Cox, Unwin, Sherwin. Where ETSU is Silent – Wind Turbine Noise

Wind Turbine Noise Impact Assessment – Where ETSU is Silent

Richard Cox, David Unwin, Trevor Sherman

“A rigorous review of how wind turbine noise is assessed is long overdue. The issues raised have far reaching implications for the industry and the entire nation”. Chris Heaton-Harris MP (Daventry, Con)

Executive Summary and Recommendations

Sections 2 and 3 of this Report examine the de facto standard method, ETSU‐R‐97, used for assessing the likely noise nuisance from wind turbines that was produced in the mid‐1990s and is usually referred to as ‘ETSU’. The basis of the assessment is that background noise will mask the turbine noise and that both the background noise and turbine noise increase with increasing wind speed. To test whether or not the noise will be within acceptable limits, ETSU requires applicants to measure the background noise at sites at risk, correlate this with the simultaneous wind speeds measured at the proposed turbines’ site, make model predictions of the likely turbine noise and then compare these two sets of values with a series of derived noise limits. If the wind farm gains planning consent, the imposed planning conditions include the noise limits and complaints procedure that apply for the life of the wind farm. Although it might be assumed that the science on which the ETSU processes rests is sound and that the data and analyses presented can be relied upon, we demonstrate this is not true.

Section 2 of this Report deals with the background to the ETSU guidance where it shows that it was originally defined by a small group of acoustics consultants (see Appendix D), many of whom have subsequently appeared as authors of the noise assessments embedded in developers’ environmental impact statements, and at a time (1997) when the turbines were much smaller than currently being installed. This was also before some of the adverse effects of turbine noise had been fully recognised. Analysis of some 67 planning inquiry decision letters since 2009 [Ref: 17] shows that likely noise nuisance has never been cited as the sole reason to disallow an application and in only a few cases has it been cited as a contributing reason for such a decision. At first sight it would thus seem that the ETSU process is reliable and that concerns about wind farm noise are misplaced. However, this Report adduces evidence to suggest that this view cannot be sustained. Uncertainties that are intrinsic to the process mean that it is far less reliable than has hitherto been assumed. Much of the evidence presented in support of this view relates to parts of the process where the ETSU document does not provide sufficient guidance, in fact to places where it is remarkably ‘silent’. Consequently, most if not all wind farm applicants have incorrectly interpreted the ETSU guidance and presented invalid noise assessments.

Section 3 examines sources of uncertainty in the measurement of the background noise upon which the entire process rests. It is shown that there are serious issues of sampling in time (is the period for which data collected truly representative?) and space (have the most at risk locations been adequately monitored or even identified?). An important finding is that, despite early advice to the contrary (see Appendix C) and statements in the ETSU document [Ref: 01] itself, it is likely that possibly all recent noise assessments use data collected from microphones that are inadequately shielded, such that the data are contaminated by the noise of the wind itself. This has two effects. First it raises the level of background noise used as reference and, second, it increases the apparent correlation with wind speed. In planning terms this will allow consents to place turbines that are closer to the nearest at risk properties than ETSU originally intended.

Section 4 and Appendix A examine a second area where ETSU is also silent and where current practice fails to recognise unavoidable intrinsic uncertainty. ETSU asks that a ‘best fit curve’ is fitted to the background noise data plotted against wind speed in order to determine the average reference background noise over the range of wind speeds. It is from these curves that the noise limits are derived. It is shown that no acoustic theory guides the choice of curve so that typically quadratic, cubic, or even quartic polynomial functions all seem to provide a reasonable statistical fit to the data and are used interchangeably. It is shown that, although it can add an uncertainty of at least ±2dB to the values used as reference, the choice of function is more‐or‐less arbitrary. Almost any curve threaded through the data could have been used. In two real examples the uncertainty is shown to be at least ±2dB and it is most evident at moderate winds. As far as the authors are aware, this is the first time this aspect of the ETSU process has been examined and it is additional to issues to do with the establishment of background noise where the variation dominated by some other factor, as for example the time of day at properties close to major traffic routes.

