Burton T. (et. al.) Wind energy Handbook

Подождите немного. Документ загружается.

Viewpoint analysis is based on sele cting a number of important locations from

which the wind farm is visible and applying professional judgement using quanti-

tative criteria to assess the visual impact. The viewpoints are selected in consulta-

tion with the civic planning authorities and for a large wind farm up to 20 locations

may be chosen. Although approaches vary, the assessment may involve considera-

tion of three aspects: (1) the sensitivity of the landscape, (2) the sensitivity of the

viewpoint, and (3) the magnitude of the change of view. Thus, for example, the

landscape within a National Park will be of ‘high’ sensitivity while a landscape

with existing discordant features such as old quarry workings may have a ‘low’

sensitivity. Similarly a viewpoint where the land use is residentia l or has high

recreational value may have a ‘high’ sensitivity, while a viewpoint that is used only

for indoor employment (e.g., a local industrial estate), might be considered to have

a ‘low’ sensitivity. The magnitude of the impact can be described in a similar

manner depending, for example, on the number of turbines visible, the distance to

the wind farm and the prominence of the development. The overall signiﬁcance of

the impact is then assessed, again using quantitative terminology (such as substan-

tial, moderate, slight, negligible, etc.), by combining these factors. Where a substan-

tial impact is identiﬁed, acceptability will depend on whether it is considered that

the wind farm will have a detrimental effect on the landscape quality.

Figure 9.7 Example of Zone of Visual Impact of a Wind Farm (Courtesy of ReSoft www.re-

soft.co.uk) Based upon the Ordnance Survey Mapping on behalf of The Controller of Her

Majesty’s Stationary Ofﬁce # Crown Copyright MC 100014737

VISUAL AND LANDSCAPE ASSESSMENT 525

Publisher's Note:

Permission to 

reproduce this image 

online was not granted 

by the copyright holder. 

Readers are kindly 

requested to refer to 

the printed version of 

this article.

Figure 9.8 shows a typical wireframe image generated from a viewpoint and

Figure 9.9 shows a photomontage. Both of these images were generated using wind

farm design software. Wireframe representations, provide an accu rate impression

of turbine position and scale while photomontages are the normal tool used to give

an overall impression of the visual effect of a wind farm. Videomontages have been

proposed in order to give an impression of the movement of turbine blades but this

tool is not yet in regular use.

Figure 9.8 Example of Wire Frame View (Courtesy of ReSoft www.resoft.co.uk)

Figure 9.9 Example of Photomontage (Courtesy of ReSoft www.resoft.co.uk)

526

WIND TURBINE INSTALLATIONS AND WIND FARMS

9.2.4 Shadow ﬂicker

Shadow ﬂicker is the term used to describ e the stroboscopic effect of the shadows

cast by rotating blades of wind turbines when the sun is behind them. The shadow

can create a disturbance to people inside buildings exposed to such light passing

through a narrow window. Although considered to be an important issue in

Europe, and recognized in the operation of traditional windmills (Verkuijlen and

Westra, 1984) shadow ﬂicker has not generally been recognized as signiﬁcant in the

USA (Gipe, 1995).

The frequencies that can cause disturbance are between 2.5–20 Hz. The effe ct on

humans is similar to that caused by changes in intensity of an incandescent electric

light due to variations in network voltage from a wind turbine (see Section 10.5.1).

In the case of shadow ﬂicker the main concern is variations in light at frequencies of

2.5–3 Hz which have been shown to cause anomalous EEG (elec troencephalogram)

reactions in some sufferers from epilepsy. Higher frequencies (15–20 Hz) may even

lead to epileptic convulsions. Of the general population, some 10 percent of all

adults and 15–30 percent of children are disturbed to some extent by light

variations at these frequencies (Verkuijlen and Wes tra, 1984).

Large modern three-bladed wind turbines will rotate at under 35 r.p.m. giving

blade-passing frequencies of less than 1.75 Hz, wh ich is below the critical frequency

of 2.5 Hz. A minimum spacing from the nearest turbines to a dwelling of 10 rotor

diameters is recommended to redu ce the duration of any nuisance due to light

ﬂicker (Taylor and Rand, 1991). However , a spacing of this magnitude is likely to be

required in any event by noise constraints and to avoid visual domination.

