Prediction of wind turbine noise propagation

Abstract

In order to meet the US Department of Energy’s projected target of 35% of the US energy coming from wind by 2035, there is a strong need to look into the management and development of wind turbine technology and its impacts on human health, wild life, and the environment. The prediction of wind turbine noise and its propagation is very critical to study the impacts of wind turbine noise for long-term adoption and acceptance by neighboring communities. This article presents a study on the prediction of wind turbine noise propagation using an open access software and the publicly available topological map of projected sites. The noise propagation prediction utilized Openwind, a software package used for wind project design and optimization, to generate a noise map based on inputs acquired from a potential wind energy demonstration site. The predicted noise maps were incorporated with the measured ambient noise data to predict the resultant noise level in the surrounding neighborhood under different scenarios.

Keywords

Ambient noise human health noise prediction noise propagation wild life wind turbine noise

Introduction

Understanding and managing the impact of wind turbine technology on the environment is a critical factor for successful long-term adoption and acceptance of the technology by the neighboring community and mitigating stress on impacted wildlife. The issues in wind turbines include environmental impact causing concern for both humans and wild life, noise and vibration caused by operation, and visual and esthetic impacts. Noise and vibrations caused by the wind turbines can be particularly important factors to consider when a wind energy generation site is being considered and thus the prediction of possible noise becomes a concern before the turbines are installed. It is imperative to consider these potential issues in view of the US Department of Energy’s projected target of 35% of the US energy coming from wind by 2035 (US Department of Energy, 2015). Noise generated by turbines can be perceived to cause annoyance, sleep disturbance, as well as self-reported instances of stress and quality-of-life issues (Health Canada, 2014; Michaud et al., 2016; Saavedra and Samanta, 2015).

There are many different methods, approaches, and procedures in order to produce a wind turbine noise propagation model. An easily understood example of turbine noise propagation can be observed in Figure 1 (General Electric, 2014). This graphic compares sound levels at different distances from a wind turbine and equates them to various appliances, for instance, the wind turbine sound at a distance of 100 m (with flat terrain, no vegetation, and no ambient noise) is about as loud as a midsize window air conditioner.

Figure 1.

Average turbine noise levels at various distances (General Electric, 2014).

The objective of this study is to present a method of wind turbine noise propagation using an open access software platform and the publicly available topological maps of projected wind energy sties. The predicted wind turbine noise maps are combined with the measured ambient noise data to estimate the potential noise levels in the neighborhood. The procedure is illustrated using a potential wind energy demonstration site for different scenarios.

The rest of the article is organized as follows. Section “Literature Review” presents a review of wind turbine noise propagation models focusing on two open access software platforms. Next, the details of noise propagation model used for this study are presented in section “Openwind^® noise propagation model.” In section “Results and Discussion,” the measured ambient noise data at a potential wind energy demonstration site are presented. Results of the predicted noise maps and the potential resultant noise levels in the neighborhood under different scenarios are presented next. The salient features of the study are summarized in section “Conclusion.”

Literature review

Wind turbine noise prediction methodologies

This section gives an overview of two open access software platforms along with the advantages and disadvantages. This section also reviews the literature in the fields of sound attenuation due to vegetation and urban form.

Openwind

Openwind is an open source software package that is used for wind project design and optimization. It is a platform that more intuitively integrates geographic information system (GIS)-based site information in order to promote industry collaboration and research in a flexible wind project design platform. Inputs for this software program include site selection and land constraints, wind resource and energy production potential, and turbine and operation characteristics. Site selection entails the construction of a wind resource grid defined within a GIS framework that will be the basis for generating data for the program. Resource and energy potential are based on hub height wind speed maps as well as meteorological data. The characteristic parameters of the turbines to be constructed are also taken into account in the model (AWS Truepower, 2017; Filippeli, 2013). The noise model in Openwind is based on ISO 9613-2, which is the international standard for the propagation and attenuation of industrial noise (ISO 9613-2:1996, 1996). The program makes several simplifying assumptions in the generation of a noise map including the following: all noises are treated as point sources, all propagation is assumed to be in the same direction as the wind, and obstacles and blocking effect of terrain can be ignored due to wind turbines being classified as aerial sources (AWS Truepower, 2017). This method of calculation results in a fairly conservative picture of wind noise propagation; however, it could have higher accuracy if all types of attenuation like downwind versus upwind effects, vegetation, and terrain topography were taken into account. Openwind was chosen for this project due to its accessibility, better documentation, and flexibility of using publicly available Google Earth GIS data for a potential wind energy demonstration site.

