Reproducing wind farm infrasound for subjective testing – Just how accurate is the reproduced signal?

Steven CooperMay 2016Australia

Reproducing wind farm infrasound for subjective testing – Just how accurate is the reproduced signal?

“As a result of the work to date the generation of a signal for test subjects that is intended to reflect that of wind farms should use WAVE files of actual wind farm and not synthesised waveforms.”

“For the purpose of assessing infrasound from turbines the frequency response from speaker manufacturer’s data is of no assistance.”

By Steven Cooper
171st Meeting of the Acoustical Society of America – Salt Lake City, Utah – May 2016


Apparently on the basis of room modes being excited in residential dwellings one concept has been to ignore high quality field measurement recordings (Wave files) and use the narrow band (FFT) Leq results of a 10-minute sample to create, by superimposing a number of sine waves, a trapezoidal time signal as the source for subjective testing and restricting the bandwidth to only infrasound. Other testing has utilised full spectrum signals but has issues with providing an accurate signal due to the limitation of the speakers. An examination of both methods and the limitations of the results have been examined and will be discussed.


In response to investigation of residents’ complaints concerning the operation of wind turbines, independent acousticians have identified the presence of a discrete infrasound/low frequency signature associated with the operation of the turbine to be present when such turbines are operating.

The discrete signature of turbines when using narrowband analysis reveals peaks at the blade pass frequency (and harmonics of that frequency) to occur in the lower portion of the infrasound frequency band, generally below 10 Hz and a peak with sidebands around what may be the gearbox output shaft speed.

Attenuation of infrasound over distance occurs at a lower rate than that of normal sound, resulting in the discrete infrasound signature of turbines being recorded up to 7 km from wind farms and in some situations even greater distances.

Infrasound measurements of the natural environment in rural areas free from the influence of wind turbines whilst revealing similar broadband levels of infrasound (for example using dBG or even 1/3 octaves) do not experience a discrete periodic pattern similar to that associated with rotating blades on wind turbines when assessed in narrow bands.

In seeking to assess the audible characteristics of wind turbine noise, being different to that of general traffic or environmental noise, laboratory studies have sought to use speakers to generate or to reproduce recorded signals for test subjects in a controlled environment.
Restriction of analysis to A-weighted noise levels or 1/3 octave bands has limitations in examining the impacts of the noise in question.

All of our wind turbine measurements have revealed infrasound levels below the nominated threshold of hearing and low frequency modulated noise as inaudible/barely audible and in some case clearly audible (to sensitised individuals) inside dwellings.

With the (re)identification of the presence of discrete infrasound signatures associated with wind turbines in the last few years, experiments have been conducted on just the infrasound components even though such levels are below the threshold of hearing. Conducting such experiments would confirm inaudibility with conclusions posed that infrasound is not an issue.

Cooper raised the possibility that the infrasound signature, whilst identifying the operation of turbines, may be the by-product of the modern day analysers and that the impact on people could be the result of audible and inaudible modulation of frequencies occurring in the normal range of frequencies associated with everyday sounds.

As the impact of the turbine’s inaudible infrasound on people has not been studied in controlled studies, of critical important in the laboratory assessment of wind turbine “noise” is the question as to whether the source signals generated in the laboratory are full spectrum and reproduce the original signal (that includes by narrowband analysis infrasound).

Experiments into wind turbine “noise” have considered just the infrasonic region, audible spectrum only (no infrasound), and possibly full spectrum (infrasound components unknown).

Tachibana used a set of reverberation chambers to evaluate full spectrum sound of wind turbines. However, the primary issue presented in the paper was looking at the A-weighted level with different low pass filtering and modulation. Reference did not examine infrasound specifically but concluded that frequency components below 25 Hz are not audible which is to be expected for the levels that were tested. As a side issue to the investigation of the A-weighted levels and audibility of the modulation, the audible modulation effects were identified as associated with low frequency.

Walker provided results of generating infrasound signals synthesised from narrow band Leq analysis to find no impact. No frequency response was provided to define the output of the synthesised infrasound signal generated by a speaker. There is an assumption the system equalisation curve resulted in a flat spectrum.

Walker started with external wind farm noise samples from the Waterloo wind farm that were then synthesised from the narrow band frequency spectrum to provide the source signal.

Tonin has used a synthesised infrasound signal applied to a pnuematic driver connected to modified hearing protectors.

Crichton has used single infrasound tones inserted into broad band noise for an assessment of “wind turbine infrasound”.

The presentation of synthesised signal of turbine noise as a result of a creating a signal based upon the individual FFT components of a time average (rather than a WAVE file of the original signal) has been used to show the inaudible infrasound frequencies do not impact upon people by suggesting that the synthesised signal or pure sine waves is a representation of the signal residents receive as a result of an operational wind farm.

Issues of concern with the use of simulated “infrasound” are:

  • Whether the synthesised signal (obtained from adding sine waves) reproduces the actual time signal that occurs in the field.
  • “Infrasound” applied as single tones and then attributed as being the signal generated by wind farms
  • Testing of synthesised signal and claiming the results apply to wind farms.
  • Accurately reproducing the Wave file signal by the use of speakers.



The wind turbine acoustic signal is not restricted to infrasound but provides frequency spectra that covers infrasound and normal audible frequencies.

The time signal of wind turbines show pulses. Due to the variable nature of the acoustic signal over a ten-minute period and the nature of pulses occurring at the blade pass frequency, the Leq result must be lower than the peak levels. Whilst the general assessment of wind turbines is based on the A-weighted Leq level there is a question as to whether people are responding to the A-weighted Leq, peak level A-weighted or linear, bands of sound in the infrasound region or the low frequency region, or responding to inaudible/audible energy.

Some experiments into wind turbine noise have been restricted to just the infrasound components.

The use of synthesised “infrasound” signals from narrow band Leq analysis does not to produce the same time signal as the original signal.

Our testing has found that the original time signal cannot be reproduced by the synthesis method as the Leq narrow band FFT results are an average over time and are devoid of any phase information in relation to the individual frequency components.

If the generation of only infrasound that is inaudible does not produce an impact, then it follows that testing should use full spectrum signals.

If infrasound and low frequency components are a significant factor in the sound field that occurs, then the use of traditional speaker boxes places restrictions on the sound field that is generated.

The presentation of wind farm noise (full spectrum, including infrasound) requires careful consideration, assessment and calibration of the resultant sound so as to reproduce the original signal.

The multiple speaker baffle arrangement has been used for infrasound threshold measurements and subjective testing of wind farm audible noise.

As a result of the work to date the generation of a signal for test subjects that is intended to reflect that of wind farms should use WAVE files of actual wind farm and not synthesised waveforms. Obtaining accurate Wave files is relatively straightforward and overcomes the inaccuracies and conflicts that arise by the use of synthesised “wind farm infrasound” signals, or the concept of single tones in the infrasound region and claiming such synthesised signals/tones are “wind turbine infrasound”.

The challenge for researchers in the laboratory (separate to measurements in the field) is to be able to reproduce the original signal. Obtaining an audio signal in the laboratory from 0.8Hz up that reproduces the original WAVE file presents many challenges. Obtaining a digital to analogue convertor and amplifier chain with appropriate signal to noise ratio is a limiting factor for laboratory testing of the full spectrum.

The preliminary results reveal interaction of individual speakers and the volume of air behind the speakers affects the linearity of the frequency response from a plane wave, that requires further work.

The transient response of the wind turbine time pulses may be a limiting factor in reproducing the original signal.

The above limitations are subject of further investigations.

Full paper


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s