‘Sensing but Not Hearing’: Latest from Steven Cooper on Wind Turbine Nuisance

By Sherri Lange
November 5-6, 2019

“But wind turbine noise issues are not just an acoustic issue. I have been trying to solve the acoustic problem to allow the medical side to then undertake the required research.” (Cooper, below)

Last year, I interviewed acoustician Steven Cooper, Australia, on wind turbine health issues related to pulsation and low-frequency noise. “In general, wind farm applications claim that turbines do not generate any low-frequency, tonal, or impulsive characteristics,” he noted, “which is a matter disputed by residential receivers.”

What has developed in the last 20 months? In this new interview, Mr. Cooper shares his most recent research and findings, which complement our current knowledge regarding the nature of “noise” impacts to real-time victims of wind power.

Mr. Cooper recently presented his new findings in Germany at the International Congress on Acoustics Conference. As Mr. Cooper states in this interview:

“I question how an authority can propose noise criteria with no fundamental basis identified as to how a stated core objective can be satisfied.”


Q: Since our previous discussion for Master Resource you have continued your research into the basis of wind-turbine noise. Your presentations to the Acoustical Society of America in December 2017 and Euronoise in May 2018 presented the results of your testing of inaudible wind-turbine sound on a group of people who have been severely impacted by turbine noise to the extent of regularly leaving or permanently leaving their home, versus a group of people never been exposed to wind turbine noise.

Cooper: Yes.

Q: You still seem to be pushing boundaries and asking fundamental questions on what is the “signature of wind turbine noise.”

Cooper: Correct. As a noise engineer you have to understand what is causing the problem before you can arrive at a solution. But wind turbine noise issues are not just an acoustic issue. I have been trying to solve the acoustic problem to allow the medical side to then undertake the required research.

Q: Congratulations are in order as you have been appointed to the European Acoustics Association Technical Committee on Noise as the group leader on wind turbine noise, and last month gave three papers at the International Congress on Acoustics in Germany.

Cooper: Right. The papers were well received and led to multiple discussions with individual experts during the conference. The three papers and a pdf of the power point for each paper are up on National Wind-Watch, two in the acoustics section and one in the general section.

Q: In our previous interview for Master Resource you identified a series of relatively simple questions in relation to criteria to protect against sleep disturbance that need to be answered by Regulatory Authorities. Any response?

Cooper: No. Still a deafening silence. Also on Wind Watch is my submission on a review of draft wind farm guidelines for South Australia that refers to those questions. I question how an authority can propose noise criteria with no fundamental basis identified as to how a stated core objective can be satisfied.

Q: Previously we have discussed that you don’t use the term ILFN (infrasound and low-frequency noise). I have trouble getting my head around that. Your position on ILFN seems to cause some problems for people in discussions on wind turbine noise. When discussing infrasound from turbines you have referred to the infrasound signature showing peaks in the spectrum being multiples of the blade pass frequency. Why this specific terminology?

Cooper: I have never used the term of infrasound sound. I used “infrasound signature” to refer to the result of a frequency analysis. This is because the noise from a turbine has a broad band signal with some tones associated with the gearbox, but the time signature of the pressure signal from turbines is a series of pluses that occur at an infrasound rate. The pulses are very short in time when compared to the wavelength of the blade pass frequency and are not of a continuous nature like an audible sound. You can use FFT analysis to show the presence of an operating wind turbine.

Q: The term FFT is used in wind turbine noise, and you reference it here. What does it mean?

Cooper: FFT stands for fast Fourier transform. It is a method used in digital analysis to extract the periodic functions in a time signal to derive a frequency spectrum. It is the modern way of creating narrow-band analysis.

This form of analysis is the basis of the “infrasound” problem. The FFT procedure identifies the occurrence of the presence of the pulsations in the time domain (say every 1.2 seconds) which is at the blade pass frequency (0.86 Hz), because they are a periodic function. That doesn’t mean there is a constant sound at 0.86 Hz.

Q: As a result of your research, you have produced a number of presentations in relation to the acoustic signature of wind turbine noise and have questioned the issue of infrasound being an actual sound signal.

