Time Series of Amplitude Modulation of Wind Farm Noise at Noise Receptor Locations
Joachim Gabriel, Steffen Vogl, Thomas Neumann
DEWI GmbH (German Wind Energy Institute), Wilhelmshaven, Germany
Gundula Hübner, Johannes Pohl
Research Group Health and Environmental Psychology, Institute of Psychology, Martin-LutherUniversity Halle-Wittenberg, Germany
Summary
Installations of wind turbines (WT) are accompanied by an increasing number of complaints
about noise from wind farms. What kind of sound characteristics makes residents complain?
Amplitude modulated aerodynamic noise (AM) turned out to be the common ground of
complaints documented within a research project about acceptance of noise from wind turbines.
The project combined psychology and acoustic noise measurements. In addition audio
recorders had been handed out to residents of a wind farm to record annoying noise situations
by themselves.
In order to get an idea of frequency, duration and intensity of AM of wind turbine noise, the
audio files have been listened to and analysed by a numerical evaluation tool. Within the tool
the modulation depth ∆L was used as a parameter for AM. In addition a second parameter for
AM quantification - fluctuation strength F* - was used. F* is a psycho acoustically motivated
measure, representing the perceived AM. Within the 28 recordings analysed, a clearly
perceptible AM occurred only sporadically and mostly for short time periods, but it might trigger
the attention and focus it on the sound of the wind farm.
An improved version of the AM assessment tool - featuring long term data analysis - will help to
identify AM critical operational conditions. Standardization of AM assessment is aimed in order
to be able to quantify the specific AM-annoyance and the efficiency of AM mitigation. An
objective assessment of AM can decouple AM from impulsivity and infrasound.
1
Introduction
Operators of wind farms are confronted with an increasing number of complaints about noise
from wind turbines. Is there any correlation of complaints and certain weather or operational
conditions? If there is, can the acceptance be improved by noise mitigation focussed on these
situations? Is there a way to change the sound performance of a wind farm for potential times of
annoyance?
The German research project “Study on impairment of residents by noise emissions of wind
turbines and deduction of transferable intervention strategies for their mitigation” [1] funded by
Deutsche Bundesstiftung Umwelt (DBU, German Federal Foundation for Environment) tries to
answer these questions. The project combines interviews with residents of a wind farm,
professional physical measurements and sound recordings by residents.
With the focus on physical aspects this paper will present findings from evaluations of sound
recordings at noise receptor locations.
2
The Project
The subject matter of the project is a wind farm of 9 two-megawatt-WTs in Northern Germany at
a distance of approximately 1500 m from a village in rural flat landscape. The hub height is
108 m.
Assuming sound power levels according to the information supplied by the WT manufacturer,
sound propagation calculations result in sound pressure levels below 33 dB(A) for houses near
the wind farm.
In March 2012, 212 residents were interviewed using a questionnaire with 450 items. Residents
also got forms to document complaints about noise from the wind farm (22 items). Audio
recorders were available in order to document times with annoying sounds from the turbines.
These recordings have been listened to and analysed in addition to temporary standardised
noise measurements. Meteorological and WT data, derived from the WT controllers and an
independent data acquisition system, were recorded for the whole time of the campaign.
The data basis for the analysis were standardised complaint forms, acoustical data and
minimum / maximum values as well as 10- minute averages of wind speed at hub height and at
10 m height, wind direction, active power, and rotational speed for each WT.
Figure 1: Residents view on the wind farm
3
Measurements and Evaluations
Figure 2: Acoustic noise measurements near the wind turbines
Standardised measurements according to IEC 61400-11 Ed.2.1 show no indication of any
unexpected acoustic noise emission of the WTs and sound power levels used for the sound
propagation calculation were proved within the measurement uncertainty.
Figure 3: Acoustic noise measurements at a noise receptor location
Temporary acoustic noise measurements according to DEWI’s accredited standard were
performed at one of the houses relatively close to the wind farm. Measured sound pressure
levels for the wind farm operating did not stand out from the background noise. This was
expected because of the large distance to the wind farm and the level of the masking noise.
Because of the high level of background noise additional measurements closer to the wind farm
were performed. The result was a better signal to noise ratio but being different from the sound
perception of the residents, the recordings did not help to understand the complaints.
Figure 4: Additional measurement location using a vertical board
according to [2] to get a better signal to background noise ratio
According to the project data and the results of measurements and evaluations there is no
indication for any noise immission problems with regard to the German immission protection
law. Compared to other German projects the wind turbines are installed at a relatively great
distance to the houses and sound pressure levels are likely to be far below legal limits.
Nevertheless residents complain about the noise from the wind farm and there even have been
judicial proceedings.
A total number of 95 complaints by 10 residents were documented. 80 % of the complaints
occurred at night or early morning. The data base did not allow the definition of a certain critical
operational mode or weather condition for complaints, but daytime patterns occurred. Most of
the complaints were made for low background noise conditions.
