Académique Documents
Professionnel Documents
Culture Documents
DYNAMICS
AND
BLADE
by
YNG-RU CHEN
May, 2016
CASE WESTERN RESERVE UNIVERSITY
SCHOOL OF GRADUATE STUDIES
Yng-Ru Chen
Committee Chair
Dr. J. Iwan D. Alexander
Committee Member
Dr. Jaikrishnan R. Kadambi
Committee Member
Dr. Paul Barnhart
Committee Member
Dr. David H. Matthiesen
Committee Member
Dr. Michael Hölling
Date of Defense
感謝父母,兄姊,兒子和愛犬的包容、支持和陪伴
TABLE OF CONTENTS......................................................................................................................... i
LIST OF TABLES ................................................................................................................................ iii
LIST OF FIGURES .............................................................................................................................. iv
LIST OF SYMBOLS .......................................................................................................................... viii
ABSTRACT........................................................................................................................................ ix
INTRODUCTION ................................................................................................................................ 1
Part I: ................................................................................................................................................ 6
I. OVERVIEW AND BACKGROUND ........................................................................................... 7
1. THE CUP ANEMOMETER .................................................................................................. 9
2. WIND RESOURCE MEASUREMENT................................................................................. 10
II. OBJECTIVE .......................................................................................................................... 14
III. THEORY .......................................................................................................................... 15
IV. CALIBRATION.................................................................................................................. 20
V. OVERSPEEDING ANALYSIS OF DIFFERENT CUP ANEMOMETERS ....................................... 26
VI. OVERSPEEDING ANALYSIS OF UNBALANCED CUP ANEMOMETERS .............................. 37
VII. PRELIMINARY FIELD DATA EXAMINATION ..................................................................... 42
VIII. CONCLUSIONS AND FUTURE RESEARCH ........................................................................ 44
Part II .............................................................................................................................................. 46
I. OVERVIEW AND OBJECTIVES ............................................................................................. 47
II. NOISE AND SOUND FUNDAMENTALS ................................................................................ 50
1. SOUND AND NOISE ........................................................................................................ 50
2. NOISE STANDARDS AND REGULATIONS ........................................................................ 52
III. WIND TURBINE AERODYNAMICS ................................................................................... 54
1. LIFT, DRAG AND MOMENT COEFFICIENTS ..................................................................... 54
2. TIP SPEED RATIO ............................................................................................................ 54
3. AIRFOIL BEHAVIOR ......................................................................................................... 55
4. TWISTED BLADE ............................................................................................................. 55
IV. NOISE MECHANISMS OF WIND TURBINES ..................................................................... 57
1. WIND TURBINE AERODYNAMIC SOUND GENERATION ................................................. 57
i
2. CLASSIFICATION OF NOISE MECHANISMS ..................................................................... 58
3. ACOUSTIC SOURCES ....................................................................................................... 61
V. NUMERICAL ANALYSIS ....................................................................................................... 65
1. PHYSICAL MODEL ........................................................................................................... 65
2. MESH GENERATION ....................................................................................................... 65
3. TURBULENCE MODEL ..................................................................................................... 69
4. FAR FIELD NOISE PREDCITION USINF THE FFOWCS WILLIAMS AND HAWKINGS
ACOUSTIC MODEL .................................................................................................................. 70
5. BROADBAND NOISE SOURCE MODEL ............................................................................ 72
VI. RESULTS ......................................................................................................................... 74
1. FLOW FIELD AND STREAMLINES .................................................................................... 74
2. SURFACE ACOUSTIC POWER .......................................................................................... 75
3. FAR-FIELD AERODYNAMIC NOISE AT DIFFERENT WIND SPEEDS ................................... 82
4. SURFACE ACOUSTIC POWER AND FAR-FIELD AERODYNAMIC NOISE PREDICTION FOR A
YAWED TURBINE .................................................................................................................... 83
VII. DISCUSSION.................................................................................................................... 87
VIII. CONCLUSIONS ................................................................................................................ 89
1. SUMMARY ...................................................................................................................... 89
2. FUTURE WORK ............................................................................................................... 90
APPENDIX A .................................................................................................................................... 91
APPENDIX B .................................................................................................................................... 93
APPENDIX C: ................................................................................................................................... 98
APPENDIX D:................................................................................................................................. 101
APPENDIX E: ................................................................................................................................. 102
BIBLIOGRAPHY ............................................................................................................................. 104
ii
LIST OF TABLES
Table II-1: Health and safety executives noise regulation guidelines on exceeding safe level ..... 52
Table II-2: Recommended noise level standards and guidelines by WHO and selected countries
Table IV-1: Wind Turbine Aerodynamic Noise Mechanisms (From Wagner, Bareib, & Guidati,
1996) .............................................................................................................................................. 61
Table VI-1: Sound Pressure Level (dB) of different rpm at receiver locations............................... 82
Table VI-2: Surface Acoustic Power (dB) of different yawed angles at 72 rpm ............................. 84
Table VI-3: The far-field aerodynamic noise of a yawed turbine at 72 rpm at receiver locations 86
iii
LIST OF FIGURES
Figure I-1: Global Energy Consumption 2014; (Data taken from in the BP Statistical Review of
Figure I-2: Annual and Cumulative Growth in U.S. wind power Capacity [11] ................................ 3
Figure I-1: Schematic of a turbulent flow with a mean horizontal unidirectional component and
showing a dye trace marking typical particle trajectories in such a flow [24] .............................. 11
Figure I-2: Two examples of velocity (U) distributions varying in time (t) [24] ............................. 12
Figure I-3: Ultrasonic wind speed measurements from the FINO platform in North Sea (Dr. M.
Figure III-2: A schematic diagram of one cup perpendicular to the wind direction ...................... 16
Figure IV-1: Cup anemometers used for the current study (a): ThiesTM First Class Cup
Anemometer, (b): Second Wind C3TM Cup Anemometer, (c): ThiesTM Sensor Compact Cup
Figure IV-2: The calibration function of ThiesTM First Class Cup anemometer .............................. 23
Figure IV-3: The calibration function of ThiesTM First Class Cup anemometer according to the
Figure V-2: Wind tunnel experiment setup (looking at outlet of the wind tunnel)....................... 27
Figure V-4: Wind speed measured by ThiesTM First Class Cup anemometer while decreasing wind
iv
Figure V-5: Two cases of overspeeding situation of ThiesTM First Class Cup anemometer at 10
m/sec ............................................................................................................................................. 31
Figure V-6: Comparison of measured and computed response to stepwise changes in wind speed
....................................................................................................................................................... 32
Figure V-7: Overspeeding curves for the ThiesTM First Class Cup Anemometer at various wind
speeds ............................................................................................................................................ 33
Figure V-8: Overspeeding curves for the Second WindTM C3 Cup Anemometer at various wind
speeds ............................................................................................................................................ 33
Figure V-9: Overspeeding curves for the ThiesTM Compact Cup Anemometer at various wind
speeds ............................................................................................................................................ 34
Figure V-10: Overspeeding curves for the ThiesTM Classic Cup Anemometer at various wind
speeds ............................................................................................................................................ 34
Figure V-11: Overspeed curves for the four anemometers at V = 10 m/sec (* denotes the mean
uncertainty).................................................................................................................................... 36
Figure VI-2: Calibration function of the Thies™ Classic Cup Anemometers with different weighted
cups ................................................................................................................................................ 39
Figure VI-3: Overspeed curves comparison of the Thies™ First Class Cup Anemometers with
different weighted cups at wind speed of 10 m/sec (* denotes “mean total uncertainty”) ........ 40
Figure VII-1: Measured wind speed vs. time (upper graph) and the inverse of the wind speed vs.
