Conference PaperPDF Available

Infrasound and Low-Frequency Noise from Wind Turbines

Authors:

Abstract and Figures

Infrasound, low-frequency noise (ILFN) and amplitude modulation of the noise are known to disturb some residents living near wind farms. However, the mechanisms responsible for ILFN and amplitude modulation are not well understood. In an attempt to shed some light on these mechanisms, acoustic measurements were taken in the close vicinity of a wind farm, at residences located two or more kilometres from the nearest turbine in a wind farm and in an anechoic chamber using a scale-model, electrically-driven, wind turbine. The measured spectra reveal distinct peaks at the blade-pass frequency and harmonics, and the characteristics of these peaks are remarkably similar for field and laboratory measurements, indicating that the zero mean flow simulation is a good representation of an actual wind turbine. Near field acoustic holography measurements on the scale-model turbine confirm that tonal components at the blade-pass frequency and harmonics are generated as a result of blade-tower interaction, suggesting that it is likely to be an important mechanism of infrasound generation for industrial wind turbines. Inaccuracies in the assumed location of noise sources on a wind turbine affect the accuracy of community noise predictions. This is because the source height affects the distance from the turbine beyond which sound rays arrive at the receiver having been reflected from the ground more than once, thus reducing the attenuation with distance from the turbine.
Content may be subject to copyright.
Infrasound and low-frequency noise from wind
turbines
Colin Hansen, Branko Zajamˇ
sek and Kristy Hansen
Abstract Infrasound, low-frequency noise (ILFN) and amplitude modulation of the
noise are known to disturb some residents living near wind farms. However, the
mechanisms responsible for ILFN and amplitude modulation are not well under-
stood. In an attempt to shed some light on these mechanisms, acoustic measure-
ments were taken in the close vicinity of a wind farm, at residences located two or
more kilometres from the nearest turbine in a wind farm and in an anechoic cham-
ber using a scale-model, electrically-driven, wind turbine. The measured spectra
reveal distinct peaks at the blade-pass frequency and harmonics, and the character-
istics of these peaks are remarkably similar for field and laboratory measurements,
indicating that the zero mean flow simulation is a good representation of an ac-
tual wind turbine. Near field acoustic holography measurements on the scale-model
turbine confirm that tonal components at the blade-pass frequency and harmonics
are generated as a result of blade-tower interaction, suggesting that it is likely to
be an important mechanism of infrasound generation for industrial wind turbines.
Inaccuracies in the assumed location of noise sources on a wind turbine affect the
accuracy of community noise predictions. This is because the source height affects
the distance from the turbine beyond which sound rays arrive at the receiver having
been reflected from the ground more than once, thus reducing the attenuation with
distance from the turbine.
1 Introduction
One of the drawbacks of wind energy is that turbines generate sufficient noise to
result in adverse reactions from some nearby residents. The noise that causes prob-
lems is infrasound and low-frequency noise (ILFN) and the associated noise levels
are highly variable as a function of time, due to meteorological factors, blade loading
Colin Hansen
The University of Adelaide, North Terrace, Adelaide, South Australia, 5000
e-mail: colin.hansen@adelaide.edu.au
Branko Zajamˇ
sek
University of New South Wales, Kensington, Sydney, NSW, 2052
e-mail: b.zajamsek@student.unsw.edu.au
Kristy Hansen
Flinders University, Bedford Park, Adelaide, South Australia, 5042
e-mail: kristy.hansen@flinders.edu.au
1
2 Colin Hansen, Branko Zajamˇ
sek and Kristy Hansen
variations, directivity variations as the turbine blades rotate, and interaction between
the sound from two or more turbines. These characteristics make wind turbine noise
more annoying at comparable levels than other sources such as industrial and trans-
portation noise. Also, low frequency noise and infrasound are less attenuated over
large propagation distances and can penetrate buildings more readily than mid- to
high-frequency noise.
In the future, the size of wind turbines is expected to increase since larger turbines
are more efficient energy generators. Unfortunately, this will lead to an increase in
the thrust force acting on the blades as well as an increase in the infrasound and
low-frequency noise due to higher hub heights, which introduce larger variations
in blade loading during a revolution. Larger wind turbines also rotate more slowly
due to limitations in the allowable blade-tip velocity and therefore mechanical noise
is expected to decrease in frequency. A downward shift of the noise spectrum by
approximately one-third of an octave has already been observed in the transition
from small to large wind turbines (Møller and Pedersen, 2011). If infrasound and
low frequency noise are to be reduced, it is important that the corresponding noise
generation and propagation mechanisms are well understood.
2 Effect of wind turbine noise on people
Many residents who live in close proximity to wind farms report annoyance and
sleep disturbance, even when the measured noise levels are relatively low. One rea-
son for this is that wind turbines are often located in rural areas where background
noise levels are very low, particularly at night time. The contrast between ambient
noise and noise due to wind farm operation is also exacerbated during the evening
and night-time due to stable atmospheric conditions (Van den Berg, 2004). During
these conditions, the wind turbines continue to operate, while the wind speed at resi-
dences is negligible, so corresponding background noise levels are often well below
20 dBA.
Stable atmospheric conditions are also characterised by high wind shear, which
has been suggested as a major factor responsible for the amplitude modulation of
wind turbine noise (RenewableUK, 2013). Residents living in the vicinity of wind
farms describe the associated noise as “thumping” (Van den Berg, 2004) or “rum-
bling” (Hansen et al., 2014) in character, indicating the presence of low frequency,
time varying noise. The difference between the audibility threshold and perceived
loudness is small when the noise is dominated by low-frequencies (Møller and Ped-
ersen, 2004) and therefore if a low-frequency noise is amplitude modulated as well
as being above the normal hearing threshold, it is likely to be annoying to many
people.
