Wind Farm Noise: Measurement, Assessment, and Control

This book is dedicated to our families without whose patience it may not have been completed.

There are three sides to every story: your side, my side and the truth. And no-one is lying.

Robert Evans, an American film producer born June 29, 1930.

If we knew what it was we were doing, it wouldn't be called research, would it?

Albert Einstein commenting on research.

Clever is the person who believes half of what he hears. Brilliant is the person who chooses the right half to believe.…

Wiley Series in Acoustics, Noise and Vibration

This book series will embrace a wide spectrum of acoustics, noise and vibration topics from theoretical foundations to real world applications. Individual volumes will range from specialist works of science to advanced undergraduate and graduate student texts. Books in the series will review scientific principles of acoustics, describe special research studies and discuss solutions for noise and vibration problems in communities, industry and transportation.

The first books in the series include those on Biomedical Ultrasound; Effects of Sound on People, Engineering Acoustics, Noise and Vibration Control, Environmental Noise Management; Sound Intensity and Windfarm Noise. Books on a wide variety of related topics.

The books I edited for Wiley, the Encyclopedia of Acoustics (1997), the Handbook of Acoustics (1998) and the Handbook of Noise and Vibration Control (2007) included over 400 chapters written by different authors. Each author had to restrict their chapter length on their special topics to no more than about 10 pages. The books in the current series will allow authors to provide much more in-depth coverage of their topic.

The series will be of interest to senior undergraduate and graduate students, consultants, and researchers in acoustics, noise and vibration and in particular those involved in engineering and scientific fields, including, aerospace, automotive, biomedical, civil/structural, electrical, environmental, industrial, materials, naval architecture and mechanical systems. In addition the books will be of interest to practitioners and researchers in fields such as audiology, architecture, the environment, physics, signal processing and speech.

Malcolm J. Crocker

Series Editor

Preface

Wind farm noise has polarised communities and is featured on numerous web sites that either dismiss its effects on people as a nocebo effect or as something in their imagination. There are just as many other web sites that claim wind farm noise has led to serious medical problems in some people and that infrasound generated by wind farms can have far-reaching consequences for the health of people who are exposed. These web sites can be found easily by typing ‘wind farm noise’ into any internet search engine.

Our intention when writing this book has been to cover all aspects of wind farm noise, including how it is generated, how it propagates, how it is assessed, how it is regulated and what effects it has on people living in the vicinity of wind turbines. Where aspects of wind farm noise are controversial, we have presented what we believe to be an unbiased assessment of the facts. None of the three authors have ever worked for the wind farm industry nor have they been members of any anti-wind-farm organisation. Only the first author has appeared as an expert witness, in a 2010 court proceedings concerned with a wind farm development. This was his only involvement in court proceedings and it was in the capacity of being asked to critique a report prepared by an acoustical consultant for a wind farm operator.

The first two authors have been chief investigators on a number of research projects, funded by the Australian Research Council, on aerodynamic noise generation and the impact of wind farm noise on rural communities. The first author has also spent over 40 years teaching, researching and consulting in acoustics and noise control. The second author has spent nearly 20 years working in the area of aerospace engineering, with a strong focus on aeroacoustics: the science of how objects like rotor blades create sound. Following completion of a PhD in fluid mechanics, the third author has spend the past four years measuring and analysing wind farm noise.

Wind farm noise is a very controversial subject, in that it has been used as a reason for delaying many wind farm projects that together are worth billions of dollars. Most court cases find in favour of the wind farm developer and very few wind farms are prevented from being constructed as a result of court proceedings based on excessive noise, although sometimes the turbine layout has had to be modified to minimise noise impacts on the surrounding communities. Nevertheless, even after wind farms have been constructed, many people complain of the noise keeping them awake at night and causing them to feel ill. In spite of the many reported cases ofadverse effects of wind farm noise on people, wind farm proponents insist that wind farm noise is so low in level that it could not possibly be a problem. They often imply that affected people must be developing symptoms as a result of feelings of jealousy over payments received by wind turbine hosts or as a result of anti-wind-farm publicity telling them that wind farms produce such symptoms. Although we are neither pro- nor anti-wind-farm campaigners, we do believe that some people in the vicinity of some wind farms are badly affected by the noise and that further research into this phenomenon is absolutely essential.

We hope that you the reader find the material in this book useful and, where it strays into areas that are controversial, that you find that we have achieved our aim of presenting a balanced point of view.

Colin Hansen

Con Doolan

Kristy Hansen

Adelaide, Australia

Chapter 1

Wind Energy and Noise

1.1 Introduction

Why write this book about noise generated by wind farms? Many people believe that wind farm noise is a non-issue and that people complain about it because they are unhappy with the lack of financial compensation they receive compared to their neighbours who are hosting the turbines. Other reasons that we often see on pro-wind-farm web sites are that the anti-wind-farm lobby has suggested a range of symptoms are caused by wind farms and that this suggestion has made some people living near wind farms develop these symptoms as a result: the ‘so-called’ nocebo effect. Although the authors of this book would consider themselves neither pro- nor anti-wind-farms, they have taken a sufficient number of their own measurements and spoken to a sufficient number of residents living in the vicinity of wind farms (including wind farm hosts) to appreciate that the character and level of wind farm noise is a problem for a significant number of people, even those who reside at distances of 3 km or more from the nearest turbine.

Although one chapter in this book is concerned with the effects of wind farm noise on people, the main focus of this book is on how wind farm noise is generated and propagated, the characteristics of the noise arriving at residences in the vicinity of wind farms, and measurement procedures and instrumentation, as well as assessment criteria that are necessary for properly quantifying the noise. As many people living in the vicinity of wind farms report ‘feeling’ vibration when they lie down, vibration generation, propagation and measurement are also discussed in sections in Chapters 4, 5 and 6.

To lay the foundation for the remaining chapters, the rest of this chapter is concerned with a description of how the wind industry has developed in various countries, followed by a brief history of noise studies (including a summary of noise levels generated by large wind turbines), a summary of some public inquiries and wind farm noise regulations, and finally a discussion of the current consensus on wind farm noise and its effects on people.