SPreAD-GIS

SPreAD-GIS is an open access software application and is implemented as a toolbox in ArcGIS software, a commercial GIS program. The main purpose of SPreAD-GIS is to model patterns of noise propagation caused by manmade sources in natural outdoor environments. This toolbox uses the datasets on land cover, topography, and weather condition from the GIS software to calculate noise propagation and excess noise above ambient conditions for the one-third octave frequency bands around one or multiple sound sources (Reed et al., 2012). The use of GIS plays an important role in noise mapping as it can greatly improve the accuracy of results obtained from noise modeling. In the field of research, it will be important for standardization in order to optimize quality and efficiency of noise effect studies. Standardization will also be important because the results of different studies can only be combined or compared if the same parameters for noise exposure and the same analysis methods are used (Kluijver and Stoter, 2003).

SPreAD-GIS emulates acoustic detection range prediction model (ADRPM), a model that was developed under the sponsorship of US Army Tank Automotive Command (TACOM) for estimating the acoustic detection range of motor vehicles (Reed et al., 2012). SPreAD-GIS can be a more accurate program to model noise propagation due to the more capable GIS functions embedded within the ArcGIS software. The actual calculation for the noise propagation is the same as in Openwind and uses ISO 9613-2; if the elevation and vegetation data can be obtained, they can be more thoroughly integrated with SPreAD-GIS than in Openwind, thus making the model more accurate. SPreAD-GIS was not used in this study due to its need of third party GIS data in order to create the model and a commercial program module to run it.

Calculating noise levels (noise prediction)

Noise levels at a receiver point can be calculated as opposed to being measured, thus allowing noise models to be created. In some cases, calculation is the preferred method even where noise measurement can be conducted. This can occur in cases such as where levels are contaminated by high levels of background noise, where future levels need to be predicted, where noise reduction scenarios need to be compared, where noise contour maps need to be produced, and where there is limited access to a measurement position (Brüel & Kjær, 2000). Outdoor environments provide a challenge in terms of noise calculation due to the lack of uniformity. Changing meteorological conditions can easily cause fluctuations in sound levels by 10–20 dB over the time periods as short as a few minutes. The longer the transmission path, the larger will be the fluctuations. Outdoor sound propagation is affected by many factors including obstructions, terrain type, atmospheric conditions, metrological conditions, and source geometry and type (Lamancusa, 2009; Larsson and Ohlund, 2014; National Physical Lab (NPL), 2006).

Vegetation’s effect on noise propagation

While not considered in the approach taken in this study, vegetation with dense foliage can have a major effect on the propagation of sound outdoors. It has been found that ground attenuation and scattering accounts for the highest amount of sound reduction from vegetation (Aylor, 1972). It has also been observed from studying tree belts’ effects on point source noise propagation, which is a negative logarithmic relationship between relative attenuation and the visibility, as well as a positive logarithmic relationship between relative attenuation and the width, length, or height of the tree belts exists (Fang and Ling, 2003).

Urban form’s effect on noise propagation

Obstructions can attenuate noise greatly from noise generation sources, which was observed by studies done on traffic noise. Barriers can effect sound attenuation based on the angle of obstruction, change in path length, observer height, and source height (Pamanikabud and Tansatcha, 2003). It has been found that urban forms in historical areas with narrower roads, complex road networks, and a higher density of intersections lead to lower traffic noise and thus lower noise pollution (Tang and Wang, 2007). Some urban areas have gone so far as to combine barriers with vegetation by introducing greenery on external building elements. This has had an even greater attenuating effect on long-distance noise propagation (Tang and Wang, 2007).

Openwind^® noise propagation model

In this section, details of noise propagation model of the open access software platform Openwind along with the standards are presented.