Some people living near wind farms also report symptoms of motion sickness,
including ear pressure, headache, nausea, dizziness and vertigo. It is possible that
these symptoms are related to exposure to infrasound, which is primarily generated
as a result of blade-tower interaction. The amplitude of vertical motion in the 2 Hz
Infrasound and low-frequency noise from wind turbines 3
to 4 Hz range required to produce seasickness in sensitive individuals corresponds
to an atmospheric pressure variation that is similar to the levels of acoustic pressure
variation experienced by people living in the vicinity of wind farms (Dooley, 2013).
The regular, periodic nature of these variations, or the symmetry of the maximum
and minimum values compared to the mean, may explain why similar levels of
random environmental infrasound do not result in motion sickness symptoms in
sensitive individuals. Another explanation for adverse health symptoms reported by
people living near wind farms is that infrasound stimulates the outer hair cells of the
human ear at levels below the audibility threshold (Salt and Lichtenhan, 2014). This
results in information transfer via pathways that do not involve conscious hearing,
which may lead to sensations of fullness, pressure or tinnitus, or lead to no sensation
at all. Salt0s work is by no means generally accepted and the subject of infrasound
is still surrounded by controversy and ongoing debate about its significance. One or
more of the above mentioned symptoms can also be attributed to excessive exposure
to low frequency noise.
3 Infrasound and low-frequency noise generating mechanisms
Wind turbines generate infrasound and low-frequency sound, as well as mid-and
high-frequency sound. However, as low-frequency sound and infrasound are not at-
tenuated by atmospheric absorption or reflection from the ground, they dominate
the noise spectrum at residences more than one or two kilometres from the wind
farm. These sounds are either aeroacoustic or mechanical in origin and can be ei-
ther tonal or broadband in nature, depending on the generation mechanism. Aero-
dynamic infrasound and low-frequency noise originate as a result of changes in
the aerodynamic force acting on the blades as they rotate. This can be caused by
wind shear, atmospheric turbulence, cross-wind conditions, or interaction with the
disturbed flow between the blade and tower as the blade passes the tower. The dis-
turbed flow is a result of the flow streamlines having to deviate to negotiate the tower
obstacle. It has been shown analytically that this phenomenon, illustrated in Fig. 1,
is responsible for variations in the blade loading and hence generation of tonal com-
ponents at the blade-pass frequency and harmonics (Doolan et al., 2012). In Fig. 1, it
can be seen that the potential flow over the support tower creates a region of reduced
velocity in front of the support tower. When the blade passes through that region,
the angle of attack varies which changes the lift force and subsequently produces
sound.
Interaction between inflow turbulence and the rotating turbine blades also leads
to aerodynamic loading fluctuations, which are responsible for the generation of low
frequency, broadband noise. The level of inflow turbulence, and hence noise gener-
ation, varies with atmospheric conditions and also with the locations of turbines
relative to the wake of other turbines. The resulting far field noise from this inter-
action is strongly dependent on the inflow turbulence spectrum. Large eddies in the
inflow cause a change in the aerodynamic force over the whole blade surface and
4 Colin Hansen, Branko Zajamˇ
sek and Kristy Hansen
rotor blade passing
in front of tower
streamlines
around tower
wind
tower
Fig. 1 Blade-tower interaction
consequently noise radiation has a dipole directivity pattern and the peak radiating
frequency is low, typically below 20 Hz for a modern turbine (Doolan et al., 2012).
A general theory, explaining sound radiation due to unsteady blade loading was
first proposed by (Tyler and Sofrin, 1962) for the case of turbo-machinery and
was later expanded to an open rotor via the concept of pressure modes (Wright,
1976). More recently, this theory has been applied to wind turbines by (Dooley and
Metelka, 2014). According to the theory, any rotor can be thought of as a system of
rotating forces whose magnitude varies with rotor azimuthal position. These rotat-
ing forces can be simulated by a point source on the blade at a location, re, which
is usually 0.8 times the blade span from the centre of the rotor plane. At any instant
in time, these pressure variations are in-phase at each blade location, but they are
not sinusoidal and thus can be described in terms of harmonics of the blade pass
frequency, which rotate with the blade — hence their name, ”spinning modes”. The
mode order indicates the number of 2πchanges during one complete blade revolu-
tion.
Each pressure harmonic radiates noise in all directions, but the radiation to the
side of the rotor is slightly greater. The tonal peaks in the radiated noise spectrum
are broadened a little by the Doppler shift as a result of the motion of the blades.
The broadening is dependent on the observer location and increases as the observer
moves from on-axis to the side of the rotor plane.
When a rotating wind turbine blade passes the tower, the blade experiences a
loading excursion that adds to the other pressure variations and increases the sound
radiation in the directions normal to the rotor plane (see Fig. 2). Although the blade
experiences the change in loading across the whole span, it can be simulated by
an equivalent point loading at an effective radius, re. The sinusoidal signal at the
top of the figure represents both the sound pressure variation as a function of time
for a specified location as well as the sound pressure distribution around the circle
containing the equivalent point sources at any instant in time. The sound field at an
arbitrary observer is a function of the path difference x1and x2(see Fig. 2), and
also linear phase variations around the spinning mode at p1and p2. It is thus the
modal phase and path difference that define the frequencies at which constructive
and destructive interference occur at any particular location. This process produces
characteristic interference lobes that radiate in specific directions, and which appear
at higher blade-pass frequency harmonics.