Figure III-1: Forces and moment on an airfoil section, α, angle of attack; c, chord [22] ............... 54
v
Figure IV-1: Horizontal axis wind turbine noise sources (Romero-Sanz & Matesanz, 2008)......... 59
Figure IV-2: Aerodynamic noise sources around a turbine blade due to wind flow, U (Brooks,
Figure V-3: Structured grids consisting of triangular, polygonal meshes and orthogonal meshes66
Figure VI-3: Surface Acoustic Power Level (dB) on blades at 36 RPM ........................................... 77
Figure VI-4: Surface Acoustic Power Level (dB) on blades at 54 RPM ........................................... 78
Figure VI-5: Surface Acoustic Power Level (dB) on blades at 60 RPM ........................................... 79
Figure VI-6: Surface Acoustic Power Level (dB) on blades at 72 RPM ........................................... 80
Figure VI-7: Maximum Sound power levels for wind speeds 7 m/s through 25 m/s at different
Figure VI-8: A wind turbine yawed to the wind direction [22] ...................................................... 83
vi
Figure VI-9: Surface Acoustic Power Level (dB) of 72 rpm at wind speed of 7 m/sec on blades at
Figure VII-1: Sound power levels as a function of standardized wind speed for the ARE 442 wind
turbine (rotor diameter is 7.2 m) (Huskey & Dam, 2010) [63] ...................................................... 87
Figure VII-2: Measured sound pressure levels of a Southwest Whisper 900 wind turbine (rotor
Figure VII-3: Sound power levels as a function of standardized wind speed for the Gai Wind 11-
vii
LIST OF SYMBOLS
𝑘- ω Turbulence model
viii
Characterization of Cup Anemometer Dynamics
ABSTRACT
by
YNG-RU CHEN
The results of two separate investigations into two issues related to the generation of
electrical power from wind are reported in this dissertation. The first, an experimental
measure local wind conditions at wind turbine array locations. This need is two-fold, one
being the need to assess wind-turbine performance (in terms of the conversion of wind
power to electrical power, the second is to ensure that predictions of the wind power
capacity of a given area of interest are based on reliable wind speed measurements. The
tunnel. It is found that the response of the anemometer to changing wind conditions is
determined by two parameters both of which represent the ratio of drag forces to
ix
inertial forces. In the recent of a sudden decrease in wind speed to zero, one of the
parameters alone controls the rate of decay of the anemometers speed to zero. This
The second part of this dissertation deals with the subject of aeroacoustic noise. An
important factor to take into consideration in the development of wind farms near
populated areas is the potential for noise disturbance. In this dissertation, the topic of
the wind velocity and pressure fields through the turbine. The Ffowcs-Williams and
Hawkings acoustic analogy is used to calculate the acoustic pressure at a far field
receiver location. The results obtained are compared with experimental data. The
purpose of the investigation is, by extending earlier work of Ranft et al., to predict the
aeroacoustic noise at different wind speeds and at different rotational speeds of the
turbine. Both rotational noise and broadband noise were analyzed. It was found that the
leading edge and the blade tip are the primary source of the broadband noise. In all
simulations, the loudest noise (80.8 dB) generated by the wind turbine blade tip and the
leading edge are higher than the World Health Organization recommended noise level
x
INTRODUCTION
One of today’s greatest global energy challenges is the need for more affordable,
reliable, clean, secure and renewable sources of electricity. There is still heavy reliance
on coal, oil, natural gas and nuclear fuels to generate electricity. The 2015 BP Statistical
Review of World Energy shows fossil fuels (Oil, gas, coal and nuclear) accounted for 86%
Figure I-1: Global Energy Consumption 2014; (Data taken from in the BP Statistical
Note: the unit ‘millions of tonnes of oil’ was converted to TeraWatt-hours using the
1
Fossil fuels are finite and non-renewable on a human socio-economic time scale [2].
They are limited in spatial extent and are expected in the long term, to become
increasingly more expensive to extract [3]. Furthermore, they are not renewable [3].
Thus, the use of fossil fuels for energy generation is not sustainable. This, along with
increasing concern about anthropogenic effects on climate change with associated CO₂
and other greenhouse gas emissions is driving governments and power companies
across the globe to diversify fuel sources for electrical power generation and, in
provides clean, renewable, and cost-effective electricity around the world. According to
the International Renewable Energy Associate, global installed wind energy capacity
reached nearly 370 GW in 2014 [7] [8]. Some countries (Germany, Demark, Spain…)
have supported wind industry heavily since the end of the twentieth century. Recently,
China and India have started to invest more and more in wind energy development and
showed the highest growing rates of wind energy installed power in the past years.
According to 2015 data reported by the Global Wind Energy Council [9] and the US
Department of Energy’s Energy Information Administration (EIA), China now has the
largest installed wind power capacity (115 GW) followed by the USA (66 GW) [10] [11].
The U.S. wind power market represented over 9% of global installed capacity in 2014.
The annual growth of global installed wind capacity has been increasing steadily for
nearly 20 years. In contrast, US installed capacity has tended to be more sporadic over
the same time period (See Figure I-2). This reflects, perhaps, the lack of a clear national
2
policy to develop wind power and a reliance on incentives on a state-by-state basis.
Nevertheless, US wind energy capacity has more than doubled over the past five years.
Recent data [6], [8] indicates that US installed capacity had reached 66 GW in 2014.
Figure I-2: Annual and Cumulative Growth in U.S. wind power Capacity [11]
The continued increase in the use of wind as a source for electrical power generation
has resulted in a variety of research activities in the public and private sectors that
address a range of issues from performance, safety, efficiency and cost to public
acceptance.
The results of two separate investigations into two issues related to the generation of
electrical power from wind are reported in this dissertation. The first, an experimental
measure local wind conditions at wind turbine array locations. This need is two-fold, one
being the need to assess wind-turbine performance (in terms of the conversion of wind
3
power to electrical power, the second is to ensure that predications of the wind power
capacity of a given area of interest are based on reliable wind speed measurements.
The latter is easily understood by considering that the power produced by the wind
turbine is proportional to the cube of the wind speed. Thus, a 1% error in the measured
15% error in the power and so on. As will be shown in the following chapters, there are
standard instrument used industry wide to measure wind speed for wind energy
tunnel.
The second part of this dissertation deals with the subject of aeroacoustic noise. An
important factor to take into consideration in the development of wind farms near
populated areas is the potential for noise disturbance. In a number of countries, wind
power capacity has been significantly limited by the so-called “Not In My Back Yard”
attitude to specific wind power projects, even when polls show that in fact the public
supports the idea wind power as an energy source [12] [13]. Noise disturbance is among
the issues raised by wind project opponents. There are two prevalent noise sources that
arise in connection with wind turbines, mechanical [12] and aerodynamic [14]. In this
dynamic model to compute the wind velocity and pressure fields through the turbine.
4
The Ffowcs-Williams and Hawkings (FW–H) acoustic analogy is used to calculate the
acoustic pressure at a far field receiver location. The results obtained are compared with
experimental data. The purpose of the investigation is, by extending the work of Ranft
et al. [15], to predict the aeroacoustic noise at different wind speeds and at different
5
Part I:
6
I. OVERVIEW AND BACKGROUND
There are many options that can be exercised to quantify wind resources at a given
wind farm site but the traditional meteorological mast equipped with anemometers,
thermometers and other sensors are still the most accurate way to obtain such data.
While limited to measurements at discrete locations (and heights) this approach is used
of the amount of electrical power that could be produced, say annually, at a specific
followed according to the IEC 61400-12 standard. The standard recommends measuring
wind speed using a calibrated cup anemometer installed within 2.5% of hub height of
the wind turbine. The anemometer may be installed on a boom attached to the tower
pointed into the prevailing wind direction. Flow distortion, say from buildings or other
7
The importance of the accuracy of the anemometer can be illustrated by considering the
following. If the actual wind speed is 10 m/s but the measured speed is 10.5 m/s the
error is 0.5 m/s or 5%. However, the power produced by the wind turbine varies as the
cube of the wind speed. It follows that the error in the estimate of the power produced
would, in this case, be over 15%. Note that, in addition to providing a quantitative
assessment of a potential wind field, anemometers are used to develop the so-called
power curve of a wind turbine and reliable data is essential to obtain the turbine’s
power curve and to monitor the performance of installed wind turbines. The most
commonly used anemometers for wind power applications are three-cup anemometers.
known as “overspeeding” [16] [17] [18] [19]. Because of overspeeding, the mean wind
8
1. THE CUP ANEMOMETER
Dr. T.R. Robinson invented the first cup anemometer in 1846. It had four hemispherical
cups. Each cup was fixed on one end of four arms that were mounted at equal angles to
each other on the vertical shaft [20]. While the air flows past the cups, the cups rotation
corresponds to the wind speed. In 1926, the Canadian J. Patterson noticed that each cup
produced the maximum torque when the cup was at 45 degree to the incoming wind.