Standards

ISO 9613-2 specifies an engineering method for calculating the attenuation of sound during propagation outdoors and is used by Openwind to predict the sound propagation of wind turbines in terms of simple A-weighted sound pressure levels (AWS Truepower, 2017; ISO 9613-2:1996, 1996). A-weighted sound pressure level accounts for the change in sensitivity in human ears at varying frequencies. A-weighting corresponds approximately to the 40-phon equal-loudness curve and applies to the human ear’s response at low to medium sound levels. It is the current International Electrotechnical Commission (IEC) acoustic standard for sound and noise measurement (Brüel & Kjær, 2000; Saavedra and Samanta, 2015). The method contained in ISO 9613-2 predicts the equivalent continuous A-weighted sound pressure level under meteorological conditions favorable to propagation. Under this standard, several assumptions must be made. All noise sources are treated as point sources; all noise propagation is assumed to be in the same direction as the wind; atmospheric conditions are assumed favorable to noise propagation; and wind speeds that are between 3 and 11 m above ground level are assumed to be between 1 and 5 m/s. ISO 9613-2-1996 (1996) considers several types of attenuation including atmospheric, geometric spreading and ground effect (porosity).

ISO 9613-2 introduces basic equations used by Openwind to predict the attenuations of noise outdoors in community environments at a distance from a variety of sources of known sound emission. These conditions are for downwind sound propagation under moderate ground-based temperature inversion (temperature rises as altitude increases) such as that occurs at night. Temperature inversion over water surfaces will not be considered accurate as it may result in higher sound pressure levels than predicted. This method is applicable in practice to a great variety of noise sources and environments.

$L_{f T} (D W)$ is the equivalent continuous downwind octave band sound pressure level at a receiver location. This is calculated for each point source, and its image sources, and for the eight octave bands with nominal midband frequencies from 63 Hz to 8 kHz (ISO 9613-2:1996, 1996)

L_{f T} (D W) = L_{W} + D_{c} - A

(1)

where $L_{W}$ is the octave band sound power level, in decibels, produced by the point source relative to a reference sound power of 1 pW; $D_{c}$ is the directivity correction, in decibel, which describes the extent to which the equivalent continuous sound pressure level from the point sound source deviated in a specific direction from the level of an omnidirectional point sound source producing sound power levels $L_{W}$ , and it equals the directivity index $D_{I}$ of the point sound source plus an index $D_{Ω}$ that accounts for sound propagations into solid angles less than 4π steradians; for an omnidirectional point sound source radiating into free space, $D_{c} = 0 dB$ ; and $A$ is the octave band attenuation, in decibels, which occurs during propagation from the point sound source to the receiver.

The attenuation term A in equation (1) is given by equation (2) (ISO 9613-2:1996, 1996)

A = A_{d i v} + A_{a t m} + A_{g r} + A_{b a r} + A_{m i s c}

(2)

where $A_{d i v}$ is the attenuation due to geometrical divergence, $A_{a t m}$ is the attenuation due to atmospheric absorption, $A_{g r}$ is the attenuation due to ground effect, $A_{b a r}$ is the attenuation due to a barrier, and $A_{m i s c}$ is the attenuation due to miscellaneous other effects (such as foliage, industrial noise, propagation through houses or buildings).

The equivalent continuous A-weighted downwind sound pressure level $L_{A T}$ shall be obtained by summing the contributing time-mean-square sound pressures calculated according to equations (1) and (2) for each point sound source, for each of their image sources, and for each octave band, as specified by equation (3) (ISO 9613-2:1996, 1996)

L_{A T} (D W) = 10 l g {\sum_{i = 1}^{n} \sum_{j = 1}^{8} 10^{0.1} [L_{f T} (i j) + A_{f} (j)]} d B

(3)

where $n$ is the number of contributions $i$ (sources and paths), $j$ is an index indicating the eight standard octave band midband frequencies from 63 Hz to 8 kHz, and $A_{f}$ denotes the standard A-weighting.

To apply this method, parameters such as the geometry of the source and of the environment, the ground surface characteristics, and the source must be known. Accuracy limitations for this method can include the attenuation of sound propagation outdoors between a fixed source and a receiver fluctuating due to variations in the metrological conditions along the propagation path (AWS Truepower, 2017).

Noise model

A noise model theoretically estimates noise levels within a region of interest under specific parameters. It is important to understand that the specific set of conditions for which the noise is being modeled will be only a “snapshot” of a certain environment. Physical environments, particularly those found outdoors, will not be fixed but have constantly varying conditions leading to constantly varying sound fields. Recognizing that modeling is a means of estimating noise, it is imperative to validate the model predictions with the measured data.

