Tong W. Wind Power Generation and Wind Turbine Design

Подождите немного. Документ загружается.

Wind Turbine Acoustics 155

ref

10 log

⎛⎞

=×

⎜⎟

⎝⎠

( 3)

where the reference pressure is p

−

= 2 × 10

Pa. The sound pressure level can be

expressed also in terms of the sound intensity, which is deﬁ ned as the sound power

transmitted through unit area and far from the source can be written as

(4)

where r

and c

are the undisturbed air density and sound speed, respectively. Thus

the sound pressure level can be written as

ref

10 log

⎛⎞

=×

⎜⎟

⎝⎠

(5 )

where the reference sound intensity is I

ref

= 10

− 12

W/m

. To characterize the total

power of a sound source, one has to integrate the sound intensity over a closed

surface surrounding the source

PIS=

∫

( 6 )

Similarly to the sound pressure level, one can deﬁ ne the sound power level as

ref

10 log

⎛⎞

=×

⎜⎟

⎝⎠

(7 )

where the reference power level is P

ref

= 10

− 12

W. Thus for a point source with

spherical wave spreading the sound intensity at a distance kR is I / k

, I being the

sound intensity at a distance R from the source (see Fig. 1 ). It is important to note,

that the sound power level, L

, is a characteristic of the sound source, it reﬂ ects

Figure 1: Illustration of sound intensity.

156 Wind Power Generation and Wind Turbine Design

the amount of acoustic energy generated by the source. In contrary, the sound pres-

sure level, L

depends on the observer position and quantiﬁ es the amount of sound

energy reaching the observer.

Pure tones are rarely found in practice, most devices emit sound composed by a

set of different frequencies:

'( , ) cos 2

ptx A

⎡⎤

⎛⎞

=±

⎢⎥

⎜⎟

⎝⎠

⎣⎦

∑

(8 )

Thus, the frequency spectrum of the acoustic pressure has to be determined as

well to characterize the sound. There are three types of commonly used spectra:

narrowband, 1/3 octave band and 1/1 octave band. In this context band refers to a

given frequency interval over which the amplitudes are averaged. For an octave band

the upper limiting frequency is exactly the double of the lower limiting frequency.

For a 1/3 octave band the ratio of the upper and lower limiting frequencies is 2

1/3

For a narrow-band frequency the width of the bands is constant and ‘small’ enough

to capture pure tones and thus gives the most details about the spectrum. Figure 2

Figure 2: Narrowband (top), 1/3 octave band (middle) and 1/1 octave band

(bottom) spectra of the same acoustic pressure signal.

Wind Turbine Acoustics 157

shows examples of narrowband, 1/3 octave band and 1/1 octave band spectra of the

same acoustic pressure signal.

Human ear is sensitive to sound in a range of about 20 − 30 Hz and up to 20 kHz.

This range varies among individuals and changes in time with reduction of the

range with age. Additionally, the sensitivity is frequency-dependent. For sounds

with frequencies of the order of 3 − 4 kHz the sensitivity is the highest, the thresh-

old of hearing being around 0 dB, while at low frequencies it might be required

a sound pressure level of 40 dB to be heard [ 4 ]. To account for this uneven sen-

sitivity of the ear, weighted sound pressure/sound power levels have been intro-

duced, the most commonly used being the A-weighted sound pressure level,

computed as

[][][]

pA p A

dB( ) dB dBLAL L=+

(9)

where the weighting function varies as it is shown in Fig. 3 . A-weighting is suitable

for most applications. In cases where low-frequency noise is dominant, B or C weight-

ing might be more appropriate. These weightings are similar to the A-weighting,

just the shape of the weighting curve is different.

4 Wind turbine noise

Acoustic studies involve three main stages: sound generation, propagation and

reception (see Fig. 4 ). In the case of wind turbines there are several different

kinds of noise sources which are discussed in Section 4.1. The generated noise

is propagating through the air, this propagation being inﬂ uenced by the air prop-

erties, the landscape, vegetation, presence of different obstacles, etc. These

issues are discussed in Section 4.2. The third stage is the sound immission, i.e.

the perception of the sound at an observer position. Beside the sound pressure

level there are other parameters as well which inﬂ uence the perception of the

Figure 3: A-weighting function.