He developed the three-cup anemometer that responded to the gusty winds more
quickly than the four-cup anemometer [16]. The simple design of a cup anemometer
consists of three cups mounted on a vertical shaft. The differences of the wind pressure,
aerodynamic drag force and acting torque between the concave side and the convex
side of the cup causes it to turn in the direction from the convex side to the concave
side of next cup. The revolution speed is proportional to the wind speed irrespective of
wind direction. Wind speed signals are generated from either a generator or a pulse
The cups were conventionally made of brass for its qualities of rigidity and rust
resistance. In recent years, however, cups made of light alloy or carbon fiber thermo-
Cup anemometers are most commonly used for measuring wind speeds in the
atmospheric boundary layer. Typically, cup anemometers can be divided into three
9
In a permanent regime, a cup anemometer turns at a frequency proportional to the
wind speed irrespective to the wind direction. One downside of cup anemometers is
their asymmetric dynamic response to accelerating and decelerating wind speeds under
turbulent wind conditions. Due to inertia the anemometer will respond faster in
magnitude. Therefore, in a turbulent flow, the cup anemometer will overestimate the
mean wind speed, especially if there are many sudden large decelerations. This
Wind is variable in space and time. It varies non-uniformly with altitude, local
topography and because air density is a function of temperature and humidity [22] the
Commercial wind turbines stand tall with typical hub heights of between 50-120 meters.
Situated in the lower reaches of the atmospheric boundary layer (ABL) wind turbines
experience wind that are influenced directly by the earth’s surface. One would expect
the horizontal wind speed to be zero at the earth’s surface and to increase with height
in the atmospheric boundary layer. The variation of wind speed with elevation, so called
vertical wind profile or vertical wind shear, is very important in wind energy engineering
since it determines the productivity of a wind turbine and also impacts the lifespan of a
10
Turbulence in the wind is caused by the kinetic energy dissipation of wind into thermal
energy via the creation and destruction of eddies (or gusts). Turbulent eddies create
fluctuations in velocity and the velocity record includes both a mean and a turbulent
component.
component and showing a dye trace marking typical particle trajectories in such a flow
[24]
As discussed in the previous section, while cup anemometers are the most commonly
used instrument for wind speed measurement their response to turbulent fluctuations
in wind speed often results the overestimation of the actual wind speed through over
speeding. Consider the turbulent flow shown schematically in Figure I-1 with a mean
̅ = (𝒖, 0), where 𝒖 = (𝑢, 𝑣), and 𝑢, 𝑣 are the horizontal and vertical
wind speed 𝒖
velocity components, respectively. Here the important quantities that are used to
𝑡+𝑇 1
𝑢̅ = ∫𝑡 𝑢(𝑡) 𝑑𝑡 = ∑𝑁
1 𝑢𝑖 (1)
𝑁
̅̅̅̅̅̅̅̅2 + ̅̅̅̅̅̅̅̅ 1 2 2
𝑢𝑟𝑚𝑠 = √𝑢′(𝑡) 𝑣′(𝑡)2 = √𝑁 ∑𝑁
1 (𝑢′𝑖 + 𝑣′𝑖 ) (4)
A larger value of turbulence intensity indicates a higher level of turbulence. The figure
below shows an example where two time series have the same mean velocity but the
Figure I-2: Two examples of velocity (U) distributions varying in time (t) [24]
In general, the highest turbulence intensities happen at the lowest wind speed but the
lower limiting values at a given location will depend on the specific terrain features and
12
surface conditions at the site. Turbulent wind may have a relatively constant mean over
time periods of an hour or more, but over shorter times in may be quite variable.
When atmospheric wind data is being analyzed, the mean velocity, turbulence strength
and turbulence intensity are usually calculated for 10-minute bins. The turbulence
intensity computed over this interval of time captures a measure of the turbulent
fluctuations but does not capture the particular nature of the turbulence. The details
may be important in that for two flows with the same turbulence intensity the specific
nature of the fluctuations may lead to more or less overspeeding. For example, two
wind measurements shown in Figure I-3 were recorded with an ultrasonic anemometer.
Both have the same turbulence intensity (about 4%). The blue curve has several sharp
decreases in wind speed (three are indicated by red arrows). It will be shown later that
such conditions can result in higher estimations of the overall wind speed.
Figure I-3: Ultrasonic wind speed measurements from the FINO platform in North Sea
13
II. OBJECTIVE
anemometers for fluctuating wind speeds and for decreasing speeds in particular. The
investigated by adding rare earth magnets as weight changes on the cups. These
characteristics are basis for the post processing of measured data under turbulent wind
conditions. Regions of the measured time series that are dominated by the specific
dynamics of the cup anemometer are typically affected by over-speeding and this
14
III. THEORY
The response and accuracy of a cup anemometer are determined by its weight, physical
dimension (such as the cups’ shape, size and the rotation radius) and internal friction.
anemometers using different techniques and mathematical models since the early work
of Robinson in 1846.
is placed into a uniform flow, consider Figure III-1. The anemometer is inserted into the
flow in a direction perpendicular to the air flow. As it is moved into the flow, the
surfaces of the cups that are concave toward wind have higher wind resistance than
convex surface. The force of the wind causes the cups to rotate and the rate of change
of the angular speed of the rotor is equal to sum of the moments acting on each cup.
15
Figure III-1 illustrates the idea that the flow of wind around the cups of an anemometer
produces a higher force on the outer side of the cup than it does on the inside of the
Over the years, scientists and engineers have investigated various shapes of cups with
different lengths of arms seeking to develop a cup anemometer that suffers neither
from overspeeding nor deviations to inclined flows. Each proposed design has its
advantages and disadvantages. For example, the larger the cups with longer arms, the
slower the changes in measured wind speed. It also means the mean measured wind
Consider only one single cup that is oriented perpendicular to a free wind speed, V, so
that the convex part of the cup surface is toward the oncoming wind (Figure III-2).
Figure III-2: A schematic diagram of one cup perpendicular to the wind direction
16
1
𝐹 = 2 𝐶𝐷 𝜌𝑉 2 𝜋𝑟 2 (6)
Here CD is the drag coefficient of the cup, ρ is the air density, V is the uniform wind
speed, and r is the radius of the cup. For these conditions, the torque on one cup is
given by
1
𝑀= 𝐶 𝜌𝑉 2
𝜋𝑟 2 𝑅 = 𝐹𝑅 (7)
2 𝐷
Here R is the distance from the center of the cup to center of the rotor (i.e. the axis of
Now consider the whole cup anemometer (Figure III-3). The rotational speed, 𝜔, of the
uniform. The angle Φi is the instantaneous angle made by the rotor arm of the ith cup
17
The drag coefficient for the cups depends on the shape of the cup, the cup surface
roughness and the cup orientation relative to the wind direction. As a first
approximation, the drag coefficient of the open (concave) cup facing the incoming wind
is set to be 𝐶𝐷1 while the drag coefficient of the closed (convex) cup facing the incoming
wind is 𝐶𝐷2 .
The frictional torque on the anemometer depends on air temperature and the rotation
speed. It is normally small in comparison to the aerodynamic torque [28] [29], and can
be neglected except at very low wind speeds (close to the ‘cut-in speed of the
anemometer). In this case the aerodynamic torque T is equal to the inertial force of
3
dw
I e = T=
dt z
å r (F i )´ F (F i ),
i= 1
(8)
dF i dF
= = w, i = 1, 2, 3
dt dt
Here r (F i )= R er (F i ) is the position vector along the rotor axis to the midpoint of the
ith cup, er (F i ) is a unit vector, I is the moment of inertia of the rotor-cup assembly
and F (F i ) is instantaneous aerodynamic force on the ith cup and e z is the unit
position vector parallel to the axis of rotation. Rearranging (8) and taking the dot
3
dw 1
dt
=
I
å k ×r (F i )´ F (F i ) (9)
i= 1
18
where
ìï æV ö2 æV ö
ïï çç cos (F )- w ÷
ïï K ÷ , if çç cos (F i )- w ÷ ÷> 0 , cos (F i )> 0
1ç
è R i ÷
ø çR
è ÷
ø
ïï
ïï
ïï
ï æ ö2 æV ö
ez ×r ´ F (F i )= ïí - K 2 çç w - V cos (F i ) ÷ ÷ , if çç cos (F )- w ÷ ÷< 0, cos (F i )> 0
(10)
ïï çè R ÷
ø èç R i
ø÷
ïï
ïï
ïï
ïï æ V ö2
÷
ç
ïï - K 2 çç w + cos (F i ) ÷ , if cos (F i )< 0
è R ÷
ø
ïî
and
r A C D 1R 3
K1 =
2I
(11)
3
r A C D 2R C
K2 = = K 1 D2
2I C D1
Here A = p r 2 and K 1, K 2 represent the ratio of aerodynamic to inertial forces for the
single cup case illustrated in Figure III-3 and the case where the rotor is spinning in zero
wind, respectively. Note that K 1 > K 2 because C D 1 > C D 2 which highlights the inherent
19
IV. CALIBRATION
manufacturers were used. The dimensions of the wind tunnel outlet (1m x 0.8m) was
large enough, compared with the dimensions of the cup anemometers, to allow the
neglect of any boundary layer effects on the cup anemometer and measured velocity
Four types of cup anemometers were calibrated. (Figure IV-1, Table IV-1). Only the
Second Wind C3TM had not been used previously. The specifications of the four
Carbon-Fiber- 45 22
ThiesTM Compact Hemisphere
Reinforced Plastic
119 39
ThiesTM Classic Hemisphere Anodized Aluminum
20
Figure IV-1: Cup anemometers used for the current study (a): ThiesTM First Class Cup
Anemometer, (b): Second Wind C3TM Cup Anemometer, (c): ThiesTM Sensor Compact
Because each anemometer has different cup shapes and different material roughness,
their drag coefficients (CD) are different. For a given wind condition, each anemometer
coefficient and the different sizes of the anemometers. As a result, each cup
21
Each cup anemometer has a manufacturer’s linear calibration function. Most calibration
procedures specify that the rotational frequency output of the anemometer or another
output variable such as voltage must be measured at certain specific wind speeds. The
V=Af+B (12)
Here V is the wind speed (m/s), A is the slope of the calibration curve (m), f is the
anemometer’s frequency output in pulses per second and B is the offset (m/s). If the
bearing is friction-free, the friction moment should be zero and B would be zero. In fact,
the bearing is not friction-free, so the friction moment is non-zero and the cup
anemometer will remain motionless until the wind reaches a finite non-zero speed. The
typical starting velocity of cup anemometer is between 0.3 - 0.5 m/sec [31]. Some
authors claim that a non-linear expression should be used instead of a linear one,
especially at low wind speeds [32] [33]. However, the linear expression is accurate in
normal working conditions and recommended in standard calibration processes [34] [35]
[36].