A noise propagation model for a potential wind energy generation site in coastal Georgia was created. The model consisted of four Bergey Excel 10 turbines, with a total sound power level of 90.18 dB(A) which is representative of L_W in equation (1), with a noise map generated at an observer height of 1.75 m (5.7 ft) above ground level. The noise analysis was conducted using the single A-weighted sound power level (ISO 9613-2) noise model in Openwind. In this model, total sound power level at source (dB(A)), atmospheric attenuation, attenuation due to geometric spreading, ground effect attenuation, and site-specific temperature and humidity information were all accounted for. A site-specific relative humidity of 52.2%, a site-specific temperature of 33.57°C, and a site-specific air density of 1.139 kg/m³ were used as inputs for A_atm in equation (2). In addition, an observer height of 1.75 m and a ground porosity of 0.75 were used and are representative of A_gr in equation (2). The total sound power level of 90.18 dB(A) (L_W) for the Bergey Excel 10 turbine was taken from manufacturer specifications (Bergey WindPower, 2010). A correction factor of −10 dB(A) was also used to compensate for miscellaneous attenuation and is representative of a negative A_misc term in equation (2). This equates to some noise amplification from the environment due to echoing and hard-packed asphalt surfaces (AWS Truepower, 2017). This noise model was used to predict sound levels at residences that may be affected by wind turbine noise. These are considered receptors for noise calculation purposes (Kwong et al., 2012).

This model does come with several simplifying assumptions with the first being that reflections are ignored as wind turbines are aerial sources of noise. This sound model is only representative of an area of land with flat or constantly sloping terrain. The model does not take into effect terrain features such as hills that can act as barriers to the sound propagation. The model does not also take other obstacles into account such as vegetation or buildings. In the model ambient noise is also ignored.

Results and discussion

Wind turbine ambient noise dataset

A relatively less populated area in southern Georgia with the potential of a small wind energy demonstration site was chosen for ambient noise measurement. A hand-held portable dual-channel sound level meter with fast Fourier transform (FFT) analysis capability from a leading international manufacturer (Reed et al., 2012) was used for ambient noise measurement for this study. The device was configured with all necessary software installed and tested. The transducers and analyzers were calibrated as per the established procedure. Measurements were taken at the proposed locations of wind turbines and the meteorological tower. The microphone provided measurements for the ambient sounds in dB(A). In total, 10 2-min measurements were taken. Five locations were designated and two measurements for 2 min were taken at each location. The designated locations were chosen at the installation points for the meteorological tower and four turbines (Figure 2).

Figure 2.

Mesurement location of the noise dataset.

The first two recordings were taken at the Met Tower Location. The second two recordings were taken at the Turbine 4 location. The third set of recordings was taken at the Turbine 3 location. The fourth set of recordings was taken at the Turbine 2 location. The fifth set of recordings was taken at the Turbine 1 location. These locations can be observed in Figure 2. All noise measurements were taken during a normal working day around 11 A.M.–2 P.M.

Analysis of initial measurement data

Five specific locations were chosen for data measurement and recording. The five locations are where the meteorological tower and four turbines are to be installed. For each location, two datasets were taken. Each dataset was with a duration of 2 min, resulting in 20 min of logged data. Table 1 represents the average and peak data of the desired parameters. Here L_Aeq is the A-weighted equivalent continuous noise level and L_Aeq is the A-weighted equivalent continuous noise level peak. It was a relatively windy day. The average wind speed varied in the range of 2.20–4.60 m/s, the minimum was in the range of 0.30–1.60 m/s, and the maximum was in the range of 5.60–12.10 m/s. The ambient noise level was in the range of 42.5–52.1 dB(A) with peak values in the range of 47.1–65.7 dB(A).

Table 1.

Ambient measurements of installation site.

	Average L_Aeq (dB)	Peak L_Aeq (dB)
Site A1	44.5	49
Site A2	44.1	47.1
Site B1	45.3	54.1
Site B2	44.5	53.5
Site C1	42.5	49.6
Site C2	44.2	52.1
Site D1	45.9	51.4
Site D2	45.7	49.8
Site E1	52.1	65.7
Site E2	43.6	52.1
Max	52.1	65.7
Min	42.5	47.1
Average	45.2	52.44

To distinguish between the different decibel measurements, their definitions are listed as follows:

L_Aeq is the A-weighted equivalent continuous sound level;

L_Apeak is the A-weighted maximum sound level.