It is not possible to usefully take part in the wind farm noise debate without having some understanding of acoustics. This is the reason for writing Chapter 2 to follow. First, basic concepts in acoustics necessary for understanding the legislation are discussed. This is followed by a discussion of the fundamentals of frequency analysis, which is an important tool for analysing wind farm noise. Chapter 2 concludes with a discussion of some advanced concepts of frequency analysis, an understanding of which is essential for practitioners wishing to undertake more advanced analyses of wind farm noise.

Chapter 3 contains an overview of how wind turbines generate noise, while Chapter 4 is about estimating wind turbine sound power levels. Chapter 5 is concerned with using turbine sound power levels and sound propagation models to estimate noise levels in the community. Several propagation models are considered, beginning with the simplest and progressing to the more complex and supposedly more accurate models. Chapter 6 is devoted to a detailed description of procedures and instrumentation for the measurement of wind farm noise and vibration, and includes a discussion of potential errors associated with such measurements. The chapter also includes a discussion on wind tunnel measurements for testing turbine models. Chapter 7 is about the effects of wind farm noise on people, Chapter 8 contains a discussion of various options that can reduce wind turbine noise, both outside of and inside residences, and Chapter 9 contains some suggestions of where we should be heading in terms of wind farm noise research and the reduction of its effects on people.

1.2 Development of the Wind Energy Industry

1.2.1 Early Development Prior to 2000

Mankind has harvested energy from the wind for over a thousand years. The first device designed for this purpose was a vertical-axis, sail-type windmill developed in Persia between 500 and 900 AD. This design appears to have been inspired by boats that used their sails to harness the wind for propulsion. Windmills have been primarily used for water pumping and grain grinding, with the mechanical power developed in the rotating shaft used directly to drive a pump or turn a grindstone. Wind turbines differ from windmills in that they convert the mechanical power into electrical power through use of a generator. They also have a smaller number of blades, since windmills require high torque at low rotor speeds (Manwell et al. 2009); for optimal electrical power generation higher rotor speeds and thus fewer blades are desirable. This is because high rotor speeds result in increased loading and reducing the number of blades reduces stresses on the rotor (Manwell et al. 2009). Another factor to consider is that wind turbine blades are very costly and therefore it is beneficial to minimise their number.

The most common wind turbine configuration that is used today is a horizontal-axis wind turbine (HAWT) and this book will concentrate on aspects of noise associated with this particular design, with a focus on large, industrial-scale wind turbines. The major components of a HAWT are shown in Figure 1.1. The basic principle of operation is that wind causes the blades to rotate and the rotor drives a shaft that is connected, generally via a gearbox, to a generator, which converts the rotational energy into electrical energy.

Figure 1.1 Schematic of typical wind turbine: LE, leading edge; TE, trailing edge.

The power output and rotational speed of a HAWT can be controlled either by designing the blades such that they begin to stall at a certain wind speed (stall control) or by having a mechanism and control system that is able to vary the blade pitch (pitch control, which involves rotation of the blades about the blade axis as opposed to the rotor axis). In a pitch-controlled turbine, the controller will continually adjust the blade pitch to ensure that the power output is optimised for the wind speed being experienced by the blade. A pitch controlled machine can also be easily ‘turned off’ to protect the turbine when the wind speed becomes too great. This is done by adjusting the pitch of the blades so that they no longer generate appreciable lift. A stall-controlled turbine blade is designed with some twist to ensure the blade stalls gradually along its length. The blade profile also has to be designed so that it stalls just as the wind speed becomes too high, thus reducing the lift force acting on the blade, which in turn limits the blade speed and power. An active stall-controlled turbine is similar to a pitch-controlled turbine in that the pitch is continually adjusted to optimise the power output. However, when the wind speed becomes too great, the stall-controlled turbine will rotate the blades so that they stall, as opposed to a pitch-controlled turbine, which rotates the blades in the opposite direction so that the lift is minimised. In some cases, turbines are also controlled using yaw control. This involves turning the rotor so the blades no longer face directly into the wind. However, this is only used on small turbines and is not relevant to the turbines that are the subject of this book.

Development of large HAWTs for incorporation into electric utilities first began in the early 1930s with the construction of the Balaklava wind turbine in Russia, which was 30 m in diameter, two-bladed and rated to a power of 100 kW. This turbine operated for around two years and generated 200 MWh (Sektorov 1934). In the late 1930s, development of the first megawatt-scale wind turbine began in the USA in a collaborative project between an engineer named Palmer C. Putnam and the Smith company, which was experienced in the construction of hydroelectric turbines and electrical power equipment. The Smith–Putnam HAWT consisted of a two-bladed rotor of diameter 53.3 m, mounted on a truss-type tower at a rotor-axis height of 33.5 m (Putnam 1948). This wind turbine was rated at 1.25 MW and included a number of technological innovations such as blade-pitch control, flapping hinges on the blades to reduce dynamic loading on the shaft, and active yaw control (Spera 2009). Several weeks of continuous operation yielded excellent power production and it was demonstrated that the wind turbine was capable of being inserted into the grid. Unfortunately, development was discontinued in 1945 when a faulty blade spar separated at the repair weld and there were insufficient funds to continue the project.

Over the next 25 years, development proceeded at a modest rate, taking place predominantly in Western Europe, where there was a temporary post-war shortage of fossil fuels that led to increased energy prices. Two HAWT designs emerged from Denmark and Germany during this time, and these would form the basis of future wind turbine development in the 1970s. The 24-m diameter, 200 kW Gedser Mill wind turbine was constructed in Denmark and was designed by Johannes Juul. The rotor consisted of three fixed-pitch blades that were connected with a support frame to improve structural integrity. This frame was removed in later years when the metal blades were replaced with fibreglass ones (Dodge 2006). The rotor was located upwind of the concrete tower and the design was notable for its simplicity, ruggedness and reliability. This wind turbine supplied AC power to the local utility from 1958 until 1967, achieving annual capacity factors of 20% in some years (Spera 2009). The annual capacity factor is defined as the ratio of the energy generated in one year to the amount that could be generated if the turbines operated continuously at their maximum power output. In 1967, a mechanical failure resulted in discontinued use of the wind turbine and the machine remained idle for the next 10 years (Auer 2013).