For some cup anemometers the raw data output is standard AC sine wave
corresponding to the turning frequency of the cup anemometer’s rotor. The cup
frequency of the anemometer, fr, by the number of pulses per revolution given by the
cup anemometer, Np. The calibration function in the datasheet has the same format as
(revolutions/s)
Figure IV-2 shows two calibration functions from ThiesTM First Class and Second Wind
C3TM cup anemometers. The output of Second WindTM C3 is standard AC output and
ThiesTM First Class has frequency output. The calibration function of Second Wind C3 has
been converted into frequency output in the equation. Both cup anemometers have the
same conical-shaped cup but cup sizes and arm lengths are different. The calibration
function for the ThiesTM First Class shows a large offset which suggests that it’s the static
friction of its bearings is greater than the Second Wind C3. (Thus, the wind speed
needed to start a static ThiesTM First Class cup anemometer will also be greater. [28].)
Figure IV-2: The calibration function of ThiesTM First Class Cup anemometer
23
Even though the cup anemometers were calibrated by the manufacturers before
shipping, they are re-calibrated individually before the experiment. The calibration
function was obtained for each cup anemometer. Figure IV-3 shows the linear
calibration of ThiesTM First Class cup anemometer in the wind tunnel vs typical
calibration function from the manufacturer. The slopes of the two curves are almost
identical but the offsets are different. The offset in the calibration by the manufacturer
is smaller than the offset in the current calibration equation obtained from the wind
tunnel experiment. The anemometer requires higher wind speeds to initiate than the
manufacturers specification. The reason could be that the bearing is probably worn.
25
20
V = 0.0468f + 0.4246
V = 0.0462f + 0.21
Velocity (m/sec)
15
Results from
10 Experiment
Results from
Datasheet
5
0
0 100 200 300 400 500 600
f (pulses/sec)
Figure IV-3: The calibration function of ThiesTM First Class Cup anemometer according
24
It is much easier to analyze the raw data for the frequency output than AC output. The
AC output was also observed to be noisier (results will be presented in the next section)
25
V. OVERSPEEDING ANALYSIS OF DIFFERENT CUP ANEMOMETERS
wind conditions by moving the anemometer out of and into a steady wind stream. After
each cup anemometer was calibrated and the calibration formulas are acquired, the
next step involved using the wind tunnel to simulate a rapid wind change. Figure V-1
shows a schematic of the experimental setup. The cup anemometer was smoothly but
rapidly moved in and out of the test section along a direction perpendicular to the wind
26
Figure V-2: Wind tunnel experiment setup (looking at outlet of the wind tunnel)
Upon removal of the anemometer from the wind tunnel, the velocity V in equation (10)
is zero and the equation (9) for the rate of change of angular speed of the rotor
becomes
dw
= - 3K 2 w2 = - k w2, t > t 0 , w (t 0 )= w0 (14)
dt
For a given cup anemometer, k will be a constant but will generally be different for
different anemometers
w0
w (t ) = (15)
1 + w0k (t - t 0 )
27
From Equation (15), it is clear that a cup anemometer having the same initial speed (t0)
but with a smaller k value will require less time to reach a given lower rotational speed.
In Figure V-1, the cup anemometer turns counter-clockwise while the generated wind
flows from left to right. If the cup anemometer is moved to the right (relative to the
wind direction) and removed from the wind stream, the rotation rate will slow gradually
and the cup will behave according to equation (15). This is the case where the
anemometer has an initial non-zero rotation tare and the wind suddenly stops blowing.
Note that, in the experiments, if the cup anemometer is moved out of the test section in
the opposite direction (i.e., to the left relative to the wind direction), rotation will stop
immediately. The reason for this is explained with the aid of Figure V-3.
dw 3 Mi
dt
= å 1 I
(16)
28
If the cups with moments M1 and M3 are pushed out the wind tunnel test section first
(moving downward from the top view in Figure V-3 (A)), the torque on these cups is
2
æ V ö
÷
M 1 = M 3 = - K 2w 2 and the torque on the remaining cup is - K 2 çç w + cos (F i ) ÷
÷
çè R ø
so anemometer will slow rapidly due the high drag force on the remaining cup. This
does not, however, represent the situation of how a cup anemometer responds to a
suddenly decreasing wind in the field. On the other hand, if the M2 leaves the wind
tunnel test section first (moving upward from the top view in Figure V-3 (B)), only one
cup has a negative torque, M 2 < 0 , while the remaining two cups are still driven by the
wind so d w / dt > 0 . In the latter case, the cup anemometer will initially speed up upon
removal from the tunnel before the velocity starts to decay after the entire
anemometer is removed from the test section. The circles in Figure V-4 and Figure V-5
show the transient increase in velocity as the anemometer is removed in this way.
29
Figure V-4: Wind speed measured by ThiesTM First Class Cup anemometer while
decreasing wind velocity situation in wind tunnel (15 m/sec wind speed)
30
Figure V-5: Two cases of overspeeding situation of ThiesTM First Class Cup anemometer
at 10 m/sec
Figure V-6 shows measurements made during a sequence that involves the removal of
the anemometer from the wind tunnel for a finite time, reinsertion into the test section,
removal and reinsertion. The cup anemometer responds more rapidly to an increase in
wind speed, faster than the response to a decrease in wind speed, overspeeding. The
cup anemometer underestimates the wind velocity as it is inserted into the test section
wind speed and overestimates the speed during the removal. This can be seen clearly
from the asymmetry in the measured and computed response to sudden changes in the
wind speed. It appears that some consideration of this asymmetry in response must be
accounted for to obtain a reliable measurement. Since in these experiments, both under
and over estimates of the wind speed are observed depending on whether the wind
31
speed increases or decreases, the corresponding curves will be referred to as under- or
overspeeding curves.
5
Wind Speed (m/sec)
3
Measured
wind speed
2
Actual wind
speed
1
0
0 20 40 60 80 100 120 140
wind speed
wind speeds (5m/s, 10m/s and 15m/s) for the four different cup anemometers listed in
Table IV-1. The goal here is to compare the degree of overspeeding for the four different
types. The experiment under the same setup is repeated at least three times and only
one set is plotted in the figures. The overspeeding curve is taken out from the set of
data and is aligned with other overspeeding curves for the comparison.
