Problems in assessing the noise of wind turbines

In both Ireland and the UK, wind farms are assessed using a set of noise limits broadly varying with wind speed. In both countries there is some flexibility between individual cases but in practice this is quite small. Most countries similarly have noise limits of some kind. The result is that the decision-maker simply needs to know whether the application complies with the limits or not.

By Dick Bowdler FIOA, specialist noise consultant

The 2022 WSP report (1) in discussing the ‘binary test’ in ETSU-R-97 points out the planning policy in England ‘is based on a hierarchy of effects thresholds’, rather than one simple limit. Nearly all other disciplines in an environmental assessment are described on a hierarchical or graduated scale in terms of their impact so that it is apparent how significant the impact is at each sensitive location. That leaves the decision-maker to assess the extent of the impact of noise together with all the other impacts as a whole. The purpose of this article is to set out the mechanism of the planning system in such a way that it identifies the options for assessing wind turbine noise.

Decision-making process
• A planning decision is a balance. Environmental impact assessments (EIA) are based on the principle that the developer sets out the impact of their development on the one hand and the need for the development on the other.
• Other parties may present different opinions on the impact and on the need and then the decision-maker weighs one against the other.
• While smaller projects may not need an EIA the principle of an assessment to describe the impact and the need, and the principle of balance, are the same. It is not the case that neighbours to a development would necessarily feel no impact. Almost all developments have an impact on somebody or on some part of the environment. The difference between the binary test and the graduated assessment lies in where the decision is made and what local and site-specific factors are taken into account. The binary test is devised in a generic manner prior to any planning decision and so the planning decision cannot take account of specific local factors.

Graduated assessment
The process in the graduated assessment is that the significance of the impact is calculated and described by specialists in the particular field – by consultants (or perhaps academics) who have studied the work that has been done on the relationship between the features of the development and the impact on people. On the other hand the decision of where the balance lies – what is the extent of the impact that would be acceptable bearing in mind the need for this particular development – is made at a political level on the basis of government or local government policy or a combination of both.

With noise, the impact depends on many factors. It depends on:
• the noise level;
• the various characteristics of the noise;
• the sensitivity of the receiver;
• the pre-existing soundscape at the receiver; and
• factors that are non-acoustic.

Some of these at least, are specific to the local environment and local circumstances, so when it comes to the debate at local level as to whether a particular development should be approved, all these things can be tabled and each side can present their case. The Institute of Environmental Management and Assessment (IEMA) produces Guidelines for Environmental Noise Impact Assessment which make clear that an assessment under ETSU-R-97 is insufficient and that the impact should be identified.

Clocaenog Inquiry,
The inspector in the Clocaenog Inquiry, where the noise impact was argued by the local authority as well as compliance with ETSU-R-97, said that “There would be significant change to the visual and aural amenities of a number of residents in the vicinity of the site” to the extent that their human rights would be breached. Though she concluded that the “interference with the human rights of the occupants of the three properties would be proportionate and justified in the public interest”. Such arguments made in detail and in public bring the decision-making process into the open. The big advantage of taking the final decision on an individual basis is transparency. Those who argue that ‘our site is a special case’ would have the advantage of putting that case. Arguably, that is a better and fairer method of making a decision than using a simple binary formula such as the set of limits in ETSU-R-97.

Above: Wind turbines in County Cork

Wind farm decisions
When it comes to wind farms in the UK and in Ireland, we don’t look at individual cases because we have ETSU-R-97 and WEDG2006. These are binary tests that are devised to provide a balance between the need and the impact. ETSU-R-97 was published in 1996, though it was not firmly seen as part of government guidance until the early 2000s, it provides a methodology for deriving a set of noise limits above which the application fails and below which it passes.

The limits seek to take account of the planning balance. This is made clear in paragraph 1 of the Executive Summary of ETSU-R-97 where it says that it ‘ … gives indicative noise levels thought to offer a reasonable degree of protection to wind farm neighbours, without placing unreasonable restrictions on wind farm development or adding unduly to the costs and administrative burdens on wind farm developers or local authorities’. ETSU-R-97 was devised by first using BS4142 to suggest that 5dB above background noise should be the limit. The edition of BS4142 current at the time said that a difference of 5dB was ‘ of marginal significance ’. Then there are separate absolute lower limits for day and for night, although there are some variations to the lower limit for daytime depending rather vaguely on:
• background noise;
• number of properties; and
• size of the development.