Figure 1.2 Schematic of rotor showing ability to teeter.

Considerable research effort, with a focus on improved rotor technology, led to the development of the Hütter–Allgaier wind turbine in Germany in the early 1960s. With a diameter of 34 m and rated at 100 kW, it was technologically advanced for its time and included an important design feature of a bearing at the rotor hub that allowed the rotor to ‘teeter’, in order to minimise the dynamic loading that results from the changes in gyroscopic inertia about the tower axis that arise when the blades of a two-bladed rotor move between the horizontal and vertical positions. A teetering rotor is illustrated schematically in Figure 1.2, which shows the bearing that facilitates the teetering motion. Despite its technological proficiency, the Hütter–Allgaier wind turbine encountered flutter in its long, slender blades, which slowed research progress.

Wind turbines were successfully connected to the grid in France in the period from 1958 to 1964 and the largest such turbine was called the Type Neyrpic, which was 35 m in diameter and rated at 1.1 MW. While this wind turbine demonstrated good performance, its operation was terminated abruptly when the turbine shaft broke.

In the UK, a number of unique 100 kW wind turbine designs were conceived and built in the 1950s with the intention of local grid connection. These turbines operated successfully for a few years, but technical and environmental factors led to the cessation of operations by 1963. Many projects were discontinued during this 25-year period due to technical issues and adverse weather conditions that resulted in expensive failures. These issues were not investigated further at this time due to a lack of interest in funding research into alternative energy sources, which was directly related to the availability of inexpensive fossil fuels and nuclear resources. Therefore, the Smith–Putnam wind turbine remained the largest in the world until the oil crisis of the 1970s prompted further development in the wind industry.

In the late 1970s, centres were established in Denmark, Germany and the Netherlands for testing of experimental and commercial wind turbines. These centres were also responsible for certification programs for tax or subsidy benefits to ensure that wind turbines met defined standards before entering the market. The International Energy Agency (IEA) was also established in the mid-1970s to encourage cooperation between Western countries on research, policy and development on wind energy. By the early 1990s, several countries had developed wind turbines with power ratings in the megawatt range, including Canada, Denmark, Germany, Italy, the Netherlands, Spain, Sweden, the UK and the USA (International Energy Agency 1989).

Canada pursued a different approach to most countries in the design and construction of megawatt-scale wind turbines, electing to focus on a Darrieus-type vertical-axis wind turbine. The Eolé Darrieus wind turbine was 64 m in diameter, 96 m in height, rated at 4 MW and was completed in 1987 (Richards 1987). Despite being rated at 4 MW, the power was limited to 2.5 MW to increase the lifespan of the turbine. The Eolé was connected to the Hydro-Quebec grid and it operated for over 30 000 h until 1993, generating over 12 GWh of electricity during its lifetime (Tong 2010). It was stopped due to damage to its expensive lower bearing.

Wind turbine development in Denmark proceeded at a modest rate and the size of turbines increased incrementally (Gipe 1995). Two upwind prototypes with a rated power of 630 kW, called Nibe A and B, were constructed in the early 1980s based on a similar design concept to the Gedser wind turbine. Nibe A was stall-controlled whereas Nibe B was pitch-controlled (International Energy Agency 1989), thus enabling a performance comparison to be made between these control mechanisms. The prototypes operated for 15 years, providing a wealth of information that contributed to later development of the wind industry in Denmark (Spera 2009).

The 54-m diameter, 2 MW Tvind wind turbine was a three-bladed downwind machine built by teachers and students from the Tvind school who collaborated with consultants, sub-contractors, volunteers and experts such as Professor Ulrich Hütter. Hütter's influence was evident in the choice of a downwind design (turbine blades downwind of the support tower) and the advanced blade technology. It was later discovered that the downwind configuration resulted in excessive low-frequency noise generation, as will be discussed in Section 1.3. The combination of design principles incorporated into the Gedser-type wind turbines such as the upwind, heavy, three-bladed, asynchronous generator and Hütter/Tvind advanced blade design and root assembly led to a successful combination that influenced future wind turbine development (Maegaard 2013).

In 1982, Germany embarked on an ambitious project to build the 100-m diameter Growian wind turbine, rated at 3 MW, which was the largest HAWT at the time. This wind turbine was a two-bladed downwind machine, which incorporated some of the latest technological innovations including full-span pitch control, carbon filament blades, a tubular steel tower and variable-speed operation. These features would later prove to be successful, but the overall design was over-ambitious for its time and was soon dismantled. In 1991, the 3 MW Aeolus II wind turbine, with a diameter of 80 m, was developed as a collaborative project between Germany and Sweden. In this two-bladed upwind design, advanced blade technology enabled a reduction in weight from 22 tons (existing Swedish Aeolus) to 6 tons (International Energy Agency 1989).

The two-bladed, 1.5 MW Gamma 60 upwind turbine constructed in Italy in 1991 was distinctive for its active yaw control, which provided a means of power regulation above rated speed. The 66-m diameter Gamma 60 design also incorporated a teetered hub as well as a direct current link between the synchronous generator and the step-up transformer. The innovative features of this wind turbine contributed to increased annual energy production, as well as eliminating control components on rotating parts to reduce complexity, thus decreasing manufacturing and maintenance costs (International Energy Agency 1989).

In the Netherlands, development was focussed on the 0.2–1.0 MW range and a number of demonstration projects were initiated by utility companies. The 1 MW NEWECS-45 wind turbine was developed in 1986 and consisted of a 45-m diameter rotor in an upwind configuration. The rotor design consisted of two-blades on which full-span pitch control was implemented and the tower was constructed from tubular steel.

Collaboration between Germany and Spain led to the successful development of the AWEC-60 wind turbine in 1989, which was a 1.2 MW, 60-m diameter, three-bladed upwind machine based in Spain. Whilst the design was based on the German 1.2 MW, WKA-60 wind turbine, further development was undertaken on the electrical system, glass-fibre reinforced blades and the control system in order to reduce the cost (International Energy Agency 1989).