32
Figure V-7: Overspeeding curves for the ThiesTM First Class Cup Anemometer at various
wind speeds
Figure V-8: Overspeeding curves for the Second WindTM C3 Cup Anemometer at
33
Figure V-9: Overspeeding curves for the ThiesTM Compact Cup Anemometer at various
wind speeds
Figure V-10: Overspeeding curves for the ThiesTM Classic Cup Anemometer at various
wind speeds
34
The value of k in equation (15) for each cup anemometer was obtained from
experimental data as follows. Each wind velocity data set obtained for anemometers
suddenly removed from the wind tunnel (i.e. the wind speed undergoes a step change
from a finite speed to zero) was processed as follows. The calibration offset velocity for
each anemometer was subtracted from the actual velocity. To convert this to a
rotational speed the resulting difference was then divided by the anemometer rotor
radius R. The k in equation (15) was then fit to these data for each anemometer. The
Anemometer k value
Figure V-11 shows the overspeeding curves of all four-cup anemometers at wind speed
of 10 m/sec. The graphs show that ThiesTM Classic Cup Anemometer is less sensitive
than the other three-cup anemometers at responding to the decreasing wind and takes
longer time to decay to zero speed. The mean measurement uncertainty for each cup
anemometer is shown in the graph in m/sec and is calculated based on the combined
uncertainty in the wind speed, uncertainty in the test anemometer and uncertainty in
35
Figure V-11: Overspeed curves for the four anemometers at V = 10 m/sec (* denotes
The same trend is also observed for the wind speeds of 5 m/sec and 15 m/sec in
Appendix A.
36
VI. OVERSPEEDING ANALYSIS OF UNBALANCED CUP
ANEMOMETERS
calibrated prior to and after the measurement campaign or after one-year. The cup
As a result it is difficult to frequently check the physical integrity of the instrument after
the installation. A rotational imbalance could occur when dirt has accumulated on the
cup or there is uneven wear of bearings. This generally would not be identified quickly,
anemometers could be identified in the wind speed times series, it might be possible to
detect such abnormalities in the output data in real time. This is the motivation of this
part of the investigation. The objective of the experiments is to assess the effect of
imbalance on overspeeding.
To simulate the unbalanced cups in the laboratory setup, four modifications of the
ThiesTM First Class cup anemometers (Figure VI-1) were tested in the wind tunnel at
different velocities (Table VI-1). To obtain an imbalance, rare earth magnets were placed
on one or two cups to change the weight of the cups, and, thus, obtain an imbalance
that would affect the rotation. Two different magnets were used, 0.578 g and 1.12 g.
ThiesTM manufactured a cup anemometer with a metal ball (2.23 g/each) glued in the
center of each cup to change the weight of the cup without altering its aerodynamics
like other configurations. The moment of inertia of the cup is altered for all cases.
37
Additional weight (g)
The “new” calibration functions for each cup anemometer with various weighted cups
are shown in Figure VI-2. Their differences are hardly noticeable as expected. The
calibration function of one-cup weighted is almost identical with the regular cup
anemometer and the calibration functions of other three conditions are off more (Table
VI-2).
38
25
Regular
20
One-Cup
Weighted
Velocity (m/sec)
15
One-Cup
Weighted
10 (Heavy)
Two-Cup
Weighted
5
Three-Cup
Weighted
0
0 100 200 300 400 500
f (pulse/sec)
Figure VI-2: Calibration function of the Thies™ Classic Cup Anemometers with different
weighted cups
The overspeeding curves of different weighted cup anemometer are shown in Figure
anemometer shows that for the lighter weight the overspeeding curves are essentially
the same. When heavier weights (6.72 grams) are used, the one-cup weighted curve
shows noticeable fluctuations. Compared the regular curve with the two-cup weighted
curve and three-cup weighted curve, it is observed that it takes longer time for the cup
39
anemometer to reach the same ending speed. The one-to-one comparison can be found
in Appendix B.
Figure VI-3: Overspeed curves comparison of the Thies™ First Class Cup Anemometers
with different weighted cups at wind speed of 10 m/sec (* denotes “mean total
uncertainty”)
Anemometer k value
Regular 0.027
40
Even though curves show only small deviations from the unweighted cup curve, it can
be seen from Table VI-3 above that the k values are different for each weighted case cup
anemometer. The largest deviation is being for the Three-cup weighted case. It takes a
longer time for the three-cup weighted anemometer to stop compared with other
unbalanced cup anemometers. By frequently obtaining k values from field data and
comparing it with the nominal value, the health of the anemometer can be determined.
41
VII. PRELIMINARY FIELD DATA EXAMINATION
Figure VII-1: Measured wind speed vs. time (upper graph) and the inverse of the wind
Figure VII-1 is a sample of actual measured wind data (u). From the experiments to
measure the constant k discussed in the previous section it is seen that the wind velocity
measured by an anemometer under a large and rapid decrease in wind speed can be
(17)
Using the idea that, under conditions when there is a large magnitude and sudden drop
in wind speed the slope of the inverse wind speed time series 1/u should be
approximately k. The 1/u graph shown below the time series for u exhibits
42
three regions where the slope corresponds to k. It is speculated that this approach could
be used to identify regions in wind time series that may have recorded large over
43
VIII. CONCLUSIONS AND FUTURE RESEARCH
The present study has shown that the dynamical properties of the cup anemometer can
result in over or underestimate of the wind speed depending on the behavior of the
wind. In addition, the time constant for the instrument (proportional to 1/k) is shown to
gust that would correspond to an increase in wind speed, the recorded speed spends a
‘long’ time at a higher speed as the disturbance decays back to the initial speed. For
turbine power curve measurements, the IEC 61400-12 standard recommends averaging
the wind data over 10-minute periods. For that reason, the 10-minute averaged wind
speed from the wind data that cup anemometer collected is higher than the actual
mean wind speed over this time period. This is results in an overestimation of power
output.
Because the overspeeding curves from the cup anemometer could be used to pinpoint
when the overspeeding happened during the post wind data analysis. It could be used
to identify where the data is corrupted due to the overspeeding during the data post-
processing. If over a certain amount of data in 10 minutes window is corrupted, the data
should be removed.
However, the current experiment assumes there is no vertical component of wind speed
which is too ideal for what really happens in the natural environment. For future work,
the same experiment should be repeated when vertical component of wind speed is
44
Even though cup anemometers react to an increase in wind speed more rapidly than for
a decrease in wind speed, the experiment for decreasing wind speed scenario should
The further step is to repeat experiments under the turbulent flow conditions. It would
behavior.
45
Part II
46
I. OVERVIEW AND OBJECTIVES
A wind turbine transforms the kinetic energy in the wind to mechanical energy in a shaft
and converts it into electricity. The power generation directly depends on characteristics
of the wind turbine blades and the interactions between the wind turbine blades and
the wind.
Wind turbines generate sound in different ways, both mechanical and aerodynamic.
Even though wind turbine technology has advanced of the last few decades and wind
turbines have become much quieter, sound from wind turbines is still one of the most
studied from the point of view of environmental impact and plays an important role in
siting criterion. The challenge for wind turbines to be installed near cities is public
acceptance and acoustic noise from wind turbines remains an issue frequently cited by
47
Wind turbine noise consists of mechanical noise that includes the noise from the fans,
generator, gear box, etc., and aerodynamic noise which is originated from the
interaction between the rotor and the wind. Manufacturers have been able to
significantly reduce the mechanical noise. Aerodynamic noise remains an issue. This part
calculate the acoustic pressure at a far field receiver location. The results obtained are
compared with experimental data. The purpose of the investigation is, to predict
aeroacoustic noise at different wind speeds and at different rotational speeds of the
turbine.
Unsteady operating conditions such as variable angles of attack produces acoustic noise
and can also cause erosion or even splitting of the blade along the leading and trailing
edges [40]. Recently, some researchers have used the sound radiated from operating
wind turbines to monitor their structural health [41]. Wind turbine designers continue
Sound measurements of actual wind turbines are impractical to perform at scale due
the size of the wind turbine. Field measurements are also difficult because it is hard to
separate wind turbine noise from other sounds generated by nearby objects. Modeling
appears to be a good option but the numerical simulation of sound generated from
turbulent wind is also difficult. Nevertheless, for certain geometries and idealized flow
48
Wind turbines are operated at different rotational speeds depending on the incoming
wind speeds and angles. Therefore, the magnitude and extent of acoustic noise
emission varies. The current study extends the work of Ranft, Ameri, Kaltschmit, &
Alexander, [15] and focuses on the prediction of aeroacoustic noise generated from
NREL Phase VI Wind Turbine Blade at different incoming wind speeds with various
rotational speeds. The goal of this study is gain a better understanding of the
The simulations are accomplished using the commercial software ANSYS FLUENT
49
II. NOISE AND SOUND FUNDAMENTALS
vibrating surface. Even flows generated by a constant pressure difference can result in
pressure fluctuations that we hear as sound [42]. A vibrating surface is a source that
emits pressure fluctuations and the fluctuations propagate as sound waves. The speed
of sound in air is approximately 340 m/s at standard temperature and pressure. Sound
waves can propagate through any phase of matter. The SI unit for pressure is Pascals
and for sound pressure denotes the force per unit area perpendicular to the direction of
the sound wave propagation. The lowest audible sound pressure for the human ear is 20
mPa (10 billionths of an atmosphere) and occurs in the frequency range 3000-4000 Hz.