Wind turbine noise prediction results

Methods

Google Earth was used to retrieve satellite imagery for the potential wind energy demonstration site (Figure 3). Polygons of the proposed site were then taken to gain an accurate geometric model of the land area with correct coordinates and spatial data (Figure 4). Points were then used to index the location of the turbines, meteorological tower, and occupied residences (Figure 5). These spatial data were used to create a shape file (.shp) to represent the potential site geometrically (Figure 6).

Figure 3.

Satellite imagery of the potential wind energy demonstration site.

Figure 4.

Polygon to represent the potential site in the model.

Figure 5.

Points to represent homes and turbines in the model.

Figure 6.

Completed shape file to represent the potential site’s spatial data.

Predicted noise maps

Given the inputs specified previously, noise maps were predicted for the potential site with one turbine (Figure 7), two turbines (Figure 8), three turbines (Figure 9), and four turbines (Figure 10). The number of turbines used is representative of n in equation (6). The models depicted show isolines of a 5-dB(A) gradation from the turbines in each case. It should be noted that the predicted noise maps are fairly conservative as the prediction model does not consider the effects of attenuation due to vegetation, terrain topology, or environmental factors other than ambient conditions (temperature, air density, relative humidity). The presence of these attenuating effects would have caused the lines of constant sound level to be closer together giving steeper sound gradient.

Figure 7.

Noise map with one turbine.

Figure 8.

Noise map with two turbines.

Figure 9.

Noise map with three turbines.

Figure 10.

Noise map with four turbines.

Table 2 depicts the various sound intensity levels at the homes on the map determined by observing the isolines at a resolution of 1 dB(A) per iteration. Table 2 also shows the maximum sound intensity level predicted at 450 ft from the turbine layout. These results could vary slightly from other future studies due to the lack of land parcel data that would accurately place the property lines of the homes in the geometry of the model. The results can also differ slightly due to several different site-specific ambient conditions chosen including temperature, site-specific air density, wind speed, and relative humidity.

Table 2.

Predicted sound intensity at homes on the potential site.

Predicted sound intensity in dB(A) from Openwind
#Turbines	Max (450 ft from turbine)	Home 1	Home 2	Home 3	Home 4	Home 5
1	46.0	22.0	14.0	16.0	15.0	15.0
2	50.0	25.0	22.0	19.0	18.0	18.0
3	50.0	28.0	25.0	22.0	21.0	20.0
4	50.0	20.9	26.0	23.0	23.0	22.0

Analysis

The ambient sound measurements and the theoretical sound predictions for the turbines were combined to generate a total sound prediction at the residential locations in the neighborhood of the potential wind energy demonstration site.

Sample calculation

The formula for the sum level of sound pressures of n incoherent radiating sources is

L_{\sum} = 10 * \log_{10} (\frac{p_{1}^{2} + p_{2}^{2} + \dots + p_{n}^{2}}{p_{0}^{2}}) = 10 * \log_{10} ({(\frac{p_{1}}{p_{0}})}^{2} + {(\frac{p_{2}}{p_{0}})}^{2} + \dots + {(\frac{p_{n}}{p_{0}})}^{2})

(4)

The reference sound pressure p₀ is 20 µPa = 0.00002 Pa = 2 × 10⁻⁵ Pa (root mean square (RMS)) ≡ 0 dB.

From the formula of the sound pressure level, we find

{(\frac{p_{i}}{p_{0}})}^{2} = 10^{\frac{L_{i}}{10}}, i = 1, 2, \dots, n

(5)

This inserted in the formula for the sound pressure level to calculate the sum level shows (Sengpiel, n.d.)

L_{\sum} = 10 * \log_{10} (10^{\frac{L_{1}}{10}} + 10^{\frac{L_{2}}{10}} + \dots + 10^{\frac{L_{n}}{10}}) d B

(6)

where L_Σ is the total level and L₁, L₂, …, L_n are the sound pressure levels of the separate sources in dBSPL. For example, adding three decibel values, that means levels 94.0 + 96.0 + 98.0

L = 10 * \log_{10} (10^{9.4} + 10^{9.6} + 10^{9.8}) = 101.1 dB

(7)

Predicted level with ambient noise

Using the method described above, the peak predicted sound intensity levels, at a distance of 450 ft from the turbines, were considered along with the peak value measured at the turbine locations (52.10 dB(A)) for the max column. The predicted sound intensity levels at the homes were considered along with the average residence values measured (46.57 dB(A)) for the home columns (Table 3).