9.2.5 Sociological aspects

There are a number of computer-based tools available for quantifying visual effects

and landscape architects and planners have developed techniques to place quanti-

tative measures on visual impact using professional judgement. However, publi c

attitudes, which ultimately determine whether a wind farm may be constructed, are

inﬂuenced by many more complex factors. Public attitudes to wind farms have

been studied on a number of occasions (e.g., ETSU, 1993, 1994) and Gipe (1995)

discusses this subject in considerable detail. In general, the large majority of people

approve of wind farms after they have been constructed although a signiﬁcant

minority remains opposed to them. In particular, there is the difﬁcult issue that

some local residents consider they are paying a high cost for a beneﬁt, either

ﬁnancial or environmental, which accrues to others. The ﬁnancial beneﬁts may be

shared with the community in a number of ways including by the development of

co-operative or community-owned wind farms, while the environmental issue has

to be addressed by consultation. Also, it is suggested that stationary wind turbines

are less acceptable than those rotating and so maintaining high availability with a

low cut-in wind speed is likely to improve public perception.

VISUAL AND LANDSCAPE ASSESSMENT 527

9.3 Noise

Noise from wind turbines is often perceived as one of the more signiﬁcant environ-

mental impacts (Wagner, Bareis and Guidati, 1996). During the early development

of wind energy, in the 1980s, some turbines were rather noisy and this led to justiﬁed

complaints from those living close to them. However, since then, there has been very

considerable development both in techniques for reducing noise from wind turbines

and in predicting the noise nuisance a wind farm will create.

The UK document Planning Policy Guidance Note (Department of the Environ-

ment, 1993) suggests that:

‘A planning application for any wind-farm development could usefully be

accompanied by the following information regarding details of the proposed

turbine(s) and pr edicted noise levels:

• predicted noise levels at speciﬁc properties closest to the wind farm over the most

critical range of wind speeds;

• measured background noise levels at the properties and wind speeds outlined

above;

• a scale map showing the proposed wind turbine(s), the prevailing wind condi-

tions, nearby existing developments;

• results of independent measurements of noise emission from the proposed wind

turbine including the sound power and narrow-band frequency spectrum; in the

case of a prototype turbine where no measurements are available, predictions

should be made by comparison with similar machines.’

9.3.1 Terminology and basic concepts

Two distinctly different measures are used to describe wind turbine noise. These

are the sound power level L

of the source (i.e., the wind turbine) and the sound

pressure level L

at a location. Because of the response of the human ear, a

logarithmic scale is used based on reference levels that correspond to the limit of

hearing. The units of both L

and L

are the decibel (dB).

A noise source is described in terms of its sound power level, L

¼ 10 log



(9:4)

where W is the total sound power level emitted from the sour ce (in Watts) and W

is a reference value of 10

12

The sound pressure level L

is deﬁned as

¼ 10 log

(9:5)

528 WIND TURBINE INSTALLATIONS AND WIND FARMS

where P is the RMS value of the sound pressure and P

is a reference value of

2 3 10

5

Pa.

By simple algebraic manipulation it may be seen that the addition of n sound

pressure levels (expressed in dB) is carried out as shown:

¼ 10 log

j¼n

j¼1

P( j)

=10

(9:6)

Thus, adding two sound pressure levels of the same magnitude results in an

increase of 3 dB. Table 9.2 gives an indication of the typical ran ge of sound pre ssure

levels.

The human ear is capable of detecting sounds between 20 Hz and 20 kHz and

spectral analysis is typically unde rtaken over this range. A narrow-band spectrum,

with a deﬁned bandwidth of measurement, gives the fullest information of the

signal and may be used to detect particular tones. However, it is conventional to

use octave and 1/3-octave bands for broadband analysis. The upper frequency of

an octave band is twice that of the lower frequency while for the 1/3-octave band

the upper frequency is

ﬃﬃﬃ

times the lower frequency.

It is common to weight the measurements to reﬂect the response of the human

ear with frequency. This is done by applying the so-called A-weighted ﬁlter.