Cooper: Correct. For the ASA (Acoustic Society of America) wind turbine working group I have given a number of presentations on the question of there being infrasound present in the acoustic signature. In our quest to accurately reproduce the wind turbine signal for subjective testing we found real issues in the frequency response of speakers/amplifiers, but as presented to the ASA in Boston we also found issues with the analysis of the pulses. We could remove all frequencies being sent to the speakers below 50 Hz and the infrasound signature was still present. There are fundamental technical issues for analysis that are not met by the short duration pulses in that what is being assessed are the pulsations of the signal, not actual infrasound.

Q: So if I understand this correctly you are saying you can take the wind turbine signal that in a normal frequency analysis shows peaks at the blade pass frequency (say 0.86 Hz) and multiples of that frequency, and reproduce that signal in a laboratory that can go down to 0.5 Hz. But having a filter to block out all the infrasound going to the speakers, you still get a frequency analysis of that sound still shows peaks in the infrasound region.

Cooper: Yes. Because the analyser is providing the results of the overall sound levels varying (modulating) at an infrasound rate.

Q: Well that concept may be giving me a headache. But that would be in my head. And this leads me to the nocebo concept, that wind turbine impacts are psychosomatic, proposed by Simon Chapman, and that it is all in people’s heads. You say he is right in one sense but wrong in another.

Cooper: Yes. In the New Orleans ASA and the Euronoise presentations we identified that the sensitive people were ableto sense the operation of the signal (even though it was inaudible) and the majority of the sensitive group identified feeling it in their head [emphasis added]. In the brainwave paper at the ICA for a test case we showed the inaudible signals affected frontal lobes. So it is right that people can sense the signal but not in the concept proposed by Chapman that they are making it up. The amplitude modulation paper goes to the work of others in relation to sensing the modulation.

Q: I see that some acousticians quote the work of Crichton as setting the bar on establishing the nocebo concept.

Cooper: Yes, and there is a problem with acousticians accepting that data without examining the material. If you look at the ICA [International Congress on Acoustics] synthesis paper and more importantly, slides 6 & 7 of the PowerPoint, you will see that the basis of the Nocebo proposed by Crichton is flawed.

Despite the title of Crichton’s paper she never used what has been identified in frequency analysis as “infrasound” for wind turbines. Look at the title of the paper then look at her signal. She used a single tone (see slide 6). I have never seen a single tone at 5 Hz (or 9 Hz) from wind farms.

The figure slide 7 of the PowerPoint superimposes Crichton’s “infrasound signal” onto the Shirley wind farm graph provided by Walker, Hessler & Hessler, Rand, and Schomer. That frequency graph is typical of a narrow-band frequency analysis of a wind turbine. Also note, the orange line is the manufacturer’s specification for the speaker used by Crichton that would seem to have some difficulty producing the “wind turbine infrasound” used in the experiment.

Furthermore, if you look at Crichton’s PhD thesis (on this work), one of her supervisors was Simon Chapman.

Q: The synthesis paper raises further issues with using just wind turbine infrasound and creating a digital signal for subjective assessments of wind turbine noise.

Cooper: Yes. If the synthesised “infrasound” wind turbine levels are well below the threshold of hearing for actual infrasound sounds and don’t produce the same time signal pressure variation as in the real world, and leave out the rest of the wind turbine signal that is in the normal audible range of sounds, then why use such signals – particularly when you have a US Standard for wind turbine (Annexure D of ANSI/ASA S12.9-2016/Part 7 “Advanced Signal Processing Techniques”) providing caution on using a signal that may have the same energy but sounds different.

Q: Your Cape Bridgewater study took a fair time to digest and comprehend. I think people may need a fair bit of time to digest this new information and carefully read the paper and also view your PowerPoint.

Cooper: Yes. This is what happened in Germany. Quite a few esteemed acousticians later in the week came to have a one-on-one session with me. I hope people will take the time to explore my clarification of this complex issue. We still do not have government authorities or wind developers being required to explain any of the impacts, nor protect people from such impacts.

For example, in Australia we still do not have the Authorities or the Wind Farm Commissioner providing noise criteria for wind turbines that is based on actual measurements/studies that identify noise levels that will not give rise to sleep disturbance. As these issues are relevant to the three ICA papers, my submission in relation to a draft noise guideline in Australia is also on Wind Watch. A review of that paper will place in context my material presented at the ICA.