158 Wind Power Generation and Wind Turbine Design

sound as acceptable or annoying. Details about these parameters are given in

Section 4.3.

4.1 Generation

There are two main categories of wind turbine noise sources: mechanical and

aerodynamic.

4.1.1 Mechanical noise

The mechanical noise is generated by the machinery in the nacelle. The main noise

source is the gearbox, but important contributions are coming from the generator

also [ 5 ]. The cooling fans and the other auxiliary devices contribute to a lower

extent to the total noise.

Depending on the transmission path, the mechanically generated noise can be

divided in two main categories (see Fig. 5 ). When the noise is directly radiated

in the atmosphere it is called air-borne , like the noise emitted by the gearbox

through the nacelle openings. Another part of the mechanical noise is due to

vibrations propagated through the transmission elements and ﬁ ttings to the

nacelle casing and tower leg. This indirect noise is called structure-borne . The

structure-borne gearbox noise constitutes the main contribution to the mechanical

noise [ 5 ].

The noise and vibrations in the gearbox are due to the transmission error of the

meshing gears. The transmission error is the difference between the desired and

the actual position of the driven gear. Thus a more accurate production of the gear-

box leads to lower noise levels. There are also differences among different gear

types, helical teeth are less noisy than the spur ones. Also, epicyclic gearboxes are

usually noisier, one of the gears being directly mounted on the casing. Due to the

wear of the gears the amplitude of the generated vibrations increases with time.

A doubling of the transmission error leads to an increase of approximately 6 dB in

the noise level.

Figure 4: Illustration of sound propagation.

Wind Turbine Acoustics 159

To reduce the gearbox noise, proper teeth proﬁ les and high accuracy products

are needed. To avoid ampliﬁ cation of the noise by the casing, the casing has to be

designed to have eigenfrequencies far from the critical frequencies from the gears.

To avoid the transmission of vibrations, elastic couplings have to be used between

the mechanical devices. For large turbines, such couplings cannot transfer the

torque and the elastic coupling is mounted between the low-speed and the high-

speed stages of the gearbox. This approach is efﬁ cient, since most of the noise is

coming from the high-speed gears, which can be elastically mounted using this

solution (see Fig. 6 ) [ 5 ]. Further reductions of the emitted noise levels can be

achieved by acoustic insulation of the nacelle.

Figure 5: Mechanical noise sources: air-borne noise (continuous arrows) and

structure-borne (dashed arrows).

Figure 6: Mechanical noise reduction using elastic ﬁ ttings (after [ 5 ]).

160 Wind Power Generation and Wind Turbine Design

Figure 7 shows typical 1/3 octave band spectra for a small 75 kW wind turbine

measured by Ohlrich (cited in [ 5 ]). One can observe that the aerodynamic noise

due to the rotor blades is dominating in almost all frequency bands, except around

1 kHz where the mechanical noise originating from the nacelle prevails. Also, the

mechanical noise spectra from the nacelle and tower is more tonal, while the rotor

noise is ‘smoother,’ it has a more broadband character.

As a conclusion, there are efﬁ cient ways to reduce the mechanical noise. Since

mechanical noise is not increasing that fast with increasing turbine size like the

aerodynamic noise [ 6 ], the current research is mainly focused on the reduction of

the aerodynamic component of the wind turbine noise.

4.1.2 Aerodynamic noise

Early theoretical studies of wind turbine noise were based on analogies with acoustic

studies of semi-inﬁ nite half-planes which were symbolizing elements of the rotor

blade [ 7 , 8 ]. Later the studies were based on results from non-rotating airfoil noise.

Lowson [ 9 ] divided the wind turbine noise sources in the following categories:

discrete frequency noise at the blade passing frequency and its harmonics •

self-induced noise sources •

– trailing edge noise

– separation-stall noise

– tip vortex formation noise

– boundary layer vortex shedding noise

– trailing edge bluntness vortex shedding noise

noise due to turbulent inﬂ ow •

The ﬁ rst group of noise sources contains low-frequency components due to the

uneven loading of the blades due to wakes, large-scale structures or velocity

gradients in the atmospheric boundary layer.