Development of large-scale wind turbines was launched rapidly and successfully in Sweden in the 1980s. The WTS-75 was a two-bladed upwind turbine rated at 2 MW, which possessed some unique design features including a drive-train system with bevelled gears that eliminated the need for power slip rings and a mechanism for raising and lowering all major components. Sweden also constructed the WTS-3 downwind turbine, rated at 3 MW, which produced a relatively large amount of energy compared with other large-scale wind turbines of the 1980s. This design incorporated a teetered hub and a spring-mounted gearbox to reduce the impact of dynamic loading associated with the two-bladed, downwind configuration.

After a number of design iterations, the UK produced the 3 MW LS-1 in 1987, which was a two-bladed upwind turbine of 60 m diameter. The rotor consisted of a teetered hub mounted on elastomeric bearings and the outer 30% of the blades were mounted on rolling element bearings, which enabled variation of the blade pitch angle (Hau 2013). The drive-train design provided control of the rotor speed to within c01-math-001 5% (Hau 2013).

Following the Arab oil embargo of 1973, the US government invested significant funds into a federal research plan directed towards wind energy development. The first large wind turbine that was developed as part of this program was the MOD-0 configuration in 1975. Over the next decade, this wind turbine was used extensively for testing to identify possible improvements that could be made to the design. It was 38.1 m in diameter, rated at 100 kW and was mounted atop a truss-type tower at a hub height of 30.5 m. The design was similar to the Smith–Putman and Hütter–Allgaier wind turbine designs, where the rotor was two-bladed and located downwind of the tower. Several modifications were made to the MOD-0 over its twelve-year lifetime, including replacement of the truss-type tower with a slender shell tower to reduce wake-induced fatigue loads and incorporation of a teetered hub. The extensive testing that was undertaken also resulted in a large volume of documentation, computer models and control algorithms that form the basis of modern wind turbine technology. During the early stages of the MOD-0 program, an upgraded version of this turbine, rated at 200 kW, was integrated into the grid at four separate locations and designated the MOD-0A. The locations were chosen to ensure that wind power would make up a significant proportion of the input power to the grid, enabling grid connectivity issues to be identified. The wind turbines collectively fed 3.6 GWh into their respective grids during their operating lives (Shaltens and Birchenough 1983) and achieved capacity factors as high as 0.48 (Spera 2009).

The first megawatt-scale wind turbine to be developed as part of the US federal research plan was the MOD-1 configuration in 1979. This model was designed before the problems with the MOD-0 had been identified and understood and therefore was dismantled after only two years of operation. The MOD-1 was 61 m in diameter, rated at 2 MW and resembled the MOD-0 configuration in that it had a downwind, two-bladed rotor, rigidly mounted on a truss-type tower. While this turbine was successfully integrated into the local grid, impulsive loading caused by the substantial wake deficit behind the truss tower resulted in a severe risk of early fatigue as well as environmental problems such as excessive low-frequency noise and electromagnetic interference. The next turbine in the series was the 91.4-m diameter, 2.5 MW MOD-2, which was an upwind design and represented a large technical leap from the earlier models in the US federal plan. The design employed partial-span pitch control on its two blades, which simplified the use of a teetered hub, as only the outer portion of the blades needed to rotate about the blade axis to enable control. A comprehensive testing program was carried out on the MOD-2 design, including investigation of wake-interaction effects, operation strategies and control algorithms. Successful integration into the local grid was also realised, with a group of three MOD-2 wind turbines installed at Goodnoe Hills in 1981, and contributing over 10 GWh to the local grid during 16 000 h of operational time. This group of three formed the first ‘wind farm’ in the world (Boeing 2015) proving conclusively that groups of wind turbines could operate in a completely automated mode.

The MOD-5B was the next wind turbine in the series to be developed under the federal wind energy program. The design was similar to the MOD-2 wind turbine, in that it consisted of a two-bladed upwind rotor, teetered hub, partial-span pitch control and a tubular steel tower. However, this wind turbine was larger, with a rotor diameter of 97.5 m and a rated power of 3.2 MW. Built in 1987, it was the first large-scale wind turbine to operate at variable speed, which led to improved efficiency and reduced structural loading (Spera 2009). From 1988, the MOD-5B wind turbine was connected to the grid on the island of Oahu in Hawaii and was fully automatic, with software changes made using the local public telephone system (Boeing 2015). The MOD-5B demonstrated excellent performance for such a large and advanced design, and during its lifetime of six years, it ran for 20 561 h and produced 26.8 GWh of electricity.

The largest wind turbine to be built before the year 2000 was the WTS-4, which was a two-bladed downwind machine, rated at 4 MW, with a hub height of 80.4 m and a diameter of 78 m. The support tower was a single 12-sided cylindrical structure of shell construction. Despite its large power rating, this wind turbine produced a relatively small amount of energy during its lifetime (Spera 2009) and it ceased operating after only four years due to a generator failure. The wind turbine was bought for a fraction of its original cost by a local engineer and wind energy enthusiast, who later watched the machine fly to pieces in a storm (Righter 1996).

1.2.2 Development since 2000

Wind power has expanded rapidly since the beginning of the 21st century to the point where there are so many different models of megawatt-rated wind turbines that further consideration of individual models is beyond the scope of this book. The rapid expansion is a result of increased awareness of global warming and eventual fossil fuel depletion, as well as rising concerns over energy security. The amount of global energy generated since the year 2000, as plotted in Figure 1.3, has consistently increased at an average annual rate of approximately 25% and was 17 times higher in 2013 than in 2000.

Figure 1.3 Global annual energy output (TWh).

When describing the relative contribution of wind energy, it is common to refer to the installed power, which is the product of the manufacturer's power rating and the number of turbines. This is also referred to as the installed capacity. However, this measure does not take into account such factors as wind variability, interactions between wind turbines, lack of grid connectivity and wind turbine malfunctions. Therefore, a more conservative measure is the actual energy generated, which is measured in TWh (terawatt-hours) for large-scale turbines. The annual capacity factor is the ratio of the energy generated over one year to the amount that would be generated if the wind farm operated continuously at its maximum power output. This gives an indication of the overall efficiency of wind energy as a whole.