Pain is experienced at sound pressures of 60 Pascals. The decibel scale (not an SI unit be
convenient way to describe sound relative to the lowest audible sound for a normal
human ear. The human ear is less sensitive to low audio frequencies, therefore, decibel
A-Weighting, dB(A), is often used to express of the relative loudness of sounds in air as
The sound power or acoustic power is the sound energy constantly transferred from the
sound source. A sound source has a given constant sound power that does not change if
50
A sound source produces sound power and this generates a sound pressure fluctuation
in the air. Sound power is the distance independent cause of this, whereas sound
pressure is the distance-dependent effect. Sound Power is the energy of sound per unit
of time from a sound source and the Sound Power Level (Lw) can be expressed as a
𝑁
𝐿𝑤 = 10 𝑙𝑜𝑔10 (𝑁 ) (1)
𝑟𝑒𝑓
where, Lw is the Sound Power level (dB), N is the sound power (W), and Nref = 10-12 W.
The Sound Pressure Level (SPL, Lp) is a property of sound at a given observer location
and is a ratio of the absolute Sound Pressure and a reference level (the lowest human
𝑝2
𝐿𝑝 = 10 𝑙𝑜𝑔10 (𝑝2 ) (2)
𝑟𝑒𝑓
where, Lp is the sound pressure level (dB), p is the sound level (Pa), and pref = 2×10-5 Pa.
Any unwanted sound is termed noise. Acoustic noise can be anything from quiet to
danger to the health and welfare of the public [43]. Human perception of noise varies.
Whether sound is identified as noise or not, it does not only depends on the sensitivity
of the listener, but also on the sound duration and the surroundings. For examples,
people can sleep in a moving train or a flying jet but get annoyed while hearing the
51
2. NOISE STANDARDS AND REGULATIONS
There are direct links between noise and health [44] [45]. Problems caused by noise
include stress related illnesses, high blood pressure, hearing loss, sleep disruption and
low productivity [43]. Although human perception of sound varies from person to
person, not only could high-decibel damage hearing, but long term-exposure to a low-
http://www.hearingcarecentre.co.uk/Info_page_two_pic_2_det.asp?art_id=6052&sec_id=3052
Table II-1: Health and safety executives noise regulation guidelines on exceeding safe
level
countries, noise regulations defined upper bounds for the noise to which people may be
exposed and it also depends on the areas and time (Table II-2).
52
Table II-2: Recommended noise level standards and guidelines by WHO and selected
53
III. WIND TURBINE AERODYNAMICS
The majority of wind turbines is horizontal-axis wind turbines (HAWT) [23] and is the
There are two forces, lift and drag, and one momentum, pitch, that act on a HAWT wind
turbine blade (Figure III-1). The lift force is perpendicular to direction of the oncoming
air flow and is the result of unequal pressure on the upper and lower airfoil surfaces.
The drag force is parallel to the direction of the oncoming air flow and is due to viscous
friction forces at the surfaces and to unequal pressure on the airfoil surfaces facing
Figure III-1: Forces and moment on an airfoil section, α, angle of attack; c, chord [22]
The tip speed ratio is the ratio of the blade tip speed over wind speed.
where Ω is the angular velocity (in radians/sec) of the tip of the blade, V is the wind
54
Generally a low speed wind turbine has the tip speed ratio value between 1 and 4 and a
high speed wind turbine has the tip speed ratio value between 5 and 9. Burton’s [22]
research suggests that the wind turbine can run at near maximum power coefficient
At the same tip speed ratio, a blade with a large span has a low rotational speed.
3. AIRFOIL BEHAVIOR
There are three basic flow regimes associated with airfoils: the attached flow regime,
the high lift/stall development regime and the flat plate/fully stalled regime [22]. In the
attached flow regime, lift increases with the angle of attack. In the high lift/ stall regime,
the lift coefficient increases and reaches the peak when it stalls. When the angle of
attack exceeds a certain value, stall occurs and flow separates on the upper surface. It is
applicable airfoil for wind turbine blade will improve the efficiency.
4. TWISTED BLADE
The blade tips of some modern wind turbines are designed using a thick airfoil for high
lift to drag ratio. If there were no structural requirements, this is how a wind turbine
blade would be proportioned. However, the blade needs to support the lift, drag and
gravitational forces acting on it. The root region of a wind turbine blade is designed
using a thick version of the same airfoil. The crucial factors for choosing airfoil are to
have the maximum lift to drag ratio and low pitch moment. The closer to the tip of the
blade, the faster the blade moves through the air and the greater is the apparent wind
55
angle. Therefore, to optimize the angle of attack along the entire blade, the blade must
twist from root to the tip. In order to achieve the maximum lift and efficiency for some
long blades, the chord length, thickness, twisted angle and the shape of airfoil varied
alone the blade (Figure III-2). These requirements are a challenge for manufacturers.
56
IV. NOISE MECHANISMS OF WIND TURBINES
The sound generated by wind turbine operation can be divided into four categories:
a) TONAL
interaction with a rotor blade surface, vortex shedding from a blunt trailing edge,
or unstable flows over holes or slits or a blunt trailing edge. The Department of
Environmental Protection of the State of Maine has, for the purposes of wind
protected location, the 10 minute one-third octave band sound pressure level in
the band containing the tonal sound exceeds the arithmetic average of the
sound pressure levels of the tow contiguous one-third octave bands by 5dB for
center frequencies between 500 Hz and 10,000 Hz and by 8dB for center
frequencies between 160 Hz and 400 Hz, and 15 dB for center frequencies
between 25 Hz and 125 Hz. (An octave band is a frequency band where the
highest frequency is twice the lowest frequency. A third octave band is a third of
57
b) BROADBAND
c) LOW FREQUENCY
Low frequency noise (20 Hz -120 Hz) was the focus of a recent study in Denmark
[47]. It was shown that large wind turbines (> 2 MW) produce low frequency
d) IMPULSIVE
thumping sounds that vary in amplitude with time. There are variety of proposed
wind speed over the rotor area, and interactions between sound fields from
neighboring turbines [48]. The mechanism for impulsive noise production is not
The noise generated from the operating wind turbines can be classified into two types,
mechanical noise and aerodynamic noise. Figure IV-1 summarizes the mechanical and
aerodynamic sound sources of a wind turbine and indicates their respective sound
power levels; a/b refers to airborne noise and s/b refers to structural noise.
58
Figure IV-1: Horizontal axis wind turbine noise sources (Romero-Sanz & Matesanz,
2008)
a) Mechanical noise
knocking against each other, such as rotating components in the gear box and
the generator, the cooling fans, yaw system, pumps and compressors.
Mechanical noise tends to be more tonal and narrowband that bother humans
more than broadband sound [50]. Therefore, many countries have regulations
which stipulate distance increases between wind turbines and the nearest
buildings. Mechanical noise can be transmitted either through the air or through
the wind turbine structure. In the latter, sound travels along the structure of the
wind turbine then into the surroundings through different surfaces, such as the
59
b) Aerodynamic noise
sound generally increases with rotor speed. Figure IV-2 shows the six main
regions along the blade that are considered to independently generate acoustic
noise. The noises from these six regions are produced different at various
Figure IV-2: Aerodynamic noise sources around a turbine blade due to wind flow, U
The six sources of aerodynamic noise identified in Figure IV-2 can be divided into
three groups: Low frequency sound, Inflow turbulence and airfoil self-noise
(Table IV-1).
60
Table IV-1: Wind Turbine Aerodynamic Noise Mechanisms (From Wagner, Bareib, &
Guidati, 1996)
3. ACOUSTIC SOURCES
Lighthill [52] [53] proposed that aerodynamic sound sources can be modeled as series of
region surrounded by a large fluid region at rest. Physically, monopoles result from a
fluctuating mass flow [54]. Dipoles are formed when there are fluctuating forces. When
61
fluctuating Reynolds stresses appear, quadrupoles are formed. It is referred to as
a) MONOPOLE
A monopole radiates sound equally in all directions and is the simplest acoustic
flow velocity in the flow field and M is Mach number). Examples include tire, and
http://en.wikibooks.org/wiki/File:Mono.GIF#/media/File:Mono.GIF
b) DIPOLE
A dipole is the superposition of two monopole sources that are out of phase.