Figure 7: Typical frequency spectra from a wind turbine (after [ 5 ]).

Wind Turbine Acoustics 161

The second group includes the noise sources generated by the airfoil itself, thus

present even in a perfectly homogeneous inﬂ ow. The trailing edge noise ( Fig. 8a )

is due to the interaction of the turbulent eddies generated in the boundary layer and

the trailing edge of the airfoil. The turbulent eddies themselves are relatively weak

sources, but their effect is ampliﬁ ed when they are in the vicinity of a plate (in this

case the trailing edge) [ 8 ]. Separation-stall noise ( Fig. 8b ) occurs when the angle

of attack is too large and the ﬂ ow separates. In such cases the separation-stall noise

dominates the other noise sources [ 4 ]. At the tip of the blade a vortex is formed due

to the pressure difference at the pressure and suction sides. The pressure ﬂ uctua-

tions due to the presence of this vortex generate the tip vortex formation noise

( Fig. 8c ). The boundary layer vortex shedding noise ( Fig. 8d ) occurs when bound-

ary layer instabilities develop along the airfoil. Such instabilities might lead to

separation and the appearance of Tollmien Schlichting waves. While the trailing

edge noise has a broadband character, the boundary layer vortex shedding noise is

tonal. Trailing edge bluntness vortex shedding noise appears when the thickness of

the trailing edge exceeds a critical limit, dependent on the Reynolds number and the

shape of the airfoil ( Fig. 8e ). When this critical limit is exceeded a Karman-type

vortex-street develops giving rise to a tonal noise.

The third group includes noise due to atmospheric turbulence. Due to the ﬂ uc-

tuations in the incoming air stream, the blades suffer an unsteady loading, giving

(a) Trailing edge noise (b) Separation stall noise

(d) Boundary-layer vortex

shedding noise

(e) Trailing edge bluntness noise (f) Noise due to turbulent inflow

Figure 8: Noise generation mechanisms for an airfoil.

162 Wind Power Generation and Wind Turbine Design

rise to noise ( Fig. 8f ). One commonly divides the noise due to atmospheric

turbulence into a low-frequency regime (when the length-scale of the ﬂ uctuations

is much larger than the size of the body) and a high frequency regime (the length-

scale of the ﬂ uctuations is much smaller then the airfoil). As it will be shown in

Section 6.2, these two regimes can be treated separately in noise modeling [ 4 ].

Due to the chaotic behavior of turbulence, these models are not deterministic.

According to Lowson [ 9 ], the self-noise sources dominate at low wind speeds,

near cut-in, while at the rated power the turbulence inlet noise source dominates.

A detailed description of all these noise mechanisms can be found, e.g. in [ 4 ].

4.2 Propagation

In the previous section the noise generation mechanisms have been discussed. The

knowledge of these noise sources , however, is not enough to predict the sound

pressure level at a receiver. While the acoustic waves are traveling through the

atmosphere, several factors inﬂ uence the propagated sound pressure levels, the

most important ones being the followings:

The • distance to the receiver. For increasing distance the acoustic energy is

spread in a larger volume which decreases the sound pressure level.

• Absorption is due to the air viscosity and converts the acoustic energy into heat.

• Reﬂ ections due to the ground and surrounding objects.

When a wave passes around a solid object • diffraction occurs. For high frequen-

cies (wavelength much smaller than the object size) a shadow zone occurs

behind the object. The shadow zone decreases with decreasing frequency, com-

pletely disappearing for wavelengths much larger than the size of the object

(see Fig. 9 ).

• Refractions are caused due to temperature gradients which cause different

densities in different layers of the air, and as a consequence impose different

propagation speeds of the sound waves.

The • wind speed and direction inﬂ uences the directivity of the noise propagation.

Figure 9: Diffraction behind obstacles high-frequency (left) and low-frequency

(right) waves.