Figures 1.4–1.6 show the installed power capacity (the rated turbine power in megawatts multiplied by the number of turbines in the wind farm), generated energy (TWh) and annual capacity factor, respectively, for the top ten countries in terms of wind energy generated in 2013. The data in these figures have been compiled from information provided by the IEA, Global Wind Energy Council and the US Energy Information Administration. Germany was leading the world with installed power and generated wind energy in the year 2000 and since then has been increasing its installed power at a rate close to linear. On the other hand, the USA and China's installed power and generated wind energy has been increasing exponentially since the year 2000. As a result, these two countries have emerged as leaders in available and generated wind power as of 2013/2014. Comparison of the figures for the USA and China reveals that although China has almost double the installed power capacity of the USA, the latter still generates a larger amount of energy. It is therefore not surprising that the annual capacity factor for China is lowest for the ten countries compared in Figure 1.6. Conversely, the USA has the highest annual capacity factor, and has been the world leader for energy generation since 2007.

Figure 1.4 Installed capacity of wind power (2000–2014).

Figure 1.5 Generated wind energy (2000–2014).

Figure 1.6 Annual capacity factor (2000–2014).

Technological development of wind turbines has focussed on reduction of costs, increased energy capture and greater reliability. To this end, wind turbines have become progressively larger to take advantage of the high-energy winds that occur at greater altitudes. Several advances have also been made in blade design. These include optimising the blade profile to increase efficiency in low winds, limiting aerodynamic loads in high winds and minimising blade fouling. Advanced composite materials have also been specified in blade designs in place of steel and wood, to improve the strength-to-weight ratio. Most large wind turbines today are variable-speed, pitch-regulated machines, which allows operation at near-optimum ratios between the blade-tip speed and wind speed, thus maximising output power. Modern designs are also predominantly three-bladed, as this number provides the best compromise between aerodynamic efficiency, cost, rotational mass, structural integrity, inertial stability, relatively low tip-speed ratios and aesthetics. The use of fewer blades results in increased aerodynamic efficiency since each blade disturbs the air for the following one. The cost and weight of each blade is substantial, so from this perspective fewer blades are also preferred. Also, the strength and stiffness of each blade is greater when there are fewer blades for a given rotor solidity (total blade planform area divided by swept area). On the other hand, wind turbines with less than three blades experience unbalanced loading during yaw, which can be overcome by using a teetered hub, although this is an extra complication that most manufacturers prefer to avoid. The rotational speed of a three-bladed design is lower than a one- or two-bladed design, resulting in lower tip-speed ratios and hence reduced trailing-edge noise. Many people prefer to look at turbines with three blades rather than one or two blades and since community acceptance is important for wind farm developers, this point is also taken into account.

The drive trains of wind turbines have become lighter and more reliable in recent times, which is an important development, since failure of drive-train components such as the gearbox is costly and the associated downtime is high (Ekwaro-Osire et al. 2011). In the late 1990s, direct-drive generators were introduced as an alternative to gearboxes, but despite their numerous advantages, their size and weight are issues that have prevented widespread use (Spera 2009). These days, the generator components of wind turbines are required to synchronise with electricity grids and they are therefore capable of producing AC electricity, in contrast to early wind turbines, which were developed as stand-alone units and employed DC generators. Control systems for wind turbines have become more sophisticated in recent times as well, with high-speed digital controls enabling processing of data from a number of sensors for optimised power generation. Advanced control algorithms have been developed to facilitate more efficient data processing and optimal actuator responses to sensor inputs. While a number of early wind turbine designs integrated steel truss-type or concrete monopole (single support cylinder or partial-cone) towers into their designs, modern wind turbines consist of a steel, monopole structure with a reinforced concrete foundation.

1.2.3 Support Received by the Wind Industry

Wherever wind energy has been developed successfully, it has been with the aid of government intervention in the form of financial, technical or regulatory support. The reason for this is that, at the time of writing, wind energy is more expensive than energy derived from coal or gas and the industry would be non-viable without financial incentives from governments. However, with many renewable energy targets in place around the globe, it seems that wind power is the least expensive way of achieving them. Of course, it appears that wind turbines are a very clean and environmentally friendly power source, as power is generated without producing greenhouse gases. But are they? To answer this question, one must consider the greenhouse gases that are produced during the construction of wind farms, from transportation of materials to the construction site, and during their maintenance and decommissioning. It is also important to consider the intermittency of wind power and the current lack of energy storage facilities, resulting in significant security and reliability concerns for electrical grids worldwide (Miskelly 2012). Through analysis of power output data provided by the Australian Energy Market Operator, Miskelly (2012) demonstrated that during the full calendar year of 2010, there were over 100 incidences where the entire Eastern Australian grid generated less than 2% of installed capacity. A consequence of these common-mode failures is the need for a rapid response from fossil-fuel-driven power stations, resulting in inefficient operation of these facilities and production of excessive greenhouse emissions at these times.

Recently Weißbach et al. (2013) compared wind energy with other energy generation facilities in terms of its energy return on investment (EROI) value and the number of years to achieve payback on the energy invested in construction, and his results are presented in Table 1.1. The EROI value is the ratio of the usable energy that the energy facility returns during its lifetime to all the invested energy needed to generate this energy. Weißbach further analysed the EROI value in terms of the cost of buffering needed to maintain a continuous power supply, considering the unreliability of the energy source. These values are also included, but must be considered in light of the economic threshold for the EROI value being about 7 (Weißbach et al. 2013), indicating that wind power produces considerably more energy than needed to construct and run the turbines, but that intermittency of supply makes it economically non-viable.

Table 1.1 EROI and energy payback times for various energy generation facilities

Data from Weißbach et al. (2013).

In Table 1.1, CSP is concentrated solar power, such as achieved by an array of mirrors directed at the apex of a tower or a large array of flat or parabolic reflectors. In the case of the more expensive but more efficient parabolic reflectors, sunlight is focussed onto a receiver tube at the focal point of the reflectors, thus heating molten salt as it flows through the tube. In all cases the heat energy generated is used to boil water to drive a steam turbine, which in turn drives a generator to produce electricity. CCGT refers to a combined-cycle gas turbine facility, in which waste heat from the gas turbine is used to generate steam to power a steam turbine, with both turbines driving an electrical generator.

