Parameters and attenuation terms

The full definition of the_​N

​ ​ eq is completed by defining the source parameter ​β, the attenuation

term _​A

​ ​ ​ ​ ​ atm,grond and the correction ​Dθ,wind.

The broadband trailing edge noise is the dominant WTN source (Oerlemans, et al., 2007). It is generated in the outer part of the blades and its power scales with the fifth power of the local flow speed. Also the inflow turbulent noise and the tip noise are other important mechanisms of noise generation that scales with the fifth power of the wind speed (Fleig, et al., 2004; Oerlemans, et al., 2009; Tangler, 2000).

The blades rotational speed (_​N

​ ) is the most important parameter affecting the noise emission

(Leloudas, et al., 2007) and the wind speed incoming on the blades is generally linearly related to _​N

​ . The total noise emission is proportional to the fifth power of wind speed and

consequently of _{​N (Oerlemans, et al., 2007; Lee, et al., 2009; Jianu, et al., 2012), even} looking at the immission noise at receiver far from the sources (Van den Berg, 2004). Thus, for the purpose of the procedure, _​β

​ was set equal to 5.

According to the ISO 9613, the air absorption is a function of frequency, temperature and humidity. The ground effect attenuation depends on the orography, on the ground typology and on the distance. It is possible to combine them in a single attenuation term linearly dependent on distance (8.5).

Different α were estimated for both flat and complex terrain. For flat terrain the Monte Carlo Method (MCM) was applied using the ISO 9613 as noise propagation model to the WTN power spectrum of the “2MW Vestas v100”, with the uncertainty reported in Table 1. The noise power spectra do not significantly vary among different turbine models (Van den Berg, et al., 2008). In details, the analysis followed these steps:

- from a distance of 50 m from the turbine up to a distance of 1500, a set of 30 calculation points was taken, one every 50 meters;

- in each of these points, a noise levels distribution was computed through an MCM approach using the ISO 9613 propagation model and a number _{​m of runs, with a random sampling of} the parameters reported in Table 8.1 within their distribution ranges. The temperature and humidity were chosen according to the typical Italian climatic values;

- _{​m sound level trends over the distances were obtained by a random sampling of a single} level value from each levels distribution over distance;

- the alpha distribution was obtained calculating an OLS linear regression over the distance for each _​m

​ sound level trend and it is reported in Figure 8.6;

- the number of runs _{​m was adjusted to assure the convergence of the alpha distribution,} about _​m

​ =10_​4​ _{runs were sufficient.}

Therefore α= 3 dB/km was obtained from the average value of the alpha distribution for flat terrain, rounded to the first integer.

Table 8.1. Distribution, mean and range for the parameter used in MCM calculation.

Figure 8.6. α distribution resulting from the MCM calculation. The continuous line is the average, the dotted lines are the 95% quantiles coverage interval.

The same approach applied to hilly terrain and considering the average over various random altitude profiles for single turbine pathways resulted in a range of different values depending to the pathway itself. An average value of α= 5 dB/km was chosen corresponding to a higher ground absorption that may happen in complex terrain compared to a flat one.

The_​D

​ θ,wind considers the influence of wind direction and meteorological effects as wind and

temperature gradients in the propagation of noise.

represents the angle between the North and the prevailing wind direction during the 10 min

θ

interval at the hub of the i _​th _{wind turbine while φ is the angle between the North and the line}

joining the receiver with the i _​th _{wind turbine. Both angles are referred with positive direction}

in a clockwise direction. The cosine expression was chosen to have a continuous variation from downwind to upwind conditions. The downwind condition correspond to a

, while the upwind to . The different sign in Equation 8.6 for 80°

θ_i− φ_i= 1 θ_i− φ_i= 0 °

daytime or night-time conditions was chosen in accordance with the γ values calculated in the following, in order to have the appropriate attenuation values.

The coefficient γ depends on the meteorological conditions, such as wind speed and atmospheric stability. Its value was estimated with (8.6) calculating the broadband sound attenuation due to the meteorological effects in the CONCAWE propagation model (Manning, 1981). A distance source-receiver of 500 m was considered, sufficient for having significant meteorological effects. The sound power spectrum used is an average spectrum (Van den Berg, et al., 2008). A reference noise level spectrum at the receiver has then been calculated subtracting the geometrical divergence, air absorption and ground effect attenuations. The attenuations extrapolated from the CONCAWE meteorological curve for each frequency band were subtracted from the reference spectrum obtaining a spectrum at the receiver for each meteorological category of CONCAWE. The broadband meteorological attenuation for each category was obtained subtracting the broadband meteorological levels from the broadband reference level. The resulting attenuations are reported in Table 8.2.

Table 8.2. CONCAWE categories and relative attenuation calculated.

The broadband attenuations were calculated for daytime/night-time and downwind/upwind condition weighting the attenuations in Table 8.2 with the percentages in Table 8.3. The percentages represent the estimated probability for each CONCAWE category corresponding to the average meteorological condition supposed. The results are in the “CONCAWE attenuations” row of Table 8.3.

Table 8.3. Probability of the CONCAWE categories supposed, with the meteorological attenuation calculated.

The estimated attenuations in the last row of Table 8.3 are calculated using Equation 8.7 in accordance with Equation 8.6 taking the opposite of _​D

​​ θ, rounded to 0.5, with:

The attenuations corresponding to the γ _​day and γ_​night​resulted comparable to the CONCAWE ones reported in Table 8.3, in a ±0.5 dB(A) range.

A quick view of the benefits brought introducing the N_​eq parameter comes from the comparison of Figure 8.7 and Figure 8.8. In Figure 8.7 the noise levels (_​L

​ Aeq,10 min ) measured

in “Scansano” and in “La Miniera” are related to the N of the nearest wind turbine, while in Figure 8.8 the same noise levels are related to the N _​eq​. The noise levels are separated into day and night periods. The dispersions in Figure 8.8 have smaller spreads of data than those in Figure 8.7, showing that noise has a better relation with N _​eq than with the N of a single

turbine.

Figure 8.7. Dispersion of L

​ ​ Aeq,10 min as a function of N of the nearest turbine in “Poggi alti -

Scansano”(upper) and “La Miniera - Scapiccioli” (lower) during daytime (blue) and night-time (red).

Figure 8.8. Dispersion of L

​ ​ ​ ​ Aeq,10 min as a function of Neq in “Poggi alti - Scansano” (upper) and “La

Miniera – Scapiccioli” (lower) during daytime (blue) and night-time (red).