The daytime lower limit is based on a somewhat tortuous discussion about whether one can sleep on one’s patio on a summer’s day. The nighttime lower limit is set at 43dB (LA90) which is derived from the earlier WHO indoor nighttime limit of 35dB to avoid sleep disturbance, which was superseded in 1999. The lower limit means that in very quiet areas it is quite possible to find turbine noise levels in excess of 10dB above background noise levels in both night and day. It is not clear how WEDG 2006 in Ireland was developed though some parts are clearly derived from ETSU-R-97. The primary statement is that the turbine noise should be limited to 5dB above background noise or a lower limit of 45dB, though nighttime should have a lower limit of 43dB. However, in low noise environments where background noise is less than 30dB the lower limit should be between 35 and 40dB. In setting these levels it says that ‘An appropriate balance must be achieved between power generation and noise impact’. The Irish Government is looking for an acceptable revision to this following two drafts in 2013 and 2019 which were, by all accounts, largely rejected by stakeholders from all sides.

So both country’s guidelines provide a methodology for deriving noise limits, one for each wind speed for night and for day, above which the wind farm is not permitted and below which it is permitted. The consequence of this method is that nearly all decisions regarding noise in wind farms are effectively decided before any hearing or inquiry because it is agreed by all sides that the noise limits are met – or can be met with appropriate mitigation which is feasible. Most of the other aspects of wind farm applications – landscape, heritage and so on, are set out as a description of the impact of the development which can take account of all the detailed local circumstances. Politicians or their representatives in the form of inspectors then make the decision as to whether the application will be allowed. Arguably, it must be correct that the decision is made by politicians because they are the policy-makers and it is they that must make the decision on the balance not acoustic consultants or academics.

It is also, perhaps, a reasonable argument that the use of ETSU-R-97 is a political decision because it has been endorsed by government for use with wind farms. There are two main questions that we need to ask. Most of the basis of ETSU-R-97 has been superseded or, even, was not reliable at the time; technology has moved on so do we need a revision? If so, do we need to move to the description of the impact in line with most other disciplines, or is it possible to develop generic guidance that produces a fair and equitable result in all cases. 

What is required for a graduated assessment
Let us look and see if it is realistic to adopt a graduated assessment for turbine noise. First, we must ask whether we have enough knowledge to be able to create a structure with which to assign levels of significance to all the aspects of the noise in the pre-existing soundscape? We need to know the degree of impact that different levels of turbine noise make on people. We do know something about the number of people annoyed by different absolute levels of wind turbine noise – the WHO Environmental Noise Guidelines for the European Region in 2018 used ‘annoyance’ to establish their guidance, setting the threshold at 10% highly annoyed. It said that the recommendation was conditional in that the quality of the data available was low. However, extensive data from Canada (2), too late to be included in the WHO study, has broadly confirmed the recommendation.

Unfortunately the WHO recommendation uses Lden which most consultants would probably feel has little relevance to the impact of the turbines on people though the original data is mostly based on the LAeq down-wind of the turbine – so effectively the worst case. The original data used to develop the WHO recommendation, including the Canadian data, can be related to percentage of people ‘highly annoyed’ and is around 39-40dB (LAeq) worst case downwind level for 10% highly annoyed.

But we have no evidence that annoyance is the right measure to use. According to Ollson (3) ‘There is at best only a weak association between wind turbine sound at the exterior of people’s homes and the reported level of annoyance. The results of European, Canadian and U.S. studies suggest that only 9-13% of the annoyance can be correlated to the sound level itself’. But if not annoyance, there is nothing else available that we can use. What is the measure that we should use at nighttime? Is it ‘annoyance’ again or should it be ‘sleep disturbance’, and what measure of sleep disturbance do we use?

