficient, Zg is the normalized ground impedance, β=kr/Zg is the surface-wave pole in the reflection coefficient and Φ is given in Eq.(2)

∫∞

−=Φ0

')',()'exp(),( dzzrikzkr φ (2)

which combined constitute the fundamental step in the GFPE-algorithm. Parameter selection The parameters were selected by suggestions from Ref. [7,9,10]. The horizontal step was set to 10λ and the vertical step size was 0.1λ in accordance to recommendations in Ref. [9]. An absorption layer that suppress spurious reflections from the upper limit of the numerical integration has an absorption parameter, A, calculated according to Ref. [9] with a depth of 75λ. The attenuation coefficients were calculated according to ISO/DIS 9613-1 [11]. To calculate the surface impedance the model by Embleton et. al. [12] was used. The ground impedance is determined by the sound frequency and a flow resistivity parameter, which was selected to a value representative of grass in a rough pasture [12]. The water surface was treated as perfectly reflecting. Turbulence Effects of wind and temperature fields were included in the GFPE model following the approach in Appendix I and J of Ref. [10]. Thus, the turbulent components of

Page 5: Wtn09 Andersson

Long distance sound propagation over a sea surface Page 5 of 9

these fields are modelled as homogeneous random fields with von Karman type horizontal wavenumber spectra. The effect of such turbulence on the GFPE solution is represented by including a random z-dependent phase factor in the GFPE propagator, without requiring explicit computation of realizations of the fields. According to this turbulence model the transmission loss to the receiver is a stochastic variable, and statistics of the transmission loss were determined by carrying out 50 Monte Carlo runs for each frequency at every hour of the experimental week. Meteorological assumptions Meteorological inputs to the GFPE model were balloon measurements (horizontal wind velocity), radio balloon (relative humidity and temperature) and the anemometers on the mast (standard deviation of wind speed and temperature). The wind and radio balloon data were used as meteorological parameters (U(z), rh, T) of the laminar atmosphere. Linear interpolation was used between measurement points in the vertical direction as well as in time. The mast data (horizontal wind speed and the variances of wind velocity and temperature) were used to estimate the turbulence intensities.

Results

As previously stated our objective is to investigate the accuracy of the model predictions compared to measurements, in particular to investigate the effect of turbulence on the model-predicted transmission loss. Turbulence excluded The black curves in Fig. 2 show the simulated TL, the average of the simulated TL during measurement periods are shown as horizontal yellow lines. Red dots show measured TL values, with their daily averages shown by horizontal green lines. It can be clearly seen that the predictions show larger variations of TL with time than the experimental data. Periods with low TL show good agreement between the measured and the modelled TL, whereas the modelled TL is severely overestimated in periods periods where TL is high. High TL values occur when the sound speed decreases as function of height, causing the emitted sound to be refracted upwards and shadow zones to occur at the receiver. An example is shown in Fig. 3. Low TL values occur when the sound speed has a local maximum at relatively low height (e.g. induced by a low level jet [6]), causing the sound to be refracted downwards and trapped within a channel below the local wind maximum. Such meteorological conditions occurred e.g. around noon of June 17 as can be seen in Fig. 4.

Page 6: Wtn09 Andersson

Long distance sound propagation over a sea surface Page 6 of 9

a) TL at 80 Hz as function of time. Black curve: b) TL at 200 Hz as a function of time Black Predicted with a laminar atmosphere curve: Predicted with a laminar atmosphere model. Red dots: Measured data. model Red dots: Measured data.

. c) Transmission loss at 400 Hz as a function of time. Black curve: Predicted with laminar atmosphere in the model. Red dots: measured data.

FIG 2: TL as function of time. (o): Measured (-): Predicted with a laminar atmosphere in the model. Daily averages for measured and predicted TL are shown as green respectively yellow lines.

FIG 3: Sound speed profile (left figure) and simulated 80 Hz sound field (right) at noon on June 16. The sound speed decreases with height and the receiver is in sound shadow. Atmosphere modelled as laminar.

Page 7: Wtn09 Andersson

Long distance sound propagation over a sea surface Page 7 of 9

FIG 4: Sound speed profile (left figure) and simulated 80 Hz sound field (right) at noon on June 17. The sound speed has a maximum at height 200 m and a sound channel below this height can be observed. Atmosphere modelled as laminar.