Table 3.

Predicted sound intensity at residences on the potential site summed with measured values.

Sound intensity in dB(A): Combined noise level of turbines and ambient (46 dB(A))
#Turbines	Max (at 450 ft from turbine)	Home 1	Home 2	Home 3	Home 4	Home 5
1	53.1	46.6	46.6	46.6	46.6	46.6
2	54.2	46.6	46.6	46.6	46.6	46.6
3	54.2	46.6	46.6	46.6	46.6	46.6
4	54.2	46.6	46.6	46.6	46.6	46.6

It can be observed from Table 3 that the effect of sound propagation from the wind turbines in the residential areas is negligible when considered along with the average (46 dB(A)) ambient noise measured at the site. Even the worst-case scenario of the maximum value at 450 ft from the turbine location would only increase the dB(A) value by 4.2–7.1 dB(A) with an ambient noise of 46 dB(A). The change of 3 dB(A) is the threshold of perception for the human ears so an increase of 4.2–7.1 dB(A) would be noticed by a human observer in the presence of background noise (Bolt, Beranek and Newman, 1973).

It can be observed from Table 4 that the effect of sound propagation from the wind turbines in the residential areas is again negligible when summed with an estimated 30 dB(A) ambient noise at night. The sound intensity levels of the wind turbines are much lower than the 30 dB(A) ambient noise. The effect observed at the residences from the noise generated by the wind turbines would be an increase in the range of 0.1–2.5 dB(A) barely noticeable to a human observer.

Table 4.

Predicted sound intensity at residences on the proposed site summed with an estimated 30 dB(A) night ambient noise.

Sound intensity in dB(A): combined noise level of turbines and predicted ambient (30 dB(A))
#Turbines	Max (at 450 ft from turbine)	Home 1	Home 2	Home 3	Home 4	Home 5
1	46.1	30.6	30.1	30.2	30.1	30.1
2	50.0	31.2	30.6	30.3	30.3	30.3
3	50.0	32.1	31.2	30.6	30.5	30.4
4	50.0	32.5	31.5	30.8	30.8	30.6

It can be observed from Table 5 that the effect of sound propagation from the wind turbines in the residential areas could be noticeable under these conditions when considered along with an estimated 20 dB ambient noise at night. The sound intensity levels of the wind turbines are much closer to the 20 dB(A) ambient noise predicted than in the other scenarios. The effect observed at the residences from the noise generated by the wind turbines would be an increase in the range of 1–9.5 dB(A) which could be noticeable to a human observer.

Table 5.

Predicted sound intensity at residences on the proposed site summed with an estimated 20 dB(A) night ambient noise.

Sound intensity in dB(A): combined noise level of turbines and predicted ambient (20 dB(A))
#Turbines	Max (450 ft from turbine)	Home 1	Home 2	Home 3	Home 4	Home 5
1	46.0	24.1	21.0	21.5	21.2	21.2
2	50.0	26.2	24.1	22.5	22.1	22.1
3	50.0	28.6	26.2	24.1	23.5	23.0
4	50.0	29.5	27.0	24.8	24.8	24.1

It can be observed from Table 6 that the effect of sound propagation from the wind turbines in the residential areas would be noticeable under these conditions when summed with an estimated 10 dB(A) ambient noise at night. The sound intensity levels of the wind turbines are much higher than the 10 dB(A) ambient noise compared to the other scenarios. The effect observed at the residences from the noise generated by the wind turbines would be an increase in the range of 6.2–19.1 dB(A) which would be noticeable to a human observer. An increase of this magnitude would seem to be 0.62–1.91 times louder in perceived loudness (AWS Truepower, 2017).

Table 6.

Predicted sound intensity at residences on the proposed site summed with an estimated 10 dB night prediction.