Wind Turbine Acoustics 163

The inﬂ uences of the last two parameters are visualized in Fig. 10 . If there

would be no wind and no temperature gradients, the sound waves would propa-

gate along straight lines ( Fig. 10a ). In windy conditions the noise propagation

paths are curved towards the wind direction ( Fig. 10b ). Negative temperature

gradients cause lower temperature regions at higher altitude, thus lower propa-

gation speeds for the noise. As a result, the noise propagation paths will be

curved upwards ( Fig. 10c ) (the opposite effect happening for positive pressure

gradients, Fig. 10d ). One can observe in Fig. 10 that in certain conditions shadow

zones appear where the noise will not propagate.

Compared to other industrial noise sources, wind turbines have two main spe-

cial features [ 5 ]. First, the sources are located at an elevated level which leads to

reduced screening and ground attenuation effects. Secondly, in windy conditions

sound propagation is difﬁ cult to predict.

4.3 Immission

There are several factors inﬂ uencing the level of annoyance from wind turbine

noise [ 6 ]. Both the odds of perceiving the wind turbine noise and the odds of being

annoyed by wind turbine noise increases with increasing sound pressure level [ 10 ].

Pedersen and Larsman [ 11 ] showed that the visual impact is important when wind

(a)

(b)

(c)

(d)

Figure 10: The inﬂ uence of vertical temperature gradient and wind speed on noise

propagation: (a) no wind, no temperature gradient; (b) wind from left

to right, no temperature gradient; (c) no wind, negative temperature

gradient; (d) no wind, positive temperature gradient.

164 Wind Power Generation and Wind Turbine Design

turbine noise is evaluated. When wind turbines are considered as ugly struc-

tures being in contrast with the surroundings, the probability of annoyance by

the noise increases, regardless of the measured sound pressure levels. Even if

the visual impact is not considered, the annoyance of the noise with the same

equivalent sound pressure level can be rated differently. Waye and Öhrström

[ 12 ] exposed several persons to noise registered from different wind turbine

types, scaled to 40 dB L

( A ). The differences in annoyance response could not

be explained by the psycho-acoustic parameters considered (sharpness, loud-

ness, roughness, ﬂ uctuation strength and modulation). In regions with high

background noise levels the wind turbine noise is considered less disturbing

[ 6 , 13 ]. Thus the acceptance level of wind turbines is higher in regions with

large trafﬁ c, industrial areas or where a lot of noise is generated by vegeta-

tion or waves, while it is signiﬁ cantly lower in recreational and rural areas. For

the same reason, wind turbine noise is considered more disturbing during the

night than during the day. As it is emphasized by Grauthoff [ 14 ], not only the

audible levels are important, high infrasound (below 16 Hz) levels might also be

percepted as annoying.

4.4 Wind turbine noise regulations

A recent survey of noise regulations in several countries is presented by Pedersen

[ 6 ]. Since it is considered unneeded to present all the details for the speciﬁ c coun-

tries, here we summarize the major strategies adopted in the legislations. There are

three different kinds of noise limitation strategies:

1. Fixed values . As an example, in Sweden the highest recommended sound

pressure level originating from wind turbines is set to 40 dB, with a penalty of

5 dB for pure tones. Although this strategy is straightforward to apply it is the

least ﬂ exible. In quiet areas even noise with sound pressure level correspond-

ing to the 40 dB limit might lead to the annoyance of the people. Contrarily,

in regions with high background noise levels even higher wind turbine noise

levels would be accepted and a ﬁ xed value of the limit leads to suboptimal

power production.

2. Relative values . Wind turbine noise limits in Great Britain recommend a maxi-

mum noise emission of 5 dB above the background noise levels. This approach

is much more ﬂ exible, but difﬁ cult to implement practically.

3. Variable values . In the Netherlands, the maximum limits of the emitted noise

vary as function of the wind speed. This method is more ﬂ exible than the ﬁ rst

one and easier to implement than the second one. Nevertheless, there are fur-

ther issues to be solved. Depending on the atmospheric stability conditions,

the wind speed measured at a ﬁ xed height above the ground might not be an

accurate indicator of the wind speed met by the rotor blades. Van den Berg

[ 15 ] presents an example where the practically emitted sound pressure levels

exceeded the ones predicted with ‘standard’ methods because of the higher

velocities at rotor height than predicted.