FIG. 5: Simulated soundfield in the 80 Hz case of Fig 3, with effects of a turbulent atmosphere included. Turbulence included As discussed in Chapter 5 of Ref [10], the effect of turbulence is a random scattering of sound, leading to increased sound levels in refractive shadow zones. This effect is illustrated in Fig. 5, showing the model-predicted 80 Hz soundfield of Fig 3, but with the atmosphere modelled as turbulent. Clearly, the shadow zone of the laminar case (Fig 3) becomes less pronounced when turbulence effects are included (Fig 5). In Fig. 6 the simulations of Fig 2 are repeated, now including effects of turbulence. The thick black curve shows the average value of the TL from the Monte Carlo simulation and the thinner black curves surrounding the average show the interval of the standard deviations. Comparing Fig. 2 and Fig. 6 it can be seen that the most prominent change in the predictions caused by turbulence is a significant decrease of the high TL values, leading to a significantly improved agreement between the predicted and measured TL levels.

Page 8: Wtn09 Andersson

Long distance sound propagation over a sea surface Page 8 of 9

a) TL at 80 Hz as function of time. Black curve: b) TL at 200 Hz as a function of time Black Predicted with a turbulent atmosphere curve: Predicted with a turbulent atmosphere model. Red dots: Measured data. model Red dots: Measured data.

c) Transmission loss at 400 Hz as a function of time.

Black curve: Predicted with turbulent atmosphere in the model. Red dots: measured data.

FIG. 6. TL as function of time. (o): Measured. (-): Predicted with a turbulent atmosphere in the model. Daily averages for measured and predicted TL are shown as green respectively yellow lines.

Conclusions The results indicate that sound propagation modeling including effects of detailed meteorological data can be used for reliable prediction of transmission loss. In particular, the predicted TL remains reasonably accurate under varying meteorological conditions, and follows the variations observed in the TL measurements in a realistic way. The results further indicate that the sound propagation model must include effects of turbulence in the atmosphere for accurate predictions of the TL into shadow zones.

Acknowledgements

The authors wish to thank Prof. Sten Ljunggren, Prof. Hans Bergström, and Civ.Ing. Hans Olsson for helpful comments and discussions. Vindforsk II is acknowledged for their financial support.

Page 9: Wtn09 Andersson

Long distance sound propagation over a sea surface Page 9 of 9

References [1] E. W. E. Association, “Ewea’s response to the European commission’s green paper”, Technical Report, (COM (2006) 275 Final) (2007). [2] L. Johansson, “Sound propagation around off-shore wind turbines -long-range parabolic equation calculations for Baltic sea conditions”, Technical Report, Licentiate thesis, KTH/ Building sciences (2003). [3] Z. M. K. Konishi, Y. Tanioku, “Long time measurementof a long range sound propagation over an ocean surface”, Applied Acoustics 61 (2000). [4] Z. M. K. Konishi, “Interpretation of long term data measured continuously on long range propagation over sea surfaces”, Applied Acoustics 62 (2001). [5] M. Boué, “Long-range outdoor sound propagation over sea, applications to wind turbine noise”, Technical Report TRITA-AVE 2007:22, KTH (2007). [6] K. T¨ornblom, “Thermally driven wind modification in coastal areas and its influence on sound propagation with application to wind power”, Technical Report, Department of Earth Sciences, Uppsala University (2006). [7] K. E. Gilbert and X. Di, “A fast green’s function method for one-way sound propagation in the atmosphere”, Journal of the Acoustical Society of America 94, 2343–2352 (1993). [8] X. Di and K. E. Gilbert, in Proceedings of the Fifth International Symposium on Long Range Sound Propagation, 128–146 (Milton Keynes, England) (24-26 May 1992). [9] E. M. Salomons, “Improved greens function parabolic equation method for atmospheric sound propagation”, Journal of the Acoustical Society of America 104, 100–111 (1998). [10] E. M. Salomons, Computational Atmospheric Acoustics (Kluwer Academic Publishers) (2002). [11] International Organization for Standardization, Geneva, Switzerland, ISO/DIS 9613-1: Attenuation of sound during propagation outdoors- Part 1: Atmospheric absorption (1995). [12] T. F. W. Embleton, J.E. Piercy and G. A. Daigle. Effective flow resistivity of ground surfaces determined by acoustical measurements, Journal of the Acoustical Society of America, 1983, 74, 1239-1244