Infrasound and low-frequency noise from wind turbines 5
spinning mode
phase variation p
2
p
1
time
Dt= one revolution
observer
x1
x2
t0
equivalent
point loading
s
t
e
a
d
y
l
o
a
d
i
n
g
r
e
loading excursion
Fig. 2 Principle of spinning modes. The dashed line represents the effective radius of the equivalent
point source and the solid line indicates the blade loading, showing a step excursion as a result of
blade tower interaction
Loading unsteadiness caused by turbulent inflow is random in nature and is
highly influenced by meteorological conditions, which affect the level of atmo-
spheric turbulence as well as the characteristics of the wake generated by upstream
turbines. Time-varying noise characteristics at the receiver can also be caused by
constructive/destructive reinforcement of sound waves arriving at a residence from
two or more wind turbines. The regions in which constructive/destructive interfer-
ence occur are defined by the relative phase of the incoming sound waves, which
varies with wind direction and blade rotation rate, leading to a time dependent am-
plitude variation of sound at any given location. This phenomenon is accentuated in
stable atmospheric conditions which have low associated atmospheric turbulence,
resulting in different turbines rotating at more similar speeds with reduced fluctua-
tions (Van den Berg, 2005).
In addition to the random amplitude variations discussed above, periodic varia-
tions in the loudness (or amplitude modulation) of wind turbine noise also occur,
with the frequency of variation generally equal to the rate at which blades pass the
tower. In cases where local stall occurs, the trailing edge noise spectra shifts to lower
frequencies and the resulting time-varying noise has been described as “thumping”
(RenewableUK, 2013). In contrast to attached flow trailing edge noise, which shows
most significant amplitude modulation in the crosswind direction at large distances
(Oerlemans and Schepers, 2009), amplitude modulation associated with stall noise
is highest in the upstream and downstream directions (RenewableUK, 2013).
Another noise source that is subject to amplitude modulation is gearbox noise.
This can occur when the planetary gear mesh noise is amplitude modulated by the
blade/planet-pass frequency and this phenomenon appears to be responsible for the
“rumbling” noise that has been measured in the vicinity of the Waterloo wind farm
in South Australia (Hansen et al., 2014).
6 Colin Hansen, Branko Zajamˇ
sek and Kristy Hansen
4 Infrasound and low frequency noise propagation
In general, sound spreads spherically from a given source, resulting in an attenu-
ation rate of 6 dB/doubling of distance. However, in the downwind direction, the
wind speed gradient causes the sound waves to bend towards the ground, reducing
the attenuation of noise in the downwind direction. Vertical temperature gradients
also give rise to sound speed gradients, but the effect of the wind speed gradient is
generally dominant in the propagation of sound from wind turbines. The effect of
a downward refracting atmosphere is that beyond a certain distance, more than one
ground reflected ray will arrive at the receiver. The first ground reflected ray will
have been reflected from the ground once, the second ground reflected ray will have
been reflected from the ground twice, etc.
The distance from the sound source at which sound will begin to attenuate at a
rate less than 6 dB per doubling of distance is a function of the source height as
well as the strength of the atmospheric sonic velocity gradient (and the consequent
strength of the downward refraction), as this determines the distance from the source
at which the receiver will experience the arrival of more than one ground reflected
wave. The effect only occurs at low-frequencies for which losses due to ground
reflection and atmospheric absorption are negligible. The path length of reflected
sound rays increases as the number of reflections increase and so the amplitude of
the sound pressure (relative to the direct, non-reflected ray) arriving at the receiver
from each reflected ray decreases as the number of reflections increases, until after a
certain number of reflections, a particular ray is no longer an important contributor
to the sound pressure level at the receiver. It has been shown that the sum of all re-
flected rays results in a decay of the sound field of approximately 3 dB per doubling
of distance (Willshire Jr and Zorumski, 1987), which is why it is often referred to
as “cylindrical spreading”. Of course there is a gradual transition over both decreas-
ing frequency and increasing distance from the source, during which the decay rate
changes from 6 dB to 3 dB per doubling of distance.
5 Experimental measurements of blade-tower interaction
To develop an understanding of how blade-tower interaction contributes to wind
turbine infrasound and low-frequency noise generation, experimental work was un-
dertaken in the anechoic chamber at the University of Adelaide using the scale-
model wind turbine shown in Fig. 3a. The anechoic chamber has dimensions of
4.79×3.9×3.94 m. The blades on the wind turbine model had a symmetrical airfoil
shape, they were mounted at 0angle of attack and they were driven by an electric
motor in conditions of zero net flow. Although this experimental configuration does
not replicate the effect of the reduced velocity field in-front of a turbine tower, which
is the major source of unsteady aerodynamic loading, the blade-tower interaction is
still adequately re-created via the interaction of the support tower with the poten-
tial velocity field surrounding the blade. The 0angle of attack is ideal for studying
Infrasound and low-frequency noise from wind turbines 7
BTI as the resulting noise spectrum is not affected by noise generated by thrust and
torque loading. However, for zero angle of attack symmetrical blades, the pressure
resulting from the displacement of fluid caused by the advancing blades, results in
pressure variations in the air that it passes through, generating ”thickness noise” that
adds to the BTI noise as the blades pass the tower.
(a) Scale-model wind turbine (b) Anechoic chamber measurement set-up
Fig. 3 Scale-model wind turbine. Parts are: (1) NACA 0012 airfoil (tripped at 10% chord length,
70 mm chord, 450 mm span), (2) Slip ring (24 channels), (3) Torque sensor, (4) AC driver and (5)
Support tower (70 mm outer diameter)
The rotor model was driven at 900 RPM, giving a blade tip speed of 47 m/s and
a blade-pass frequency of 45 Hz. The rotor plane was located 70 mm away from the
closest point on the tower. This distance is comparable to one blade chord length and
is found to be a good scaled representation of the blade-to-tower distance on a utility
scaled wind turbine. The blade-tower interaction was investigated using point micro-
phone measurements and a 1.5 m diameter circular microphone array containing 64
GRAS 40PH phase and magnitude matched microphones on a plane parallel to the
rotor plane and 10 cm from it. The data obtained with the microphone array was pro-
cessed using statistically optimised near-field acoustic holography (SONAH) (Hald,
2009) to obtain low-frequency sound source visualisation in the rotor plane.