4
Amplitude Modulation as Common Ground of Complaints
Neither loudness of the broadband WT-noise, nor tonality, nor impulsivity (at least according to
standardised definition of impulsivity) is the reason for the documented complaints.
Interviews and sound descriptions summarized in table 1 show that residents complain about
sound identified as different from the “natural” background noise, even if the loudness of this
special sound is very low and hardly perceptible.
Thus the common ground of the documented complaints is a WT-specific sound characteristic
directly correlated to the rotation of the rotor blades. This sound - described as swishing in the
frequency of the rotor blades - is amplitude modulated aerodynamic noise of the wind turbines
(AM).
Description as modulated noise
(e.g.: loud pulsating noise; Pfsch - Pfsch - Pfsch ...)
68
Perhaps modulated noise
(e.g.: hissing, roaring, recurring, irregular ...)
21
Description as a steady noise
(e.g. constant noise, aircraft noise constantly…)
Total number of complaint sheets
6
95
Table 1: Evaluation of complaint sheets with regard to AM
Only 6 % of the complaints described the annoying sound as a „steady noise“.
There are studies and findings about amplitude modulation (AM) in the near field. But can
results be transferred to the far field? Long distance AM is hard to identify in measurements.
Currently there are no filters for automatic measuring techniques available.
As for every acoustic measurement far from the WT, the level of turbine noise can be very close
to background noise. Thus you cannot be sure results refer to noise from the wind farm, unless
you identify wind turbine noise by listening.
It is difficult to get relevant professional recordings of far field AM by temporary measurements.
A common scenario for temporary professional sound measurements at houses of people
complaining about noise from WTs are comments like: “Believe me, the other day the noise was
worse!” It turned out to be a good idea to hand out audio recorders to residents to record
annoying noise situations by themselves. Thus a total number of 28 wav-files was available to
assess far field AM.
It was necessary to listen to all these recordings to find out if wind turbine noise causes the
annoyance. This ear-based identification applies individual listening and visual experiences.
Sounds changing with the frequency of the rotation of rotor blades can easily be recognized as
wind turbine noise. But a steady broadband low level sound cannot be filtered even by a trained
listener.
5
Quantification and Assessment of AM
In order to get an idea of frequency, duration and intensity of AM in wind turbine noise, the
audio files have been listened to and analysed by a numerical evaluation tool documented in
[3]. Within the tool the modulation depth ∆L was used as a parameter for AM. The modulation
depth ∆L is defined as the difference between the maximum and minimum sound pressure
level. In addition a second parameter for AM quantification - fluctuation strength F* - was used.
F* is a psycho acoustically motivated measure, representing the perceived AM. The origin of F*
is the fluctuation strength, developed by Zwicker [4], which was slightly modified for the analysis
of wind turbine noise.
Figure 5 gives an overview of the two approaches.
Figure 5: Two approaches for AM assessment
Figure 6 shows the analysis of one of the 28 sound samples. Modulation depth ∆L was used as
parameter for AM. The black line is the A-weighted 1/3 octave band spectrum of the sound
pressure level showing a maximum at 160 Hz. 14 dB as the highest value for ∆L have been
assessed for this 1/3 octave band.
Frequency / Hz
Figure 6: Example of spectrum and evaluated modulation depth of wind farm noise
An example for a resulting time series of modulation depth ∆L and fluctuation strength F* is
given in Figure 7. The assessment tool produces results for ∆L and F*, which correlate
reasonably with the subjective impression of AM perception and annoyance. However the tool
cannot distinguish between AM caused by a WT or by human voices and birds. Therefore
human listening remains necessary.
Figure 7: Example of a time series for AM
The audio file of Figure 7 was recorded at 03:20 am and is the only record of 28 showing AM
longer than one minute. In the other recordings a clearly perceptible AM only occurred for short
time periods (typically < 10 s) or was not at all audible. But it must be noted that the recordings
are just snapshots and no statement about the sound situation “beyond the record” can be
made.
Evaluated modulation frequencies of 0.7 Hz to 0.9 Hz correspond to rotational speeds of 14 to
18 rpm for the wind turbine.
Figure 8: Example of a time series with short AM events
Figure 8 shows two short time periods of less than 5 seconds of relevant AM. An interpretation
of this time series is given in the next chapter.
6
Interpretation of AM noise perception
The recordings produced within this project still are an insufficient data base for final
conclusions. Still we need to know more about the phenomenon amplitude modulation of wind
turbine noise to say how often, how long and how intensively it occurs. But what we know is:
human beings are very sensitive for AM sound. It is a prehistoric protective mechanism. AM is a
carrier of information. Awake or asleep we all “scan” our environment for signs of danger. A
“Pfsch - Pfsch – Pfsch” sound might potentially be steps of an enemy approaching.
Thus, even a short period with noticeable amplitude modulation might trigger the attention.
Fortunately enemies creeping up on you are not very common these days. So it is just the
sound of the wind farm, but the attention is focused on it. The sound perception may remain
focused on the wind farm noise, even when the amplitude modulation has fallen again below
the perception threshold. Figure 9 illustrates this interpretation of AM noise perception.