Sound intensity in dB(A): combined noise level of turbines and predicted ambient (10 dB(A))
#Turbines	Max (450 ft from turbine)	Home 1	Home 2	Home 3	Home 4	Home 5
1	46.0	22.3	15.5	17.0	16.2	16.2
2	50.0	25.1	22.3	19.5	18.6	18.6
3	50.0	28.1	25.1	22.3	21.3	20.4
4	50.0	29.1	26.1	23.2	23.2	22.3

In conclusion, the noise produced by the wind turbines is predicted to be negligible on all accounts when observed from the residences unless it is an extremely quiet night (10–20 dB(A) or equivalent to the sound of falling leaves to whispering), an increase in the 1–19 dB(A) range could be expected under these conditions which would be perceived as louder than pure ambient noise by a human observer. This would be a very rare occasion for ambient noise to reach these very low levels. However, the results presented are a conservative estimate of actual sound levels. The model does not take into account hills or mountains, vegetation, or other obstacles to sound propagation such as buildings and other constructions near the site. It should be noted that the ambient noise levels of common residential areas will have also diurnal and seasonal variations that need to be considered for a more detailed prediction of noise level.

Discussion

This experimentation began with ambient sound measurements at a potential wind energy generation site. Table 1 summarizes the data obtained in the data collection stage. The data comprised 10 2-min measurements of ambient noise data as well as wind speed acquired at the locations for potential wind turbines. The average wind speed varied in the range of 2.20–4.60 m/s, the minimum was in the range of 0.30–1.60 m/s, and the maximum was in the range of 5.60–12.10 m/s. The ambient noise level was in the range of 42.5–52.1 dB(A) with peak values in the range of 47.1–65.7 dB(A). These data are significant because the noise data were observed during a time of acceptable wind levels for power generation.

The noise model was based off of four Bergey Excel 10 turbines with other inputs as specified previously. The noise model included simplifying assumptions as follows: it does not account for terrain features and assume that ground is flat or constantly sloping, does not take into account vegetation or buildings, and reflections of noise are ignored. The model stated that the homes in the area would all receive noise propagation less than 30 dB(A). The maximum value of noise was calculated at 450 ft from the turbine and was no greater than 50 dB(A) similar to the ambient noise levels recorded at the site with only ambient noise sources.

A summation of the ambient noise values and the predicted noise values was performed. It was noted that when the measured ambient noise levels were summed with the expected noise values of the turbine, the increase in loudness was negligible. A worst-case scenario 450 ft away from the turbine with the highest peak ambient noise measurements resulted in an increase of 6.5–7.6 dB(A) which would be noticed by a human observer. It was further investigated how in the event of a quiet night in the range of 10–30 dB(A) would that affect the noise levels. It was found that an extremely quiet night between 10 and 20 dB(A) (equivalent to the sound of falling leaves to whispering) would produce the right conditions for the wind turbines to be noticed by the observer at the location. Considering vegetation, hills, buildings, and other obstacles, it is still unlikely that the noise created by the turbines would disturb residents close in the area of a potential wind farm.

The maximum wind turbine noise determined by the noise map generated by Openwind was about 50 dB(A) at a distance of 450 ft (137 m). The predicted noise level was similar to that of Figure 1 where the expected noise level would be in the range of 50 dB(A) and 40 dB(A) at a distance of 100 m and 400 m (General Electric, 2014). After 400 m, the wind turbine noise would have been unnoticed by a neighboring observer as the ambient noise recorded at the site was too great.

Conclusion

In this work, ambient noise data were collected from a potential wind energy generation site as well as a noise propagation model for wind turbines at that location was created. It was found that during a period of time where wind levels were appropriate for power generation the noise generated by the turbines would be outweighed by the ambient noise present at residences close to the site. The noise produced by the wind turbines was predicted to be negligible on all accounts when observed from the residences unless it is an extremely quiet night (10–20 dB(A) or equivalent to the sound of falling leaves to whispering) and the 1–19 dB(A) range of increase could be expected under these conditions which would be perceived as much louder than pure ambient noise by a human observer. This would be a very rare occasion for ambient noise to reach such low levels. However, the results presented are a conservative estimate of actual sound levels. The model does not take into account hills or mountains, vegetation, or other obstacles to sound propagation such as buildings and other constructions near the site.

There are several potential areas of improvement. The primary improvement would be employing a more comprehensive software, such as SPreAD-GIS, in order to produce a more comprehensive noise model; however, this comes with increased costs. Secondarily, more ambient noise data and weather data could be collected at various times in order to get a more accurate dataset to test. Another major improvement to the study would be to obtain wind turbine sound propagation data, using the sound level meter, in order to compare it to the predicted values. This would allow one to truly determine the accuracy of the model.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

AWS Truepower (2017) Openwind® user manual. Available at: https://www.awstruepower.com/assets/2016/12/Openwind_user_manual_2017.pdf

Aylor

(1972) Noise reduction by vegetation and ground. Journal of Acoustical Society of America 51: 197–205.