Measurements made with this ﬁlter are referred to as dBA or dB(A). Table 9.3

shows the centre frequencies of the octave bands together with the A-weighting in

dB. It may be seen that frequencies below 250 Hz and above 16 kHz are heavily

attenuated.

An equivalent sound pressure level L

eq,T

is the value of a continuous steady

sound that, within the speciﬁed time interval (T) has the same me an square sound

pressure level as the sound under consid eration which varies with time (Interna-

tional Energy Agency, 1994)

eq,T

¼ 10 log

(9:7)

A similar calculation may be undertaken using A–weighted values to give L

Aeq,T

Table 9.2 Examples of Sound Pressure Levels

Example Sound pressure level

(dB(A))

Threshold of hearing 0

Rural night time background 20–40

Busy general ofﬁce 60

Inside factory 80–100

Jet aircraft at 100 m 120

Wind farm at 350 m 35–45

NOISE 529

Aeq,T

¼ 10 log

(9:8)

where L

Aeq,T

is the equivalent continuous A-weighted sound pressure level deter-

mined over time T and P

is the instantaneous A-weighted sound pressure.

The exceedance level, L

A90

, is deﬁned as the A-weighted sound pressure level

which is exceeded for 90 percent of the time. Some planning authoritie s prefer the

use of the L

A90

sound pressure level, particularly for measurements of backgrou nd

noise, as the L

measurement may be heavily inﬂuenced by short-term effects such

as passing aircraft or trafﬁc. A wind farm is a fairly constant source of noise and so

its contribution to the L

A90

sound pressure level (measured over a 10 min period) is

likely to be some 1.5–2.5 dB(A) less than to the 10 min L

Aeq

value (ETSU, 1997a).

The sound intensity I is the power transmitted through a unit area, A,

I ¼

(9:9)

and far from the source of sound with a uniform ﬂux

I ¼

(9:10)

where P is the RMS value of the sound pressure level, and Z

is the characteristic

acoustic impedance.

Choosing a suitable value for I

ref

(10

12

W=m

), the sound pressure level may be

expressed in terms of sound intensity as

¼ 10 log

ref



(9:11)

For spherical spreading

Table 9.3 Centre Frequency of Oc-

tave Bands and A-weighting

Octave band centre

frequency (Hz)

A-weighting (dB)

16 56.7

31.5 39.4

63 26.2

125 16.1

250 8.6

500 3.2

1000 0.0

2000 1.2

4000 1.0

8000 1.1

16000 6.6

530

WIND TURBINE INSTALLATIONS AND WIND FARMS

I ¼

4r

(9:12)

where r is the distance from the source.

Hence, under conditions of ideal spherical spreading

¼ 10 log

4r

12



¼ L

 10 log

(4r

)(9:13)

and similarly, if hemispherical spreading is assumed

¼ 10 log

2r

12



¼ L

 10 log

(2r

)(9:14)

Sound pressure level, from a point source, decays with distance according to the

inverse square law under both spherical and hemi-spherical spreading assumptions.

Hence for each doubling of distance the sound pressure level is reduced by 6 dB.

For a line source of noise the sound pressure level is given by

¼ 10 log

2r10

12



¼ L

 10 log

(2r)(9:15)

is the sound power level per unit length of the source.) Hence for cylindrical

spreading the decay is only proportional to distance, resulting in a 3 dB reduction of

sound pressure level for each doubling of distance per pendicular to the line source.

9.3.2 Wind-turbine noise

Noise from wind turbines is partly mechanical, and partly aerodynamic.

Mechanical noise is generated mainly from the rotating machinery in the nacelle

particularly the gearbox and generator although there may also be contributions

from cooling fans, auxiliary equipment (such as pumps and compressors) and the

yaw system. Mechanical noise is often at an identiﬁable frequency or tone (e.g.,

caused by the meshing frequency of a stage of the gearbox). Noise containing

discrete tones is more likely to lead to complaints and so attracts a 5dB penalty in

many noise standards. Mechanical noise may be air-borne (e.g., the cooling fan of

an air-cooled generator) or transmitted through the structure (e.g., gear box meshing

which is transmitted through the gearbox casing, the nacelle bed-plate, the blades

and the tower).

Pinder (1992) is quoted by Wagner, Bareis and Guidati (1996) as giving the values

of sound power level as shown in Table 9.4 for a 2 MW experimental wind turbine.