1.3 History of Wind Turbine Noise Studies

Here, some of the earliest reported studies concerning wind farm noise are explored. A large proportion of the work reported here was carried out in the 1980s in response to a noise issue associated with operation of the MOD-1 wind turbine. This was the first well-documented case of acoustic disturbance from a wind turbine that was significant enough to provoke complaints from neighbours. While the noise issue was exacerbated by the fact that the rotor of the MOD-1 was located downwind from the tower, it was shown that a similar mechanism was at play for an upwind rotor (Spencer 1981). The main difference between the two rotor configurations was shown to be the magnitude of the flow deficit encountered by the blades, and consequently the level of noise generated, which is much greater for the downwind configuration (Spencer 1981). Another difference, described by Kelley et al. (1985), is that the blades of a downwind rotor experience transient lift fluctuations due to the periodic vortex shedding that occurs behind the support tower, although the effect is smaller than the flow deficit effect. These differences in the blade inflow conditions cause the noise levels associated with a downwind configuration to be much higher. On the other hand, since upwind turbines also experience a flow deficit as well as inflow turbulence, many of the findings from the studies on downwind turbines are still relevant to modern upwind wind turbine designs.

As mentioned in Section 1.2.1, the MOD-1 was a downwind machine with a two-bladed rotor that was rigidly mounted on a truss-type tower. Detailed investigations carried out by Kelley and his colleagues culminated in a comprehensive report, which identified the issue as unsteady loading imparted to the rotor blades as they passed through the tower wake (Kelley et al. 1985). This phenomenon resulted in high levels of low-frequency impulsive noise that excited structural resonances and interior air volume modes of nearby houses, sometimes causing loose objects to vibrate (Kelley et al. 1985). Measurements indicated that the impulsive character of the noise was directly related to the presence of blade-pass frequency components. Noise propagation was found to be governed by a combination of atmospheric refraction and terrain reflection, which were responsible for focussing the noise towards locations occupied by residences. Due to noise complaints received from about a dozen families living within a 3-km radius of the wind turbine, MOD-1 was slowed down from 35 to 23 RPM and it was found that an 11-dB reduction in sound pressure levels could be achieved (Viterna 1981). On the other hand, there was a corresponding increase in the level of impulsive noise in the 8 and 16 Hz octave bands and while annoyance was reduced, it was not eliminated (Kelley et al. 1988). Noise issues with the MOD-1 turbine prompted a number of investigations on this specific configuration, including field measurements, modelling and wind tunnel experiments.

A predictive model for determining the amplitude of the blade-pass harmonics was developed by Viterna (1981) and implemented in computer software called WTSOUND. The approach was based on theory for aircraft propellers first developed by Gutin in 1937 (Gutin 1948). In summary, the process developed by Viterna (1981) involved the following steps:

1. Calculating the steady aerodynamic blade forces.

2. Determining the variation in these forces due to unsteady aerodynamics.

3. Carrying out a Fourier analysis of the force variation.

4. Calculating sound pressure levels in the acoustic field, by assuming the aerodynamic source to be compact with an effective radius of 75% of the blade span.

The calculated results were in good agreement with the MOD-1 data in the vicinity of two rotor diameters from the wind turbine. However, in the far field, the model underestimated the actual levels of the MOD-1 by 6 dB or more due to propagation effects related to the terrain and atmospheric conditions. Nonetheless, the model accurately recreated the c01-math-002 spectrum shape characteristic of a pulse of finite width in the time domain, where c01-math-003 is the frequency. For the MOD-1 wind turbine, this pulse of finite length and relatively steep edges resulted from the blade lifting surface passing through a flow deficit. Metzger and Klatte (1981) found that the spectrum envelope was very sensitive to the shape of the flow deficit and that harmonics of the blade-pass frequency in the higher frequency range could be avoided by ensuring that the shape followed a Gaussian profile. One possibility for achieving this was by modifying the tower shape in the vicinity of the rotor blades. However, this would only be effective for one wind direction unless the tower had a lightweight external shell that could rotate as the wind direction changed (Tocci and Marcus 1982).

A model of the MOD-1 wind turbine was constructed and tested in the anechoic wind tunnel at the NASA Langley Research Centre. Researchers carefully scaled the tower details to ensure that the wake would be recreated as accurately as possible (Greene 1981). Results from these experimental studies indicated that the impulsive noise associated with the MOD-1 wind turbine could be significantly reduced by using an upwind configuration. Therefore, one of the primary motives for using an upwind configuration in the MOD-2 wind turbine design was to avoid the impulsive noise issues that were associated with its predecessor (Kelley et al. 1988). The acoustic emissions of the MOD-2 wind turbine were investigated extensively by Kelley et al. (1988) and it was found that the impulsive noise was significantly reduced. Further reduction in the levels and impulsiveness of the low-frequency noise emitted by the MOD-2 machine was achieved by incorporating vortex generators and pitch schedule changes. It is worth noting that the impulsive noise was not eliminated entirely and that the degree of impulsiveness was strongly correlated with the vertical atmospheric stability, the vertical or upwash turbulence length scale and the blade loading (Kelley et al. 1988). Spencer (1981) found that the spectrum shape of a MOD-2 wind turbine with a two-bladed upwind rotor was very similar to the MOD-2 with a downwind rotor in the frequency range from 0 to 45 Hz. The main difference between these spectra was the relative amplitude of the blade-pass harmonics, since the flow deficit associated with the downwind case was much larger than the flow deficit for the upwind turbine.