Bergey WindPower (2010) Acoustic characteristics of the Bergy Excel-S 10 kW wind turbine. Available at: http://bergey.com/documents/2012/03/excel-10-bwc-turbine-acoustics-data.pdf

Bolt

Beranek

Newman (1973) Fundamentals and abatement of highway traffic noise. Report no. PB-222–703, June. Prepared for the Federal Highway Administration. Cambridge, MA. Available at: https://ntrl.ntis.gov/NTRL/dashboard/searchResults/titleDetail/PB222703.xhtml

Brüel & Kjær (2000) Environmental Noise Booklet. Brüel & Kjær. DK-2850 Naerum, Denmark. Available at: https://www.bksv.com/media/doc/br1626.pdf

Fang

Ling

(2003) Investigation of the noise reduction provided by tree belts. Landscape and Urban Planning 63: 187–195.

Filippeli

(2013) Wind project design and optimization: Introduction, validation, and applications of Openwind. In: 2nd NREL wind energy systems workshop, Broomfield, CO, 30 January.

General Electric (2014) How loud is a wind turbine? GE Global Research and National Institute of Deafness and Other Communication Disorders (NIDCD part of NIH), Niskayuna, NY. Available at: https://www.ge.com/reports/post/92442325225/how-loud-is-a-wind-turbine/

Health Canada (2014) Wind Turbine Noise and Health Study: Summary of Results. Government of Canada. Ottawa, Ontario. Available at: https://www.canada.ca/en/health-canada/services/environmental-workplace-health/noise/wind-turbine-noise/wind-turbine-noise-health-study-summary-results.html

10.

ISO 9613-2:1996 (1996) Attenuation of sound during propagation outdoors. Part 2: General method of calculation.

11.

Kluijver

Stoter

(2003) Noise mapping and GIS: Optimizing quality and efficiency of noise effect studies. Computers, Environment and Urban Systems 27: 85–102.

12.

Kwong

Zhang

Romero

et al . (2012) Wind farm layout optimization considering energy generation and noise propagation. In: Proceedings of the ASME 2012 international design engineering technical conferences & computers and information in engineering conference, Chicago, IL, 12–15 August.

13.

Lamancusa

(2009) Outdoor noise propagation. Available at: http://www.mne.psu.edu/lamancusa/me458/10_osp.pdf

14.

Larsson

Ohlund

(2014) Amplitude modulation of sound from wind turbines under various meteorological conditions. Journal of Acoustical Society of America 135(1): 67–73.

15.

Michaud

Feder

Keith

et al . (2016) Exposure to wind turbine noise: Perceptual responses and reported health effects. Journal of Acoustical Society of America 139(3): 1443–1454.

16.

National Physical Lab (NPL) (2006) Guide to Predictive Modelling for Environmental Noise (NPL Acoustic Technical Guides). NPL. Teddington, England. Available at: http://www.npl.co.uk/upload/pdf/guide-to-predictive-modelling-env-noise-assessment.pdf

17.

Pamanikabud

Tansatcha

(2003) Geographical information system for traffic noise analysis and forecasting with the appearance of barriers. Environment Modelling and Software 18: 959–973.

18.

Reed

Boggs

Mann

(2012) A GIS tool for modeling anthropogenic noise propagation in natural ecosystems. Environmental Modelling & Software 37: 1–5.

19.

Saavedra

Samanta

(2015) Noise and vibration issues of wind turbines and their impact: A review. Wind Engineering 39(6): 617–626.

20.

Sengpiel

(n.d.) Adding acoustic levels of sound sources. Sengpiel Audio. Available at: http://www.sengpielaudio.com/calculator-spl.htm

21.

Tang

Wang

(2007) Influences of urban forms on traffic-induced noise and air pollution: Results from a modelling system. Environmental Modelling & Software 22: 1750–1764.

22.

US Department of Energy (2015) Wind vision: A new era for wind power in the United States. Available at: https://www.energy.gov/sites/prod/files/WindVision_Report_final.pdf