Figure 4 shows the power spectral density of the blade-tower interaction noise
measured on an axis perpendicular to the rotor plane centre. As can be seen, the
spectrum consists of a series of “peaks” which are harmonically related to the blade-
pass frequency of 45 Hz. It is also evident that the noise magnitude is inversely pro-
portional to the distance dbetween the rotor plane and the support tower, which is in
accordance with previous research (Madsen, 2010). This indicates that the support
tower does have an effect on the blade loading via interaction with the potential field
8 Colin Hansen, Branko Zajamˇ
sek and Kristy Hansen
surrounding the blade and the noise production mechanism is identical to the one
experienced by full-size wind turbines.
45 90 1000
30
40
50
60
70
Frequency [Hz]
PSD [dB re 20µPa/Hz]
d = 70
d = 40
d = 20
Fig. 4 Effect of the distance, d, between the support tower and the blade on the blade-tower inter-
action noise at 0angle of attack
Figure 5 shows the directivity pattern for 45, 135, 360 and 450 Hz. These fre-
quencies correspond to the 1st, 3rd , 8th and 10th harmonics of the blade-pass fre-
quency, respectively. It can be seen that sound radiation at 135 Hz is omnidirec-
tional, assuming symmetry across the rotor plane. For 360 and 450 Hz, a significant
reduction in the sound pressure level is observed at angles <30and >150. For
45 Hz, constructive interference is seen on each side of the rotor plane (between 0
and 30and between 150and180) and destructive interference is seen at 60. The
constructive and destructive interference at any particular harmonic frequency is a
result of the spinning mode acoustic interference in the rotor plane.
Fig. 5 Blade-tower interaction noise directivity pattern for selected frequencies. Dashed circular
contours denote sound pressure level in dB and are 5dB apart
Infrasound and low-frequency noise from wind turbines 9
The SONAH sound field visualisation shown in Figure 6 indicates that the ma-
jority of energy at the blade-pass frequency, f= 45 Hz, in Fig. 6a originates at the
support tower position. Radiation at higher blade-pass frequencies, f= 135, 360 and
450 Hz in Figs. 6b, 6c and 6d, respectively, is no longer highly concentrated at the
support tower location but is rather spread out in the lower or middle part of the rotor
plane. The most likely reason for the peak sound source being located slightly to the
left of the tower in Fig. 6a, is that the peak loading occurs close to the leading edge
(Wright, 1971). Since the blades are rotating in a clockwise direction, the leading
edge is thus on the left side of the support tower when the blade is directly in front
of the tower. The appearance of two sources in Fig. 6a, separated by the microphone
spacing between the nine microphones in the outer circle of the array is an artefact
of the SONAH process. This should be interpreted as a single source. Similarly the
separate sources shown on the outer circle in 6b and 6c should be interpreted as a
single source with intensity varying from a maximum at the bottom to a minimum
at the top of the rotor plane.
(a) f= 45 Hz (b) f= 135 Hz
(c) f= 360 Hz (d) f= 450 Hz
Fig. 6 Near-field acoustic holography. The white circle indicates the outer edge of the rotor plane
and the thin white vertical line represents the support tower
The spread of the apparent source location at higher frequencies is an effect of
the spinning modes and the frequency at which it begins to occur is a function of
how close the observation plane is to the rotor plane.
10 Colin Hansen, Branko Zajamˇ
sek and Kristy Hansen
6 Field measurements of propagation
Continuous indoor and outdoor measurements were carried out for periods of ap-
proximately one week at three residences located near the Waterloo wind farm,
which is made up of 37 operational turbines. Outdoor measurements were made
using a G.R.A.S. type 40AZ microphone with 26CG preamplifier with an elec-
tronic noise floor of 16 dB(A) and a low frequency linear response down to 0.5 Hz.
Hemispherical secondary windshields of 450 mm diameter were used to minimise
wind-induced noise, and they were designed to be consistent with the IEC-61400-
11 (2012) standard. Wind speed and direction were measured at heights of 1.5 m
and 10 m using Davis Vantage Vue and Vantage Pro weather stations, respectively.
The weather measurements were collected in 5-minute intervals and then the 10-
minute average was calculated during post-processing. Wind speed and direction at
hub height were measured using a SODAR unit which was located on the ridge-top
in the gap between the Northern and Southern wind turbine group shown in Fig. 7.
The wind farm operator also provided hub height wind speed and direction for the
period in which data for House 1 and House 2 were collected.
The location of the residences relative to the wind farm is shown in Fig. 7. House
1 is situated 3.5 km from the nearest wind turbine, which is near the centre of the
main turbine group. The downwind direction from the closest wind turbine to the
residence is 88. House 2 is 8.7 km from the nearest wind turbine which is the north-
ernmost turbine of the main group. The downwind direction from the closest wind
turbine to the residence is 268. House 3 is 3.3 km from the nearest wind turbine,
which is the southernmost turbine in the smaller northern group. The downwind
direction from the closest wind turbine to the residence is 300. The wind speeds
and directions at 1.5 m, 10 m and hub height are presented in Table 1 along with
the overall power output of the wind farm and stability factor for the measurements
shown in Fig. 7. For each measurement, the residence was located in a downwind
direction (±45)from the nearest wind turbine. It can be seen that the most stable
conditions occurred during the measurements at House 3 according to the definition
of stability factor, m.
m=log10 vh/vre f
log10 h/hre f (1)
where vhis the velocity at 80 m hub height, vref is the velocity at 10 m, his the
height of wind turbine hub above a given residence and href is 10 m.