“The description of wind turbine noise needs a terminological shift. The language should be pulsations at an infrasound rate with modulation of the entire signal at an infrasound rate (as in sensation detected by the ear).” [Cooper, below]

Q: You referred earlier to your second paper at the ICA that is linked to the synthesis paper. Maybe my question on ILFN ties in with the paper on amplitude modulation that raises questions as to terminology?

Cooper: Yes. The second paper on amplitude modulation is very important. Because what people are calling AM (by reference to the dB(A) signal) is incorrect. An electrical engineer will tell you that AM is the modulation (variation in the amplitude) of a carrier frequency (being a high frequency) that is modulated at a lower rate. The dB(A) is not a single frequency.

Turbines exhibit a tone at discrete frequencies of between 25 and 32 Hz (dependent upon the turbine model). In the ones I have measured it relates to the speed of the shaft that drives the generator. The gearbox is subject to changes in loading as the rotors rotate that must be transferred to the generator. The change in loading occurs at the blade pass frequency (bpf) – which is in the infrasound region.

If you take the definition of AM, then for a wind turbine having a carrier frequency of say the gearbox output shaft (a low audible frequency) then that signal can have variations in time at the bpf. Different wind speeds and different blade angles will give different levels of variation in the amplitude of that frequency. The pulsations should be there all the time the blades are turning.

The loading on the gearbox will change as the rotor runs around and these pulses will get transferred through the gearbox. (Whether it is the blade passing the tower or the change in wind loading along the blade to create a specific point where there is equal pressure along the blade will still give pulses).

So the gearbox output shaft speed is the carrier frequency and the modulation is the blade pass frequency. A narrow-band analysis of the frequencies around this gearbox output shaft speed will show sidebands spaced at multiples of the blade pass frequency. That is classic AM.

In the Cape Bridgewater study, when the turbines are producing power there is a peak around 31.5 Hz. At Capital WF it is 25.5 Hz (different turbines). FFT [fast Fourier transform] for both turbines show the gearbox output frequency with sidebands. That is AM as defined.

Q: So you say that turbines do produce amplitude modulation (AM) but it is only related to low frequency, and the modulation occurs at an infrasound rate?

Cooper: Correct.

Q: But isn’t this ILFN [infrasound and low-frequency noise]?

Cooper: No, it is LFN being modulated at an infrasound rate, so it is not infrasound as a sound.

Q: But the swish noise is not low-frequency, is it?

Cooper: Correct. And this leads to the next point. The modulation of the dB(A) value cannot be AM because it is a broad-band noise. Hence this is the issue of terminology.

If you take the swish noise in the regions of 800 Hz to 2 kHz there are no discrete frequencies and therefore that noise cannot be AM. You can use 1/3-octave bands to see the modulation of the amplitude, modulated at the bpf. But not AM by definition.

The modulation of the amplitude for the remaining frequencies in the audible range of the acoustic signature (such as the swish or thump) is not AM. This modulation is defined by Zwicker and Fastl [Psychoacoustics: Facts and Models, 1999] as “fluctuation” because the modulation rate is less than 10 Hz. Zwicker and Fastl say people sense the modulation by the hearing mechanism, but may not hear it.

Q: You are saying that in Psychoacoustics: Facts and Models, Professors Zwicker and Fastl identified “fluctuation” as the modulation of a sound where the modulation occurs at an infrasound rate – but specifically below 10 Hz?

Cooper: Yes. They show that the ear has a particular sensitivity to the rate of modulation around 4 Hz and that the modulation has an excellent correlation between speech and the hearing system.

Q: So acousticians have had knowledge of modulation for a while? Does this modulation cause a greater degree of annoyance?

Cooper: Correct. Leventhal (in the UK) provided a report in 2004 (for DEFRA) that cited work by Bradley (in 1994) on modulation of low-frequency broad-band noise at an infrasound rate to significantly increase the annoyance of the broad-band noise. But it seems to have been ignored. Leventhal has stated in his evidence in Australia (a long time ago) that in relation to annoyance of wind farms, it is not infrasound but modulation of low frequency.

Q: Doesn’t this tie in with Rand and Ambrose’s work at Falmouth, where the issue was the pressure pulsations occurring at an infrasound rate?