Apart from impulsive noise generated by blade–tower interaction, a number of other aeroacoustic noise sources associated with wind turbine operation were identified and modelled in the late 1980s and early 1990s. The investigated sources were mainly broadband in nature and resulted from inflow turbulence and airfoil self-noise. Grosveld (1985) found good agreement between predictions of broadband noise and far-field measurements in the vicinity of the two-bladed MOD-OA, MOD-2 and WTS-4 wind turbines and the three-bladed US Windpower Inc. wind turbine. The prediction model considered contributions from inflow turbulence to the rotor, trailing-edge effects and the wake due to a blunt trailing edge and it was found that at low frequencies the dominant source was inflow turbulence noise (Grosveld 1985). Glegg et al. (1987) developed a prediction method for wind turbines that included the source mechanisms of unsteady lift noise, unsteady thickness noise, trailing-edge noise and the noise from separated flow. To determine the inflow turbulence, which is a required input for the unsteady lift and thickness calculations, a detailed model of the atmospheric boundary layer was implemented. Good agreement was obtained between the atmospheric boundary layer model and anemometer measurements, but a 10-dB discrepancy was noted between the measured and calculated acoustic results. Improved correspondence between measurements and predictions was achieved by assuming a turbulence length scale equal to the blade chord. The authors also observed that the presence of the tower on an upwind turbine caused significant acoustic scattering when the rotor blades were close to the tower and hence this effect was also incorporated into their theoretical model (Glegg et al. 1987). However, due to the short duration of this effect, it was found to have a negligible contribution to the average level. On the other hand, the authors noted that it would increase the detectability of the signal.

A review of the aeroacoustic noise generated by large wind turbines was presented by Hubbard and Shepherd (1991) and an additional mechanism of impulsive noise generation was attributed to rotor inflow velocity gradients. Various wind velocity profiles were assumed as inputs to the model developed by Viterna (1981) and the results were compared to measurements recorded up to 80 m from the two-bladed WWG-0600 upwind turbine. There was good agreement between the results, for a specific assumed atmospheric wind velocity profile resulting from the atmospheric boundary layer. On the other hand, the actual wind velocity profile was not measured and therefore it is not known if the assumed velocity profile was accurate.

Propagation of noise from the MOD-1 wind turbine was investigated in detail by Thompson (1982) through analysis of field measurements and development of a computational model. The results indicated that the primary mechanism responsible for enhanced far-field noise levels was atmospheric refraction of acoustic waves caused by vertical wind shear. The influence of ground and surface wave propagation on the enhanced noise levels was found to be negligible in comparison to this effect. Based on the collected data, conditions of adverse noise propagation were predicted to occur about 30% of the time at complex terrain sites. A similar investigation was carried out by Willshire (1985) and Willshire Jr and Zorumski (1987) on the WTS-4 wind turbine. It was shown that low-frequency sound was refracted in the downwind direction, resulting in an attenuation rate of 3 dB per doubling of distance for frequencies below 20 Hz. Predictions of both ray tracing and normal-mode theoretical models supported this observation. In the upwind direction, the absence of a shadow zone was noted for these infrasonic frequencies and the propagating signals indicated a spherical spreading characteristic, resulting in an attenuation of 6 dB/doubling of distance.

The acoustic and vibratory response of buildings to wind farm noise was explored by Stephens et al. (1982) and it was shown that in some circumstances, low-frequency wind turbine noise could be perceived more readily indoors than outdoors. A number of reasons for this phenomenon were presented, including selective attenuation of higher frequencies by the building structure, room modes, structural resonances and noise-induced vibrations. Other complicating factors mentioned included the role of stiffness and air leaks at low frequencies (Stephens et al. 1982). Enhanced perception of indoor low-frequency noise was attributed to an increase in the indoor noise level relative to the outside level at specific frequencies and large variations in sound pressure level as a function of room position (Hubbard and Shepherd 1991).

Thresholds of perception were determined by Stephens et al. (1981) by exposing subjects to a range of impulsive stimuli of the type associated with blade–tower interaction. The test stimuli were synthesised based on MOD-1 data and blade–tower interaction calculations and presented to the listening subjects in an anechoic chamber. The resulting spectra consisted of harmonics of the fundamental frequencies of 0.5 Hz and 1 Hz that were dominated by specific frequencies. The resulting perception thresholds were found to be lower than the pure-tone threshold and it was observed that the chosen fundamental frequency influenced the results. A lower fundamental frequency gave rise to a lower perception threshold. In a later publication (Stephens et al. 1981), the authors presented additional perception curves for use when various levels of background noise were present in addition to the impulsive wind turbine signal. These curves indicated that the perception threshold increased with the level of background noise but that this increase was relatively less for lower frequencies. For a background level of 35 dBA, the perception threshold was still below the pure-tone threshold at all frequencies, according to the ISO389-7 (2005). Comparison of various metrics used in the assessment of low-frequency noise was carried out by Kelley (1987). Evaluators were exposed to simulated signals characteristic of wind turbine noise emissions and they subsequently recorded their perception of the noise according to specified categories and rankings. The researchers then determined the correlation between the stimulus sequences and the evaluators' responses. It was found that people reacted to a low-frequency noise environment and that the A-weighting (see Section 2.2.11) is not an adequate measure of annoyance when low frequencies are dominant (Kelley 1987). The results also indicated that the low-frequency sound level (LSL) (Tokita et al. 1984) and C-weighting (see Section 2.2.11) metrics were the most ‘efficient’ descriptors of low-frequency noise annoyance.

A method to control the impulsive noise associated with the flow deficit encountered by the wind turbine blades was proposed by Tocci and Marcus (1982). The method involved reduction of the flow deficit through use of airfoil-shaped fairings for the tower, which could be rotated into the appropriate direction to minimise the wake deficit. This technique would also reduce the flow deficit in front of the tower and is therefore relevant for reducing aerodynamic noise associated with blade–tower interaction for modern upwind turbines.

1.3.1 Modern Wind Turbine Sound Power Levels

Readers unfamiliar with acoustics terminology should consult Chapter 2 prior to reading further. More detail on the estimation of turbine sound power is provided in Chapter 4.

The noise output of many modern wind turbines has been reported by Søndergaard (2013) in the form of overall A-weighted sound power levels in dB re c01-math-004 Watts. He showed that the total A-weighted sound power level of turbines with a rated power greater than 2 MW could be described by Eq. (1.1) (within c01-math-005 5 dB).

1.1

where c01-math-007 is the turbine rated power in kilowatts.

Søndergaard (2013) also provided low-frequency data in the range 10–160 Hz and showed that the A-weighted sound power level of turbines with a rated power greater than 2 MW could be described by Eq. (1.2) (within c01-math-008 dB).

1.2

The data provided by Søndergaard (2013) were used to derive the relationships in Eqs. (1.1) and (1.2), which show that as the turbine rated power increases, so too does the noise it produces over the entire frequency spectrum, with low-frequency noise increasing by the same amount as mid- and high-frequency noise.