Table 1 Wind and turbine operational conditions for the data shown in Fig. 8. Wind speed at hub
height is for the turbine nearest to the residence
Wind speed (m/s) Wind direction () Power Stability
Description 1.5 m 10 m hub height 1.5 m 10 m hub height Output (%) factor
H1 1.8 3.4 10.5 135 135 133 44 0.4
H2 1.6 2.9 8.7 293 281 305 56 0.4
H3 0 0.4 10.4 - 22.5 287 53 1.1
Infrasound and low-frequency noise from wind turbines 11
-4 -2 0 2 4 6 8 10
0
2
4
6
8
10
12
14
16
Distance from origin [km]
Distance from origin [km]
Wind turbine
H1
H2
H3
Fig. 7 Field measurement locations
Figure 8 shows the measured infrasound in the vicinity of the Waterloo wind
farm at the three different locations. The harmonics of the blade-pass frequency of
0.8 Hz extend up to at least the 8th order at all three residences, including the one
located 8.7 km away from the wind farm. The relative amplitude of the peaks is con-
sistent with their distances from the wind farm, and maximum attenuation occurs at
H2. Propagation to H2 would also be affected by a ridge which is located between
the residence and the wind farm and runs parallel to the line of wind turbines. The
fundamental blade-pass frequency is not consistently visible in the outdoor spec-
tra and this is attributed to the presence of wind-induced noise at some locations.
The wind-induced noise raises the level of the broadband infrasound, particularly
at very low frequencies and this causes masking of the peak at 0.8 Hz at H1 and
H2. The blade-pass frequency harmonics can also be observed in the indoor results
but the magnitude of the peaks below 6 Hz is not representative of the true signal
due to the roll-off characteristic of the B&K 4955 microphones used for the indoor
measurements.
Comparison between Figs. 4 and 8 indicates that there are striking similarities
between the experimental and measured results. In both cases, the amplitude of the
infrasonic peaks is significantly higher than the sound pressure levels at adjacent
frequencies and the harmonics extend to similar orders. For the blade-tower spacing
in Figure 4 of d=70 mm, which represents the spacing for an industrial wind
turbine based on scaling considerations, it can be seen that the relative amplitudes
of the first six harmonics are similar to those in Figure 8, which were measured on
an industrial turbine.
To investigate the effect of source height, receiver position and atmospheric ve-
locity profile on the distance at which multiple reflections first occur, a ray trac-
ing model was used and the results are shown in Fig. 9. Field measurement data
published in a wide range of studies (Ljunggren (1996); Sondergaard and Plovsing
12 Colin Hansen, Branko Zajamˇ
sek and Kristy Hansen
Frequency (Hz)
0.8 1 2 4 6 8
SPL (dB re 207Pa)
0
20
40
60
Frequency (Hz)
0.8 1 2 4 6 8 10
H1 H2 H3
Outside Inside
Fig. 8 Measured infrasound
(2005); Willshire Jr and Zorumski (1987); Bou´
e (2007)) are also plotted for compar-
ison. Differences between the model and measurements by others can be attributed
to the vertical atmospheric sonic velocity profile associated with the measurements.
Since this information was not available from the literature, the profile used in this
study was specified according to SODAR data measured under stable atmospheric
conditions (Hansen et al., 2015) and according to the logarithmic velocity profile
used in the Nord2000 propagation model (Plovsing and Kragh (2006) and Plovs-
ing (2007)). It is evident that the assumed vertical sonic velocity profile has a large
impact on the model output, particularly at large propagation distances.
As indicated in Fig. 9, it is important to take into account the height of the source
when considering propagation of noise from a wind farm. Since noise at infrasonic
frequencies is attributed to blade-tower interaction, the source height for these fre-
quencies is the point at which this interaction noise is maximum. In the case of the
experiments presented in Sect. 5, the maximum noise occurred close to the blade-tip
as the blade passed the tower. It is possible that this point would vary for an indus-
trial wind turbine since the blades have significant twist and taper, which has a large
impact on the corresponding aerodynamic forces. Nonetheless, it is anticipated that
the source height for blade-tower interaction noise would be appreciably lower than
the hub height, which is generally assumed as the relevant source height in propaga-
tion analyses. Therefore, the distance from the source at which multiple reflections
first occur would be reduced. If the blade-tower interaction noise source is in fact
located at the blade-tip for a Vestas V90 3MW wind turbine model (installed at Wa-
terloo), the corresponding source height would be 36 m. According to Fig. 9, this
would reduce the distance at which multiple reflections first arrive at the ground by
0.5 - 1.3 km, depending on the relevant velocity profile. Therefore, due to the con-
tribution of an increased number of ray paths, the attenuation of noise at infrasonic
Infrasound and low-frequency noise from wind turbines 13
Fig. 9 Comparison between ray tracing results and published data for the distance from the source
at which multiple ground reflections begin to occur
frequencies would be less than would be predicted if the source were assumed to be
at hub height.
Determination of the relevant source height is also an important consideration for
the other aerodynamic and mechanical noise sources that were discussed in Sect. 3.
It is therefore important to determine the blade positions at which blade stall oc-
curs, how mechanical noise is radiated by structural components such as the blades
and tower and which blade location the noise due to inflow turbulence would be
maximum. The directivity of these noise sources is also important.
7 Conclusions
Comparison between the field measured noise data near a wind farm and labora-
tory measurements using a motor-driven model turbine in zero incident flow reveals
striking similarities in the infrasonic acoustic spectra. The accurate prediction of the
propagation of this noise is dependent on the source height that is used, which is
why the relative source height determined from laboratory measurements is a useful
input to noise prediction models. The source height together with the vertical wind
speed profile determine the distance from the source beyond which sound rays that
have experienced more than one ground reflection will arrive at the receiver, thus
resulting in increased noise levels over what would be expected due to a sound field
decay of 6 dB per doubling of distance.
14 Colin Hansen, Branko Zajamˇ
sek and Kristy Hansen
References
Bou´
e, M. (2007). Long-range propagation over the sea with application to wind turbine noise.
Technical report, KTH.
Doolan, C., Moreau, D. J., and Brooks, L. A. (2012). Wind turbine noise mechanisms and some
concepts for its control. Acoustics Australia, 40(1):7–13.
Dooley, K. A. (2013). Significant infrasound levels a previously unrecognized contaminant in land-
mark motion sickness studies. The Journal of the Acoustical Society of America, 134(5):4097.