Cooper: Yes, and this ties in with Prof Salt’s work on the inner ear. My pulsation analysis concept came after Prof Salt’s work using pure tones. Prof Salt’s work shows the mechanism where the ear can respond to the fluctuations – being modulation at an infrasound rate.

Q: In the UK there has been a discussion on excessive AM giving rise to a greater level of disturbance, but it uses the dB(A) value which as discussed above is not really AM?

Cooper: Right. What they have used is the modulation of the dB(A) level at an infrasound rate. It is simply incorrect terminology. Being an electrical engineer first permits me to understand AM and filter theory.

Q: Your AM paper discusses a modification of one format for deriving AM used in the UK and the modulation index.

Cooper: Yes. A number of years ago we developed a method to show the variation in the wind turbine signal over time that showed the pulsations, amplitude modulation and frequency modulation. It helped acousticians understand the time-varying nature of the signal. In the synthesis paper there is a link back to our website that has examples of narrow-band and 1/3-octave band “movies”.

The UK method is time consuming in determining the modulation index.

We used the underlying analysis concept for our “movies” to determine the statistical analysis in 1/3-octaves for multiple 10-minute samples of wind turbine noise and derive the modulation index of the A-weighted peaks in the spectrum by taking the L1 level minus the L90 level (to agree with parameters used in Australia). The results agree with the labour-intensive method of manually determining the modulation index using multiple sets of 10-second samples.

I presented the modulation index information for the various wind turbine noise samples (and non-wind turbine samples) in the brainwave paper. The third paper on the brainwave monitoring is just a pilot study but identified the impact on the frontal lobes.

Q: Yes: the third paper is different, and whilst only a single person test pilot study it would seem to show the automatic response of the brain to the presence of inaudible sound.

Cooper: Yes. The paper can only discuss the acoustic content of the testing. It was the next step from the work presented in ASA (American Society of Acoustics) in New Orleans and Euronoise and the brain wave results presented to the audience are not in the power point for public viewing as they are the results from the psychologist who undertook the measurement and analysis.

Q: Well, there is a lot to digest in your three papers given at the ICA (International Congress on Acoustics) and I see that we have to consider a change in terminology as more is learned about wind turbine noise. I see that your work is practical field work using real wind turbine noise and you are still undertaking this work with no funding. Could you summarise the key technical points from your latest work?

Cooper: There are other researchers working at the ‘coal face’ in getting to the bottom of the wind turbine noise issue and their work is important and blends in with my investigations.

I suggest that with respect to the description of wind turbine noise it is a matter of terminology that needs a shift as follows:

As a result of work I conducted in 2013 as to detected by residents of the operation of the turbines when the FFT fast Fourier transform] levels in the region of 4–6 Hz were above 50 dB that led to the re-discovery of Kelley’s work of the 1980’s, one can expand Zwicker and Fastl’s work, and Bradley’s work, to determine the annoyance adjustment for wind turbine noise. That is one of the next projects I am looking at.


We sincerely thank Mr. Steven Cooper for again explaining not only his technical work, but also for his diligence in pursuing the in-depth nature of the complexities of thepeculiaritiesof wind turbine “noise,” and relating those to actual real time experiences of people living near wind turbine factories.

Above all, noted is the volunteerism of Mr. Cooper’s ongoing research. We emphasize that Mr. Cooper’s re-definitions and research augment the measurement of ILFN (infrasound and low-frequency noise) and do not in any manner diminish the validity of the general discussion in relation to impacts. The shift in identifying the correct terminology and definition of amplitude modulation and fluctuation complements the work of other notable acousticians and researchers (as did Cooper’s identification of sensation and his double-blind studies on inaudible wind turbine noise).

The impacts are repeatedly proven to seriously impact persons, and animals. Cooper’s recent articles reinforce the groundbreaking work of Zwicker and Fastl (which has been forgotten/ignored by the wind industry) and that of Kelley (which had also been ignored by the wind industry until Rand, Ambrose, James and Cooper “rediscovered” the Kelley work in 2014).

[Originally published at Master Resource: Part 1 & Part 2.]

July 30, 2020: Health Effects of Industrial Wind: The Debate Intensifies (update with Steven Cooper)

Click here for computer translation

Click here for computer translation