This results a sound radiation field such as that shown in Figure IV-4. The
62
are normally the result of vortex shedding as would occur in turbulent flow over
a rod or cavity. The order of the acoustic power of dipole is similar to U3M3.
http://en.wikibooks.org/wiki/File:Dipole.GIF#/media/File:Dipole.GIF
c) QUADRUPOLES
source can be obtained by the superposition of two dipole sources of the same
strength that are out-of-phase (Figure IV-5). Quadrupoles arise from turbulence
and the order of the acoustic power of quadrupole is similar to U 3M5. One
63
http://en.wikibooks.org/wiki/File:Quadpole.GIF#/media/File:Quadpole.GIF
64
V. NUMERICAL ANALYSIS
1. PHYSICAL MODEL
Ranft et al. (2009) carried out computational simulations of the acoustic noise emitted
by S809 blade, and validated the computations by comparing the results with
experiments carried out at NASA AMES Research Center [55]. The S809 blade airfoil
blade has a baseline span of 5.029 m and the chord lengths vary from 73.7 cm at 25% to
35.6 cm at the blade tip (Appendix E). The numerical model was validated. The same
2. MESH GENERATION
Two types of meshes were used corresponding to structured and unstructured grids.
A structured grid consists of regular mesh lines while an unstructured grid is identified
by its irregular connectivity. The disadvantage of structured grids is that it takes more
65
Figure V-2: An example of unstructured grids
An unstructured grid is more flexible and adapts well to complex geometries. The mesh
can be any shape (Figure V-3). Due to the irregularity of the structures there is, however,
Figure V-3: Structured grids consisting of triangular, polygonal meshes and orthogonal
meshes
It is usual practice to have a refined mesh close to solid surfaces in the near-field region
so that the flow structures (usually involving large velocity gradients) in these areas can
be sufficiently well resolved. Depending on the type of flow, the mesh can be gradually
coarsened away from boundaries. To reduce computational costs for this study, an
unstructured grid is used for the calculation of the mean flow away from the blade and a
66
structured grid is used in the region close to the blade to permit resolution of the
The grid was created using the mesh generator "Pointwise". The computational domain
tetrahedral and prismatic elements as displayed in Figure V-4. The mesh is refined
67
To be able to compare the simulation results with the experimental results, one blade of
the two-bladed turbine is modeled in a 180° sector with the boundary conditions as
show in Figure V-5. The velocity vectors are set to be perpendicular to x-z place as the
upper stream boundary condition. The ambient pressure of 101,325 Pa is defined by the
A non-slip condition is applied on the wall of the blade (Figure V-6). In this model, the
blade is not moving so a moving frame of reference is set to simulate the rotating wind
turbine blade.
68
3. TURBULENCE MODEL
Figure V-7 shows a schematic of the wind velocity impinging on the blade.
Here Vrot is the rotational velocity of the blade, Vw is the incoming wind speed
V Vrot
2
Vw2 is the resultant velocity, ω is the angular velocity (RPM).
The Reynolds number is taken to be 𝑅𝑒 = 𝑉∞ 𝐿/𝜐, where L is the blade’s chord length
and υ is the kinematic viscosity. The lowest wind speed used in the simulation is 7 m/sec.
which, for a blade chord of 0.5 m, gives Re = 1.3 x 106. Hence, the flow is assumed to be
turbulent.
There are several turbulence models in FLUENT package [56]. The k-ε model proposed
by W. P. Jones and B. K. Launder is widely used in industry due to its stability and
convergence [57]. This model includes two equations, the k (turbulence kinetic energy)
69
equation and ε (dissipation rate) equation. The model is good for high Reynolds flow
The k-ω turbulence model, first introduced by Kolmogorov in 1942 [58], is also a two-
parameter turbulence model with the turbulence kinetic energy term (k) and the
dissipation per unit turbulence kinetic energy (ω). The k-ω turbulence model performs
better at near wall layers than the k-ε turbulence model and has been shown to fit best
for the type of application considered here [59]. To avoid the effects of pressure
induced separation a version of the k-ω model known as Shear Stress Transport k-ω or
SST k-ω turbulence model was employed in this study. The SST k-ω model is a two
equation eddy-viscosity model that combines both k-ε and k-ω turbulence models. The
k-ε model part is in effect in the free stream while the k-ω model part accounts better
Simulations were performed in a steady state simulation and the Moving Reference
There are two steps for computing sound using the FW-H (Ffowcs Williams and
solution must be obtained so the pressure, velocity and density on the selected source
surface are obtained. Next, sound pressure signals are computed using the source data
collected during the first step at the user-specified receiver locations. For the current
70
study, the model has a single rotating reference frame; therefore a steady state FW-H
The wind turbine used in the actual experiment has a hub height of 16 m. The ground
effects of sound propagation were not taken into account in the computation. Table V-1
& Figure V-8 shows the four receiver locations related to the wind turbine.
The wind turbine was set as the acoustic source and four acoustic receivers were
located to measure the noise from near-to far-field using the results of the FW-H
71
acoustic model. Upon completing of the simulation a Fast Fourier Transform (FFT)
analysis was applied to interpret the spectral distribution of the sound pressure level.
Broadband noise occurs of the frequency range of 5-6 and 10 kHz and includes the
inflow turbulence noise and the airfoil self- noise [60]. In many practical applications
involving turbulent flow, the noise has no distinct tones and the sound energy is
Using the broadband noise source model available in the ANSYS FLUENT software
package enables the quantification of the local contribution to the total acoustic power
level generated by the flow. The source model can be used to extract useful diagnostics
on the noise source to determine which portion of the flow is primarily responsible for
the noise generation. However, in contrast to the FW-H model, the broadband source
The acoustic power generated by isotropic turbulence without mean flow is calculated
using Proudman’s formula which has been derived from Lighthill’s acoustic analogy
(quadrupole sources). Proudman’s formula gives the acoustic power (in W/m 3) per unit
PA 0M t5 (5)
2k
where M t ,
a0
72
βε is a model constant that is set to 0.1 in ANSYS FLUENT based on the calibration of [61]
using direct numerical simulation of isotropic turbulence. k and ε are constants from the
ANSYS FLUENT can also show acoustic power (Lp), in dB, which is computed from
P
L p 10 log A (6)
P
ref
where Pref = 10-12W/m3 is the reference acoustic power. Proudman’s formula gives an
approximate measure of the local contribution to total acoustic power per unit volume
73
VI. RESULTS
In this section results are presented first for the sound pressure distribution on the wind
turbine blades using the broadband model and then the far field results using the FW-H
model. The simulations were performed for 5 different wind speeds, 7m/s, 10m/s, 15
The flow field and streamlines at different wind speeds and rotational speeds were
computed using and are presented below. At lower wind speeds (lower angle of attack)
of 72 rpm, the flow is fully attached on the active part of the rotor blade. However,
when the wind speed increases, the central portion of the blade begins to stall. Blade
surface streamlines in Figure VI-1 show that the boundary layer separates at the leading
edge from root to tip. When the wind speed is 25 m/s, the streamlines show it is a fully
detached flow. It is also observed that the boundary layer flow is not only spanwise
from root to tip, but also streamwise from trailing edge to leading edge.
Figure VI-2 exhibits the flow at section z = 5 m is attached at 7 m/sec but starts to
separate at 10 m/sec.
74
10 m/s
25 m/s
The surface acoustic power level in Figure VII-3~6 describes the acoustic power per unit
area generated by boundary layer turbulence on the surface of the blade using the
75
Broadband Sound Source model described in Section V-5. The strength of the sound
sources on the suction sides of the blade increases with higher wind speeds. This
explains why flow separation occurs on the suction side of the blade at higher wind
speeds. It can also be observed that the sound sources on the pressure side move
towards the trailing edge at higher wind speeds. A maximum in the Surface Acoustic
Power Level always occurs at the leading edge of the blade tip which can be explained
by higher velocity and pressure values occurred at the blade top due to the rotation.
76
Figure VI-3: Surface Acoustic Power Level (dB) on blades at 36 RPM
77
Figure VI-4: Surface Acoustic Power Level (dB) on blades at 54 RPM
78
Figure VI-5: Surface Acoustic Power Level (dB) on blades at 60 RPM
79
Figure VI-6: Surface Acoustic Power Level (dB) on blades at 72 RPM
80
The maximum surface acoustic power levels under different conditions are shown in
Figure VI-7. It is observed that the maximum surface acoustic power level increases with
the increasing wind speed in the beginning as expected. At low rotation rates, a local
minimum occurs as the wind speed increases and then the surface power level rises
again.