It may be seen that the gearbox is the dominant noise source through structure-

borne transmis sion.

Techniques to reduce the mechanical noise generated from wind turbines include

careful design and machining of the gearbox, use of anti-vibration mountings and

couplings to limit structure-borne noise, acoustic damping of the nacelle, and liquid

cooling of the generator (see Section 7.4.8).

Aerodynamic noise is due to a number of causes (ETSU, 1997a):

NOISE 531

• low-frequency noise,

• inﬂow turbulence noise,

• airfoil self noise.

Low-frequency noise is caused by changes in the wind speed experienced by the

blades due to the presence of the towe r or wind shear. Although this effect is very

pronounced with downwind turbines it is also signiﬁc ant with upwind machines.

The spectrum of the noise is dominated by blade-passing frequency (typically up to

3 Hz) and its harmonics (typically up to 150 Hz). It may be seen from Table 9.3 that

the A-weighted ﬁlter heavily attenuat es these frequencies and so this source does

not make a major contribution to audible noise. However, it was reported that

experimental, downwind turbines, constructed in the 1980s excited low-frequency

vibrations in adjacent buildings. For upwind turbines increasing the clearance

between the blades and tower will reduce this effect. Hubbard and Shephard

(Spera, 1994) show interesting experimental evidence of low-frequency noise from a

large 80 m diameter downwind turbine and provide a comparison with a 90 m

diameter upwind machine.

Inﬂow turbulence creates broadband noise as the blades inte ract with the eddies

caused by atmospheric turbulence. It generates frequencies up to 1000 Hz which

are perceived by an observer as a ‘swishing’ noise. Turbulent inﬂow noise is

considered to be inﬂuenced by the blade velocity, airfoil section and turbulence

intensity. Wagner, Bareis and Guidati (1996) describe this phenomenon in some

detail, remarking that it is not yet fully understood, and quote the results of one

ﬁeld experiment where the sound pressure level actually decreased with increasing

turbulence intensity.

Airfoil self-noise is generated by the airfoil itself, even in steady, turbulent-

free ﬂow. It is typically broadband although imperfections in the blade surface may

generate tonal components. The main types of airfoil self-noise include the

following.

Trailing edge nois e. This is perceived as a broadband swishing sound with frequen-

cies in the range 750–2000 Hz. It is due to interaction of the turbul ent boundary

Table 9.4 Sound Power Levels of Mechanical Noise of a 2 MW Experi-

mental Wind Turbine (after Wagner, Bareis and Guidati, 1996)

Element Sound power level

(dB(A))

Air-borne or structure-borne

Gearbox

97.2

84.2

Structure-borne

Air-borne

Generator 87.2 Air-borne

Hub (from gearbox) 89.2 Structure-borne

Blades (from gearbox) 91.2 Structure-borne

Tower (from gearbox) 71.2 Structure-borne

Auxiliaries 76.2 Air-borne

532

WIND TURBINE INSTALLATIONS AND WIND FARMS

layer with the trailing edge of the blade. Trailing edge noise is a major source of

higher frequency noise on wind turbines.

Tip noise. The literature is not clear as to whether three-dimensional tip effects are a

major contributor to wind turbine noise. However, the majority of the blade noise,

as well as the turbine power, emanates from the outer 25 percent of the blade and

so there has been very considerable investigation of nove l blade tips to reduce

noise. Imperfections in the blade shape due to tip-brakes or other control surfaces

are another potential source of noise.

Stall effects. Blade stall causes unsteady ﬂow around the airfoil which can give rise

to broadband sound radiation.

Blunt trailing edge noise. A blunt trailing edge can give rise to vortex shedding and

tonal noise. It can be avoided by sharpening the trailing edge but this has implica-

tions for manufacturing and erection.

Surface imperfections. Surface imperfections such as those caused by damage during

erections or due to lightning strikes can be a signiﬁcant source of tonal noise.

The obvious approach to reducing aerodynamic noise is to lower the rotational

speed of the rotor, although this may result in some loss of energy. The ability to

reduce noise in low wind-speed conditions is a major beneﬁt of variable-speed or

two-speed wind turbines. An alternative technique would be to reduce the angle of

attack of the blade although again with a potential loss of energy.