Dooley, K. A. and Metelka, A. (2014). Acoustic interaction as a primary cause of infrasonic
spinning mode generation and propagation from wind turbines. volume 20, page 2272.
Hald, J. (2009). Basic theory and properties of statistically optimized near-field acoustical holog-
raphy. The Journal of the Acoustical Society of America, 125(4):2105–2120.
Hansen, K., Hessler, G., Hansen, C., and Zajamsek, B. (2015). Prediction of infrasound and low
frequency noise propagation for modern wind turbines: a proposed supplement to ISO 9613-2.
In Proceedings of WTN15, the sixth international conference on wind turbine noise.
Hansen, K., Zajamsek, B., and Hansen, C. (2014). Comparison of the noise levels measured in the
vicinity of a wind farm for shutdown and operational conditions. In InterNoise 2014.
IEC-61400-11 (2012). Wind turbine generator systems - part 11: acoustic noise measurement.
Technical report, International standard IEC 61400-11.
Ljunggren, S. (1996). Ljudutbredning kring havsbaserade vindkraftverk. resultat fraan en litter-
aturstudie. Technical report, Department of Civil and Architectural Engineering, KTH.
Madsen, H. A. (2010). Low frequency noise from wind turbines mechanisms of generation and its
modelling. Journal of Low Frequency Noise, Vibration and Active Control, 29(4):239–251.
Møller, H. and Pedersen, C. S. (2004). Hearing at low and infrasonic frequencies. Noise and
Health, 6(23):37.
Møller, H. and Pedersen, C. S. (2011). Low-frequency noise from large wind turbines. The Journal
of the Acoustical Society of America, 129:3727–3744.
Oerlemans, S. and Schepers, J. (2009). Prediction of wind turbine noise and validation against
experiment. International journal of aeroacoustics, 8(6):555–584.
Plovsing, B. (2007). Proposal for nordtest method: Nord2000–prediction of outdoor sound propa-
gation. DELTA Acoustics, Report AV, 1106(07).
Plovsing, B. and Kragh, J. (2006). Nord2000, comprehensive outdoor sound propagation model.
part 2: Propagation in an atmosphere with refraction. Delta Acoustics for Nordic Noise Group,
Report AV 1851/00.
RenewableUK (2013). Wind turbine amplitude modulation: research to improve understanding as
to its cause & effect. Technical report.
Salt, A. N. and Lichtenhan, J. T. (2014). How does wind turbine noise affect people. Acoust Today,
10:20–28.
Sondergaard, B. and Plovsing, B. (2005). Noise from offshore wind turbines. Environmental
Project 1016.
Tyler, J. M. and Sofrin, T. G. (1962). Axial flow compressor noise studies. SAE Trans., 70:309.
Van den Berg, G. (2004). Effects of the wind profile at night on wind turbine sound. Journal of
Sound and Vibration, 277:955–970.
Van den Berg, G. (2005). The beat is getting stronger: The effect of atmospheric stability on low
frequency modulated sound of wind turbines. Journal of Low Frequency Noise, Vibration and
Active Control, 24(1):1–24.
Willshire Jr, W. L. and Zorumski, W. E. (1987). Low-frequency acoustic propagation in high winds.
In NOISE-CON 87; Proceedings of the National Conference on Noise Control Engineering,
volume 1, pages 275–280.
Wright, S. (1971). Discrete radiation from rotating periodic sources. Journal of Sound and Vibra-
tion, 17(4):437–498.
Wright, S. (1976). The acoustic spectrum of axial flow machines. Journal of Sound and Vibration,
45(2):165–223.
... A number of authors have mentioned infrasound as a noise-related issue resulting from the wind farm operation [14][15][16][17][18][19][20], while some others are more concerned with lowfrequency sound [21]. In their comprehensive study of literature on the possible effects of wind farms on the health of the population, Freiberg et al. [22] conclude that when it comes to infrasound and low-frequency noise produced by turbines in wind power plants, major impacts on people include physical symptoms, general health issues, and affective impact (temper). ...
... Certain authors [15,16] point out that the levels of infrasound pressure produced by wind turbines are well below the generally accepted hearing threshold of a human ear, concluding that they cannot produce negative effects on human health. However, many authors [15][16][17][18] still maintain that just as there is no conclusive proof of the potentially detrimental effects of long-term exposure to the wind turbine-generated infrasound, especially in the low frequency range, there is no proof of the opposite either. Therefore, it is not possible to draw strict conclusions on the nature of the impact. ...
Article
Full-text available
Although wind farms have an undeniable beneficial impact on the environment, certain negative environmental implications do appear as a consequence of their operation. One of them is the production of noise. The wind farm noise values decrease with distance, so that at a certain point they are within the legally prescribed limits. This is the case for individual wind farms noise impact assessments. However, with two or more wind farms in the same area, there is a superposition of noise and a consequential change in the noise value. The focus of the paper is on the results of modeling noise propagation in space in the case of the cumulative impact of two neighboring wind farms. The results are modeled during the process of strategically assessing the environment so as to determine territorial impacts and make informed decisions about future development. The paper presents the strategic answer to the model of the spatial propagation of noise in cases of cumulative impact with a view to including the preventive protection principle in the planning of several adjacent wind farms.
... Relatively poor damping, wavelength and frequency, combined with the possibility of standing waves that can be formed in field specific conditions, as well as the possible resonance phenomenon lead to objective difficulties in the unambiguous and adequate location of the source of infrasound generation. An important element is also related with the need to take into account the acoustic background during the measurements of infrasound emitted by wind turbines, the level of which, in many cases, may be close to the useful signals [4][5][6][7][8]. In particular, this applies to the conditions when wind speeds with values above 12-15 m/s occur during infrasound recording. ...