Figure 9: Illustration of an interpretation of AM noise perception
The maximum human sensitivity for amplitude modulation is at 4 Hz. Still typical modulation
frequencies of wind turbine noise in the range of 0.6 Hz to 1 Hz potentially cause a kind of
“alarm”. This might explain the comment of a resident, saying that he wakes up the night
because of conspicuous wind turbine noise.
In the example of figure 9 he might wake up with the first peak of AM not knowing wind farm
noise was the reason. His concentration is on listening if there might be some “danger”. With
the next “peak” of AM - two minutes later - he might identify the wind farm as origin of the noise.
This conclusion is based on his experience with wind farm noise (rotation with the same
frequency) and the “training” of his perception. Up to this point the story is independent of the
individual attitude towards the wind farm. A local resident who has complained before court
against the wind farm will react differently from a shareholder of the project. Dependent on the
individual attitude towards the wind farm, reactions may differ from “fine - this sound is money in
my pocket” to annoyance and anger. It even might be hard to rest or to fall asleep again.
According to this, complaints could be reduced if it was possible to eliminate short “Trigger-AM”.
7
Ideas for Mitigation of the Annoyance
Noise reduced modes usually base on reduction of the rotational speed of the WT rotor. How
this affects far field AM is unknown. Based on investigations of (near field) AM, reduction of
trailing edge noise is in the focus when thinking of AM mitigation. With regard to the above
described findings, there are some more ideas for AM mitigation.
If AM as an attention trigger is a short time phenomenon investigations of the causes of far field
AM are worth to be carried out. It is likely necessary to look at
-
dynamic aero acoustics as well as
-
wake effects,
-
sound propagation and
-
wind turbine interaction
in combination to understand this dynamic phenomenon.
Once AM-critical operational modes or sound propagation situations are identified specific
modification can reduce or extinguish AM.
Another idea can be described as “sound design”. As it is more the “quality” than the “quantity”
of the wind turbine sound that causes complaints, it is conceivable to achieve an equalization of
the WT noise either by single blade pitching or even by active generated synchronised
broadband noise.
Hearing experiments [4] showed that AM is most annoying at 4Hz. Therefore reduction of
rotational speed might reduce annoyance. The above mentioned active systems might also be
used to decouple the modulation frequency from the frequency of the rotational speed of the
rotor. The sound would then no longer be associated with the optical impression of turning rotor
blades.
8
Conclusion and Outlook
Rating scale for the efficiency of any mitigation action must be a reproducible method for AM
assessment which takes into account that human AM sensitivity is frequency-dependent. It can
help to objectify the discussion about a WT specific annoying sound characteristic. Especially
because AM of WT noise is often mixed up with impulsivity and infra sound. There are
examples of AM being regarded as impulsivity even in legal proceedings and infra sound
discussion might just be a confusion of modulation frequency and spectral frequency.
In addition to previous work of e.g. Jeremy Bass [5] the above mentioned DEWI software tools
might be a start for standardisation. The idea is to develop a standard for an objective
assessment of AM in noise from WTs comparable to the German DIN 45681 for tonality. It must
be emphasized that this standard is for noise at receptor locations only and must not be mixed
up with “near field” guidelines like IEC 61400-11. The discussion of AM standardisation is on the
agenda of the national German WT noise working group.
A couple of questions about far field AM might be answered by means of a far field sound
recording station and adapted AM assessment tools. With the help of AM audibility data and in
correlation with high resolution meteorological and wind turbine data manufacturers can
improve their knowledge about AM and optimize the sound performance of their turbines.
References
[1]
Research project: Untersuchung der Beeinträchtigung von Anwohnern durch
Geräuschemissionen von Windenergieanlagen und Ableitung übertragbarer
Interventionsstrategien zur Verminderung dieser (Study on impairment of residents by noise
emissions of wind turbines and deduction of transferable intervention strategies for their
mitigation);
Gundula Hübner, Johannes Pohl, Research Group Health and Environmental Psychology,
Institute of Psychology, Martin-Luther-University Halle-Wittenberg;
Joachim Gabriel, DEWI GmbH (German Wind Energy Institute);
funded by Deutsche Bundesstiftung Umwelt (DBU, German Federal Foundation for
Environment), wpd windmanager
[2]
Osten, Tjado (1994): Schallimmissionsmessungen in der Umgebung von
Windkraftanlagen bei unterschiedlichen meteorologischen Bedingungen und Erprobung einer
neuartigen Meßmethode, Diplomarbeit am DEWI
[3]
Vogl, Steffen (2013): Erfassung und Bewertung von Amplitudenmodulation in
Windenergieanlagengeräuschen, Bachelorarbeit am DEWI
[4]
Zwicker, E. und Fastl, H. (1999): Psychoacoustics: Facts and Models, Springer series In
information sciences, Springer.
[5]
Bass, J. (2012): Wind Turbine Noise - from Source to Receiver Far-field Noise Issues AM, Tonality & Impulses, EWEA Wind Turbine Noise Workshop, Oxford, England.