Whether we use annoyance or some other impact there is the question of whether we use absolute noise or relate the specific noise to the background noise – or a combination of both as in ETSU-R-97. In ETSU-R-97 this was originally based on BS4142:1990 which describes the likelihood of complaints. The latest version of BS4142 says that a margin of 5dB ‘ is likely to be an indication of an adverse impact, depending on the context’ so it specifically measures the impact. An adaptation of BS4142 could be used to produce a graduated assessment. An advantage of BS4142 is that it uses the rating level so it can take account of tonal noise and, perhaps with some adaptation, AM. A disadvantage is that, even though it has now been around in one form or another for nearly 60 years, it has no evidence base to back it up, though it might be argued that it appears to have been used largely successfully for that period. Perhaps the answer is to use both the absolute and relative methods.

Tonal noise and a penalty for it has been in ETSU-R-97 from the beginning and there are other methodologies which can effectively assess the additional impact of tonal noise. The methodology for measuring audibility of tones is now formalised in IEC TS 61400-11 (4) though there is no measure of the impact. However, there are examples of available penalty systems, including that in ETSU-R-97. Amplitude modulation (AM) has become a more prominent concern as time has passed since the publication of ETSU-R-97. A method for measuring AM has also now been formalised in IEC TS 61400‑11‑2. We do have some information about the sort of penalty that might be applied for AM, for example, as summarised by Lotinga (5).

The problem with penalties, is that they do not readily translate into graduations of effect though one could foresee that a methodology could be devised to do so. The alternative would be to add any penalties to the absolute level and then try to relate the rating level rather than merely the sound level to percentage highly annoyed or whatever measure is used.

There are other issues that might be relevant at a local level such as whether sensitivity varies according to time of day or day of the week or whether the type of pre-existing soundscape influences the impact. While there is no obvious, readily available framework for making an assessment of the impact of turbine noise or for identifying LOAEL and SOAEL there are methodologies that could be assembled into one whole to provide such a framework.

Above: ETSU-R-97 was devised by first using BS4142 to suggest that 5dB above background noise should be the limit. The daytime lower limit is based on a somewhat tortuous discussion about whether one can sleep on one’s patio on a summer’s day

Conclusion
There are several relationships we can use to describe the impact of wind farm noise on neighbours though there are many gaps in our knowledge. But that is little different from assessing any type of noise and possibly little different from the position with other disciplines in the planning process. On balance, it seems possible to construct an acceptable methodology that could be used to achieve a reasonable planning balance. Arguably the graduated method is more transparent in that it is clearer how decisions are made and it produces a more tailored decision specifically applicable to each application.

On the other hand, it might be preferable to carry on using a binary method either by reviewing ETSU-R-977 or preferably starting again from scratch. Where the binary method is used it has the advantage that it is clear to all sides what the noise levels need to be before any planning decision is made. However, to adopt the same level of transparency and equity we need to know as much to develop a binary methodology as we do to develop a graduated methodology. Further work is certainly needed to establish relationships between noise and its effects whichever method is used but it would be unrealistic to suggest we should do nothing.

Reminders
• The difference between the binary test and the graduated assessment lies in where the decision is made and what local and site-specific factors are taken into account.
• The process in the graduated assessment is that the significance of the impact is calculated and described by specialists in the particular field who have studied the work that has been done on the relationship between the features of the development and the impact on people.
• Arguably the graduated method is more transparent in that it is clearer how decisions are made and it produces a more tailored decision specifically applicable to each application.

References
1 A Review of Noise Guidance for Onshore Wind Turbines – WSP, October 2022
2 Various papers by David Michaud et Al, Health Canada.
3 Establishing Sound Limits for Wind Energy: What is the Role of Annoyance? Christopher Ollson, Mark Bastasch. 9th International Conference on Wind Turbine Noise.
4 IEC TS 61400‑11‑2:2024: Acoustic noise measurement techniques — Measurement of wind turbine sound characteristics in receptor position.
5 Subjective responses to wind turbine noise amplitude modulation: pooled analysis of laboratory listening studies and synthesis of an AM character rating penalty. Michael J B Lotinga, Toby Lewis.