... The dominant frequency bands in the range from 2 to 4 Hz are visible on the scalograms obtained on the basis of the study. There is also a noticeable increase in power for frequencies in the range of around (6)(7)(8) Hz and from about 12 to 16 Hz. We can also note that the highest values in the spectra occur for infrasound frequencies. ...
Article
Full-text available
The study reported in this paper is concerned with areas related to developing methods of measuring, processing and analyzing infrasound noise caused by operation of wind farms. The paper contains the results of the correlation analysis of infrasound signals generated by a wind turbine with a rated capacity of 2 MW recorded by three independent measurement setups comprising identical components and characterized by the same technical parameters. The measurements of infrasound signals utilized a dedicated measurement system called INFRA, which was developed and built by KFB ACOUSTICS Sp. z o.o. In particular, the scope of the paper includes the results of correlation analysis in the time domain, which was carried out using the autocovariance function separately for each of the three measuring setups. Moreover, the courses of the cross-correlation function were calculated separately for each of the potential combinations of infrasound range recorded by the three measuring setups. In the second stage, a correlation analysis of the recorded infrasound signals in the frequency domain was performed, using the coherence function. In the next step, infrasound signals recorded in three setups were subjected to time-frequency transformations. In this part, the waveforms of the scalograms were determined by means of continuous wavelet transform. Wavelet coherence waveforms were calculated in order to determine the level of the correlation of the obtained dependencies in the time-frequency domain. The summary contains the results derived from using correlation analysis methods in the time, frequency and time-frequency domains.
... As expected and already found by Klein et al. [4] the SPL is increased when deformation is considered. The main reason for that was identified as the decreased blade-tower distance [22]. Adding turbulence to the inflow further increases the SPL but only in upstream and downstream direction. ...
... In fact, human hearing does not stop at 20 Hz, but the perception of pitch is lost and sensitivity lowers as the frequency drops. According to various sources, the range of infrasound frequencies is from 0.1 to 20 Hz and also includes audible low frequencies [7][8][9][10][11][12]. ...
Article
Full-text available
The development of wind energy and the increasing number of installed wind turbines make it necessary to assess them in terms of the nuisance of the emitted infrasound noise generated by such devices. The article presents the results of measurements and analyses of infrasound emitted during the operation of wind turbines installed in various locations in Poland. Comparative analysis of noise levels in the infrasound and audible range has shown that acoustic energy is mainly in the low and infrasound frequency range, and the measured levels depend significantly on the weighting curves used. On the basis of the results, it was confirmed that the sound pressure level of infrasound signals emitted by the operation of high-power wind turbines, regardless of wind velocity, weather conditions, design solutions of turbines, operating time, rated capacity, does not exceed the criteria specified in the applicable legislation dealing with the assessment of infrasound noise on the working environment.
... Problems that have stood out in importance in practice so far relate to the acoustic impact of noise and sleep disturbance to which local populations living near wind farms are exposed [31,32]. Some authors point out that part of the problem related to noise from wind farms relates to infrasound [33][34][35][36][37][38][39][40][41][42][43], while others highlight the significance of low frequency sound [44,45]. After an extensive analysis of the literature, Freiberg et al. [46], in an assessment of the potential effects of wind turbines on human health, concluded that in the case of infrasound and low frequency noise generated by wind turbines, the main effects of their impact on humans are physical symptoms, health effects in general and affective influence (mood). ...
Article
Full-text available
Determining the spatial position of wind turbines is the initial and most important phase in the development of a wind farm project. In this sensitive phase, all potential problems that may arise in the later stages of project development should be prevented by means of spatial and urban planning instruments. This makes it possible to achieve maximum use of the potential of wind in a particular space and, thus, fulfil the technical and economic requirements of the project while respecting the goals of environmental protection in that same area, through preventive protection. Therefore, it is essential, even at the earliest planning and development stage of a wind farm project, for the requirements that are important for optimal spatial solutions to be balanced. In this process, strategic environmental assessment (SEA) is a support to the planning process and an invaluable instrument for finding optimal spatial solutions for the possible key spatial impacts of wind power with regard to noise, shadow flicker, ornithofauna and chiropterofauna. The weakness of SEA can be seen in its predominant application of expert qualitative methods that bring with them subjectivity, since they depend on expert knowledge and skills. This paper presents the aspect of noise impact assessment and its inclusion in the SEA for the Maestrale Ring wind farm in Serbia. The results of the research indicate how it is possible to achieve the principle of objectivity in the process of multicriteria expert evaluation by including the results of a partial impact assessment of the noise from wind farms, using results obtained from software modeling of the spatial dispersion of wind turbine noise in the SoundPlan 8.1 software package in the SEA process. These quantitative results predicting the noise level were used in a semi-quantitative method of multicriteria evaluation in the SEA through the definition of criteria to determine the ranking of impacts, which is elaborated in the paper. The results also show the significant of the contribution of applying a methodological approach based on a combination of qualitative and quantitative evaluation methods in SEA. These methods positively affect the application of the principle of preventive protection through the optimal selection of the number and position of wind turbines on one hand and the objectivity of drawing conclusions based on which strategic decisions are made in the final phase of the SEA process, on the other.
... As a consequence, it can cause inappropriate attitude to infrasound of people living near wind farms. Therefore, intense research is being conducted in this area by numerous scientific institutions in order to systematize knowledge with regard to the broadly understood problem area of infrasounds emitted by working wind turbines [33][34][35][36][37][38][39]. The final effect should be the development of a reference measurement methodology and the indication of unambiguous criteria of assessment of their acceptable impact. ...