Section 5 deals with uncertainties that arise from hypothetical (‘model’) calculations of the turbine noise at the ‘at risk’ locations and times. Typically, developers do not specify which model of turbine is to be installed relying instead on noise emission data for a ‘candidate turbine’. Here it is shown that there is uncertainty in the ways by which manufacturers report the emitted noise, between different ‘candidate turbines’, and in all probability engineering variations between turbines of nominally the same design. These data are then input into a standard noise prediction model defined in ISO9613‐2 and used, together with essentially arbitrary assumptions about the absorption of the noise en route, to predict the noise at the at risk locations. Even the ISO9613‐2 model itself recognises at least ±3dB of calculation uncertainty in these predictions, but in any specific implementation the engineering and turbine choice variability will add to this perhaps ±2dB giving a total uncertainty in this part of the process of around ±4dB.

Section 6 and Appendix B looks at the effects of wind shear, which is shown to be higher at some sites than would have been assumed. Wind shear is the change of wind speed and direction with height above ground level caused by a combination of ground roughness and atmospheric stability. Although seemingly aware of the issue, ETSU is silent on how this is to be factored into noise assessments. High wind shear is the condition when the wind speed at upper elevations is much higher than at lower elevations. When using just a single wind speed at some reference height to predict both the turbine and background noise it is necessary to make an assumption about the rate at which this change with height takes place. If high shear is present and not factored into the predictions, the result is to underestimate wind speed at the turbine hub height and to overestimate it at near ground level. The effects are that we have more noise emitted and less masking of this noise near ground level than under low shear conditions. Analysis of the available data for the proposed wind farm at Winwick (Northants) and at nearby sites shows that high levels of shear occur for around 10% of the time, with a particular concentration during the evenings and at night. Noise assessments by wind farm developers have either failed to consider wind shear or applied very low levels of correction such that they consistently under estimate the likely noise nuisance. In common with other recent authors, this Report argues that the so‐ called ‘Article Method’ for considering wind shear being promulgated by some acousticians is not only unnecessarily complex but it leads to an even more permissive regime than currently in force. This is being seized upon by the wind power consulting community since it would allow them to demonstrate justification for reduced separation distances between turbines and houses.

Section 7 examines uncertainty. Noise assessment is no different from any other area of science or engineering in that any process requiring measurements and modelling will be subject to error and uncertainty. Recognising that such errors can work in different directions, by adding or subtracting, it is suggested that a sound difference of ±10dB from prediction could easily occur that will either double or halve the noise loudness received and is additional to the errors arising from microphone wind induced noise and wind shear.

Section 8 summarises the growing evidence related to excess amplitude modulation (EAM) that also seems to be associated with times of high wind shear. The phenomenon is compared to the well‐understood and analogous one of ‘blade slap’ associated with helicopters. It is concluded that the ‘standard’ condition applied at planning appeals is ineffective at detecting the pulsing nature of EAM and that only by measuring using a much shorter sampling time period can it reliably be detected.

Finally in Section 9 a summary is presented of some other reasons that have been suggested why ETSU should be reviewed. There are continuing debates about the use of the LA90 measurement rather than the others such as the LA50, over the use of the ‘A’ rather than the ‘C’ weighting, and over the appropriateness of the levels of the minimum day time noise limit of 35–40dB (A) and more particularly the higher night time limit of 43dB (A) that is out of line with World Health Organisation sleep disturbance limits.

Appendix A: Background Noise Data Analysis Appendix B: Wind Shear Case Study Appendix C: Malcolm Hayes blog – wind noise at microphones Appendix D: Noise Guidance and Working Groups Appendix E: Hansard Written Answers 19 June 2012 Appendix F: Court of Appeal; Hulme – Rix


The main recommendations that arise from the presented evidence in this report are:

  • There is a compelling case for retrospective background noise surveys to be repeated in properly controlled conditions using suitable wind screens to enable more representative noise limits to be established. There is a strong case for commissioning a truly independent controlled study to assess the full magnitude of the effect. We propose that this study is completed at three classes of onshore wind farm site: 1) where permission is granted and the turbines have not been constructed; 2) at sites where there have been noise complaints; 3) at a sample of operating sites.
  • It is clear that ETSU as currently defined is simply not fit for the purpose for which it was intended. It should be subjected to independent review from a truly independent multi‐disciplinary working group of experts who have no financial interest in wind farm development and under the auspices of a government agency whose objectives do not include the encouragement of renewable energy. The Institute of Acoustics wind farm working group currently preparing technical guidance for DECC on the implementation of ETSU as reported by Charles Hendry MP on 19 June 2012 (Appendix E) clearly does not meet these requirements.

10 July 2012

Download the Report →