The relative frequency distribution of the noise for turbines with a rated power greater than 2 MW is shown in Figure 1.7, where the decibel difference between the 1/3-octave band sound power level and the total sound power level is plotted.

Figure 1.7 Wind turbine 1/3-octave band A-weighted sound power level (dB) normalised to the total A-weighted sound power level, for turbines with a power rating greater than 2 MW. The error bars show the 95% confidence interval around the mean. Data from Søndergaard (2013).

An interesting result from Sondergard's analysis is that modern pitch-RPM regulated turbines do not produce any more noise as the hub height wind speed increases above about 8 m/s. However, it is quite a different story for stall- and active-stall-regulated turbines. For these turbines, the noise output increases markedly as the wind speed increases above 8 m/s, by anything from 8 to 12 dBA for a wind speed increase to 12 m/s. The 8 m/s wind speed was chosen because this is the wind speed at which all turbines reach 95% of their rated power.

1.4 Current Wind Farm Noise Guidelines and Assessment Procedures

Worldwide, there is a plethora of standards, guidelines and recommended assessment procedures for wind farm noise and its allowable limits. As many as possible are reviewed here but unfortunately it is not possible to capture every single one that has been published. Here we are concerned with assessment procedures and allowable limits—standards that concern sound power measurements or sound propagation predictions are discussed in relevant sections elsewhere in this book and only the parts relevant to allowable limits and assessment guidelines are discussed here.

In the following paragraphs and later in the book, the term wind shear is used quite often, so it is useful to discuss its meaning here. Wind shear is a way of describing the variation of wind speed with altitude. For high wind shear conditions, the wind speed increases much more rapidly with altitude than it does under low wind shear conditions. Low wind shear generally occurs during the day, when solar radiation heats the ground and causes mixing in the atmospheric boundary layer within a few hundred metres of the ground, resulting in only a small variation of wind speed with altitude in the boundary layer. At night, in the absence of solar radiation, there is reduced mixing in the boundary layer and it is quite common for wind turbines to be generating a significant amount of power while there is little or no wind close to the ground. One method of quantifying wind shear is with the use of a shear exponent (or coefficient), c01-math-010 , which relates wind speed to altitude. The ratio of the wind speed, c01-math-011 , at height, c01-math-012 , to the wind speed c01-math-013 at height c01-math-014 is given by,

1.3 equation

where the wind shear coefficient can vary from 0.1 to 0.6, depending on the ground surface roughness and the atmospheric stability. This is discussed in much more detail in Section 5.2.4.

1.4.1 ETSU-R-97 (used mainly in the UK and Ireland)

ETSU-R-97, a detailed report published in 1996 (ETSU 1996), was perhaps the first government-sponsored work written with the specific purpose of providing detailed planning guidance for the assessment and rating of noise from wind farms. It was prepared by a working group sponsored by the UK Department of Transport and Industry. It follows on from a UK Department of Planning Policy Guidance note (PPG22) published in 1993, which was written at a time when not much was known about wind farm noise and its effects on people, although there was a considerable body of work available from NASA on noise emissions and health effects from the much noisier downwind turbines. ETSU-R-97 is important in that many existing local guidelines for wind farm noise, both in the UK and other countries, have been based on it. However, much more has been discovered about wind farm noise, its characteristics and its effects on people in the past two decades, so the guidance in ETSU-R-97 has become outdated and inappropriate in many areas. The main problems with the recommendations in ETSU-R-97 are listed below.

1. The recommended use of the c01-math-016 metric for measurement of wind farm noise, which is the sound pressure level exceeded 90% of the time in a 10-min period, is based on the assumption that wind farms produce a steady noise at residences, in other words, one that does not vary very much in level over a 10-min period. Many more recent studies, including those of the authors of this book, have shown this to not be the case and although many current guidelines use the more acceptable c01-math-017 metric, there are still a number that continue to use the c01-math-018 metric. The c01-math-019 metric represents a noise energy average and generally produces wind farm noise levels in the range 1.5 to 3 dB above those obtained using the c01-math-020 metric. However, the c01-math-021 metric is generally considered appropriate for measurement of background noise prior to construction of the wind farm, to minimise contributions from transient noise sources that are not representative of actual background noise levels occurring for the majority of the time. Both of these metrics are defined in Section 2.2.12.

2. ETSU-R-97 recommends that the acceptable nighttime limit for noise should be 43 dBA, based on the outdated WHO guideline that 35 dBA indoors is an acceptable level for people trying to sleep. The WHO has since updated their acceptable indoor noise level for sleep to 30 dBA (with windows open), but most guidelines based on ETSU-R-97 do not account for this change. It is strange that the recommended limit for nighttime is greater than the 35 to 40 dBA limit for daytime and this has received the appropriate criticism in the literature (Cox et al. 2012).

3. Allowable levels (day and night) can be increased to 45 dBA for residents receiving financial benefit as a result of the wind farm.

4. The noise-level limits described above do not account for the low-frequency nature of wind farm noise by the time it reaches the bedrooms in residences more than 1–1.5 km from the nearest turbine.

5. ETSU-R-97 states that noise from a wind farm should be limited to 5 dBA above background noise levels. However, it is stated that this limit should not apply in environments where the daytime background noise levels are less than the limits mentioned in items 2 and 3 above. It was assumed that audible noise above background is not disturbing to people living in low-noise environments, which is why this aspect of ETSU-R-97 has drawn considerable criticism. However, it is clear that an industrial noise source of 35 or 40 dBA superimposed on a typical rural environment of 20 dBA or less would be intrusive to most people.

6. The allowance of 5 dBA above background is based on use of the c01-math-022 metric for measurement of wind farm noise, rather than the usual c01-math-023 metric, which means that when comparing with other standards, the allowed increase of c01-math-024 over background c01-math-025 is actually at least 7 dBA, and could be more, depending on the difference between c01-math-026 and c01-math-027 for wind farm noise. This then exceeds the limit beyond which complaints may be expected.

7. There is no consideration of the common nighttime situation where wind shear is high and there is no (or very little) wind at a residence, while at the same time there is sufficient wind to drive