Article
Full-text available
The subject matter of the paper is concerned with improvement in the methods of measuring, processing and analyzing the infrasound noise caused by working wind turbines. The article presents results of measurements and analyses of infrasound signals emitted by work of a wind turbine with 2 MW nominal power, during its normal operation. Infrasounds were recorded at the same time in three independent and identical, in terms of the applied devices, measurement tracks. For the measurements of infrasound signals a dedicated measurement system was used. The measurements were made at variable weather conditions and for different distances between the measurement points and the turbine being examined. In the results part of the paper the following are subsequently presented: the time courses of the recorded infrasound signals, and then courses of the periodograms designated for them in frequencies and the spectrograms calculated for time-frequency data. Special attention was paid to the analysis of the possible application of time-frequency ridge transformation, the application of which enables effective extraction of energy-wise dominant instantaneous frequencies. In the summary, an analysis was made of the results obtained by means of frequency, time-frequency and time-frequency ridge analysis methods.
Conference Paper
Full-text available
A number of propagation models of varying complexity have been developed for environmental noise predictions. While more complex models are correspondingly more accurate, the continued widespread use of the simpler ISO 9613-2 model suggests that its relative simplicity is valued highly. Therefore, to extend the use of ISO 9613-2, it is proposed that a correction should be added to take into account that at some distance from a point source, the propagation of low frequency noise may begin to decay at a rate of 3 dB per doubling of distance rather than 6 dB, during atmospheric conditions where the sonic velocity near the ground increases with altitude, resulting in downward refraction of sound rays and reduced attenuation with distance. Since low frequency sound is poorly attenuated by the ground and atmosphere, a distance from the source will exist beyond which multiple sound rays, which have been reflected from the ground more than once, can arrive at the receiver thereby increasing the sound pressure level at that location. There is some disagreement as to the transition distance at which the attenuation rate changes from 6 dB/doubling of distance to 3 dB, due to the influence of source height, frequency, and the strength of the wind and temperature gradients. This study aims to determine whether decay rates of 3 dB/doubling of distance are applicable to modern wind turbine noise propagation for a range of atmospheric conditions and frequencies and to determine a typical transition distance for the investigated conditions. This investigation focuses on comparison between ISO-9613-2 predictions using different transition distances and a large data set that has been collected at a number of different locations in the vicinity of a modern wind farm. A theoretical analysis is also carried out to calculate the radii of curvature of rays that leave the source at different angles. This enables prediction of the transition distance for various wind profiles and source heights.
Article
Full-text available
The objective of the present paper is to present an overview of LFN characteristics of modem MW turbines based on numerical simulations. Typical sizes of modem turbines are from 1-3 MW nominal generator power and a rotor diameter ranging from 80-100 m but larger prototypes up to 5 MW and with a rotor diameter of 126 m have now been installed. The numerical investigations comprise the common upwind rotor concept but also the turbines with a downwind rotor are considered. The reason to include the downwind rotor concept is that this turbine design has some advantages which could lead to future competitive designs compared with the upwind threebladed rotor. The simulation package comprises an aeroelastic time simulation code HAWC2 and an acoustic low frequency noise (LFN) prediction model. Computed time traces of rotor thrust and rotor torque from the aeroelastic model are input to the acoustic model which computes the sound pressure level (SPL) at a specified distance from the turbine. The influences on LFN on a number of turbine design parameters are investigated and the position of the rotor relative to the tower (upwind or downwind rotor) is found to be the most important design parameter. For an upwind rotor the LFN levels are so low that it should not cause annoyance of neighbouring people. Important turbine design parameters with strong influence on LFN are the blade tip speed and the distance between rotor and tower.
Conference Paper
Full-text available
Outdoor and indoor microphone measurements have been taken in the vicinity of the Waterloo wind farm at a number of locations during periods when the nearby wind farm was operational as well as when it was shutdown. The majority of the shutdowns were of short duration and deliberate on the part of the wind farm operator, as they were associated with the recent EPA noise impact study. However, one of the shutdowns lasted for several days as it was related to a cable fault. Comparisons are made between both the third-octave spectra and narrowband spectra measured during the shutdown and operational periods. Operational times immediately adjacent to the shutdown times, as well as at other times when the wind conditions at hub height and at the residence matched the conditions recorded during a shutdown time, are considered in the analysis. It is shown that there are consistent and significant differences in noise spectra at the residence for the shutdown and operational cases, particularly for frequencies below 100 Hz. These differences can be observed at distances up to 8.7 km from the wind farm. Keywords: Wind farm noise, amplitude modulation, shutdown I-INCE Classification of Subjects Num-ber(s): 14.5.4
Article
Full-text available
Airborne Infrasound at any given point can be accurately described as fluctuations or cyclic changes in the local barometric pressure. Variations in a motion sickness test subject's elevation result in fluctuations in the surrounding barometric pressure by similar degrees to that experienced on a ship in high seas. Cyclic variation in the lateral or linear velocity of a subject in a vehicle or platform in atmospheric air may also be subject to infrasonic pressure fluctuations due to the Bernoulli principle and possibly vortex shedding effects. Calculations presented demonstrate that in at least one landmark study (McCauley et al., 1976) test subjects were exposed to infrasonic sound pressure levels in excess of 105 db at discrete frequencies between 0.063 and 0.7 Hz. The infrasonic sound pressure level necessarily present in cyclic motion in free atmospheric air does not appear to have been accounted for as a nausea influencing factor in the McCauley et al. (1976) motion sickness studies.
Article
Full-text available
Relatively balanced load and velocity related pressure waves from the rearward facing surface of each rotor blade, are at a frequency of 1 cycle per revolution of the turbine and are phase shifted by 120 degrees from each other. The superposition of these infrasonic waves destructively interfere. This action results in a non-propagating rotor locked mode; however, the shielding (reflecting) effect of the tower as each blade passes, interrupts the balanced destructive interference for a small portion of rotor angle three times per revolution. The momentary un-balance between the destructive interfering waves results in the generation of Tyler-Sofrin spinning mode series, which propagate into the far field. The spinning mode radiation angles, coupled with the low decay rate of infrasound, result in higher far field sound pressure levels than would be predicted for a point source. An analysis approach partially derived from Tyler-Sofrin (1962) is presented. Field microphone data including phase measurements identifying spinning modes are also presented.