Stiesdal and Kristensen (1993) provide a comprehensive description of noise

control methods applied to a 300 kW stal l-regulated wind turbine. The gearbox was

identiﬁed as an important source of tonal, mechanical noise. Control measures

included detailed modiﬁcations to the design and manufacture of the gears and an

additional insulating covering over the gearbox housing. Air-borne mechanical

noise was minimized by total enclosure of the nacelle and careful design of sound

bafﬂes in the ventilation openings. Structure-borne noise was reduced by rubber

mounting both the gearbox and generator and using a rubber coupling on the high-

speed shaft. The long low-speed shaft (2 m) reduced the drive train vibrations

transmitted to the rotor. Tip noise was minimized by the use of a ‘tip torpedo’

to control the tip-vortex while trailing-edge noise was controlled by specifying a

1–2 mm traili ng edge thickness. Stall noise was reduced by using a turbulator strip

on the leading edge of the blades near the tip that init iated stall in a controlled

manner. Overall these measures resulted in a reduction of the sound power level of

3–4 dB and the elimination of signiﬁcant tones. The most important noise control

methods were considered to be the tip modiﬁcation, the controlled stall and the

gearbox improvements. The measured sound power level of the improve d wind

turbine was 96 dB(A) at a wind speed of 8 m/s (measured at 10 m height). This was

an impressive improvement as typical sound power levels of a modern 600 kW

wind turbine may be up to 100 dB(A) at a wind speed of 8 m/s (measured at 10 m

height) increasi ng by 0.5–1 dB(A) per m/s.

NOISE 533

9.3.3 Measurement, prediction and assessment of wind-farm noise

The sound power level of a wind turbine is normally determined by ﬁeld experi-

ments. These were originally speciﬁed in the IEA Recommended Practice (Interna-

tional Energy Agency, 1994) but are now described in an international standard (BS,

1999, IEC, 1998). Outdoor experiments are necessary because of the large size of

modern wind turbines and the necessity to determine their noise performance

during operation. The sound power level cannot be measured directly but is found

from a series of measurements of sound pressure levels made around the turbine at

various wind speeds from which the backgrou nd sound pressure levels have been

deducted. The method provides the apparent A-weighted sound power level at a

wind speed of 8 m=s, its relationship with wind speed and the directivity of the

noise source for a single wind turbine. It does not distinguish between aerodynamic

and mechanical noise although there is a method proposed to determine if tones are

‘prominent’. In addition to A-weighted sound pressure levels, octave or 1/3-octave

and narrow-band spectra are measured.

The measurements are taken at a distance, R

, from the base of the tower:

¼ H þ

(9:16)

where H is the hub height and D is the diameter of the rotor. This distance is a

compromise to allow an adequate distance from the source but with minimum

inﬂuence of the terrain, atmospheric conditions or wind-induced noise. The micro-

phones are located on boards at ground level so that the effect of ground

interference on tones may be evaluated. Four microphone positions are used as

shown in Figure 9.10.

Simultaneous A- weighted sound pressure level measurements (more than 30

measurements each of 1–2 min duration) are taken with wind speed. All wind

speeds are corrected to a reference height of 10 m with a terrain roughness length of

¼ 0:5 m. The preferred method of determining wind speed, when the turbine is

operating, is from the electrical power output of the turbine and the power curve.

The main sound pressure level measurement is that of the downwind positi on

while the other thr ee microphones are used for determining directivity. Measure-

ments are taken with and without the turbine operating over a range of at least

4m=s which includes the 8 m=s reference. The sound pressure levels, with and

without the turbine in operation, are then plotted against wind speed and linear

regression used to ﬁnd the 8 m=s values. The method of bins may be used to group

the data. The sound pressure level of the turbine alone at the reference conditions,

eq,ref

, is then calculated using

Aeq,ref

¼ 10 log

(10

SþN

=10

 10

=10

)(9:17)

where L

SþN

is the sound pressure level of the wind turbine and the background

sound at 8 m=s, and L

is the sound pressure level of the background with the

wind turbine parked at 8 m=s.

534 WIND TURBINE INSTALLATIONS AND WIND FARMS