Figure VI-7: Maximum Sound power levels for wind speeds 7 m/s through 25 m/s at
There is currently no way to measure the acoustic noise caused by boundary layer
turbulence on blade surface in the lab or in the field and so validation of the code in this
81
3. FAR-FIELD AERODYNAMIC NOISE AT DIFFERENT WIND SPEEDS
The noise radiated to the observer in a far-field is investigated by using the integral
The observer locations are shown and listed in Figure V-8 and Table V-1. The simulation
results of aerodynamic noise under different wind speeds at the various rotational
It is expected that the emitted Sound Pressure Level (SPL) is louder for higher wind
speeds and rotational speeds. The receiver 4 has the longest liner distance from the
82
noise source (blade), the acoustic noise in the minimum compared with noise at other
receiver locations.
Wind direction changes continuously and so it is often the case that the rotor axis of a
wind turbine is not always aligned with the wind direction that is, the wind turbine is
yawed. When a wind turbine is in the yawed condition, even in a steady wind, the angle
of attack on each blade is continuously changing and the loads on the blades are varied.
There are six (6) yawed angles are simulated for a wind turbine at 72 rpm. The Surface
Acoustic Power results are shown in Table VI-2. It is noted that the Surface Acoustic
83
Power is the lowest at lower wind speeds when the wind turbine is not yawed. This is
Table VI-2: Surface Acoustic Power (dB) of different yawed angles at 72 rpm
Figure VI-9 shows the Surface Acoustic Power Level (dB) on blades at different yawed
angles. There is no obvious difference. The maximum Acoustic Power Level always
Table VI-3 shows the far-field aerodynamic noise of a yawed turbine at 72 rpm. The
receiver locations are shown in Table V-1 and Figure V-8. Even though the yawed angles
vary; there is no significant difference in Sound Pressure Level at the receiver locations.
84
Figure VI-9: Surface Acoustic Power Level (dB) of 72 rpm at wind speed of 7 m/sec on
85
Receiver -1 Receiver-2 Receiver-3 Receiver-4
7 m/s 64.2 58.5 56.1 44.6
10 m/s 66.2 60.3 58.0 46.4
15 Degree 15 m/s 68.5 62.1 60.0 48.4
20 m/s 71.1 64.1 62.3 50.8
25 m/s 72.8 65.5 63.9 52.4
7 m/s 64.0 58.4 56.0 44.5
10 m/s 66.3 60.4 58.1 46.5
10 Degree 15 m/s 68.7 62.2 60.1 48.6
20 m/s 71.2 64.1 62.4 50.8
25 m/s 72.9 65.5 63.9 52.4
7 m/s 64.0 58.4 56.0 44.5
10 m/s 66.4 60.4 58.1 46.5
5 Degree 15 m/s 68.7 62.2 60.2 48.6
20 m/s 71.2 64.1 62.4 50.9
25 m/s 72.9 65.5 63.9 52.4
7 m/s 64.0 58.4 56.0 44.5
10 m/s 66.3 60.3 58.1 46.5
0 Degree 15 m/s 68.6 62.1 60.1 48.5
20 m/s 71.1 64.0 62.3 50.8
25 m/s 72.8 65.3 63.8 52.3
7 m/s 63.9 58.3 55.9 44.3
10 m/s 65.8 59.9 57.6 46.0
-5 Degree 15 m/s 68.6 62.1 60.0 48.5
20 m/s 70.6 63.5 61.7 50.2
25 m/s 72.8 65.3 63.7 52.2
7 m/s 63.8 58.2 55.8 44.3
10 m/s 65.6 59.7 57.4 45.8
-10 Degree 15 m/s 68.3 61.8 59.8 48.3
20 m/s 70.5 63.4 61.7 50.1
25 m/s 72.3 64.8 63.2 51.8
7 m/s 63.6 58.1 55.7 44.1
10 m/s 65.3 59.5 57.2 45.6
-15 Degree 15 m/s 68.0 61.6 59.5 48.0
20 m/s 70.1 63.1 61.3 49.8
25 m/s 71.8 64.4 62.8 51.4
Table VI-3: The far-field aerodynamic noise of a yawed turbine at 72 rpm at receiver
locations
86
VII. DISCUSSION
There are not many acoustic measurement experiments involving wind turbine that can
be found in the open literature. The wind turbines tested in the literature have different
sizes and the blades are not representative of an S809 airfoil. Also, the acoustic noise
was measured at a distance from the turbine and not on the blade surface. For that
reason, the experiment results could not be used to quantitatively validate the acoustic
In Figure VII-1 and Figure VII-2, Huskey & Dam (2010) [63] and Rogers, Manwell &
Wright (2006) [64] reports that the sound pressure levels at the receiver locations do
not increase with wind speeds. The sound pressure level displays a minimum at
intermediate wind speeds which is in qualitative agreement with the 36 RPM rotor
100
Sound Power Level (dB)
95
90
85
80
75
4 5 6 7 8 9 10 11 12
Standardized Wind Speed (m/s)
Figure VII-1: Sound power levels as a function of standardized wind speed for the ARE
442 wind turbine (rotor diameter is 7.2 m) (Huskey & Dam, 2010) [63]
87
Figure VII-2: Measured sound pressure levels of a Southwest Whisper 900 wind
turbine (rotor diameter is 2.1 m) (Rogers, Manwell, & Wright, 2006) [64]
In contrast, another field study by Huskey (2011) [65] for which the turbine rotor
diameter is 13 m, sound pressure levels increase with the wind speed (Figure VII-3) and
Figure VII-3: Sound power levels as a function of standardized wind speed for the Gai
88
VIII. CONCLUSIONS
1. SUMMARY
The noise generated by the S809 airfoil has been investigated using a combination of
computational fluid dynamic simulation and two acoustic models. While a qualitative
Both rotational noise and broadband noise were analyzed. As in the earlier study (Ranft
et al., 2009) it was found that the leading edge and the blade tip are the primary source
of the broadband noise (Section VI-2). In all simulations, the wind turbine with the
slowest rotational speed (36 rpm) and lowest incoming wind speed (7 m/sec), the
loudest noises (80.8 dB) generated by the wind turbine blade tip and the leading edge
are still higher than the World Health Organization recommended noise level standards
(Table II-2). An increase of 3dB in Acoustic Power Level will double the sound source
energy. Therefore, the noise will increase with higher wind speed and/or faster rotation.
Modifications of the blade tip and the leading edge are recommended to reduce the
noise generated by a working wind turbine. Modifications of the leading edge could be
made to avoid or delay the flow separation that adds additional sound sources on the
Table VI-2 shows that the sound pressure level at Receiver 4 is below the WHO
recommended noise level (55 dB in daytime) even with the fastest rotational speed (72
89
rpm) and incoming wind speed (25 m/sec). This result could be used as a baseline to
define a zone around the wind turbine that where excessive noise will occur.
2. FUTURE WORK
In this work the wind profile is assumed to be uniform and the interaction of the wind
with the ground is not considered. The effect of the ground will be a velocity profile that
increases with height and, thus, the loading conditions on each blade will be different
depending on their position. The ground can also absorb and reflect sound waves
although this is expected to decrease rather than amplify the propagating sound waves.
The noise produced by the wind turbine tower also needs to be assessed since the flow
is also separated by the tower. This produces wakes behind the tower and low-
frequency noise (Table IV-1) is generated. Three-blade wind turbines are the most
frequently used wind turbines. Future modeling of a three-blade wind turbine would be
90
APPENDIX A
Overspeed curves comparison of the Thies™ Classic, Second Wind™ C3, Thies™ First
Class and Thies™ Compact Cup Anemometers at wind speed of 5 m/sec (* indicates
91
Overspeed curves comparison of the Thies™ Classic, Second Wind™ C3, Thies™ First
Class and Thies™ Compact Cup Anemometers at wind speed of 15 m/sec (* indicates
92
APPENDIX B
The overspeed curves comparison of manufactured ThiesTM First Class and one-cup
The overspeed curves comparison of manufactured ThiesTM First Class and one-cup
93
The overspeed curves comparison of manufactured ThiesTM First Class and two-cup
The overspeed curves comparison of manufactured ThiesTM First Class and three-cup
94
The overspeed curves comparison of one-cup weighted and one-cup weighted (heavy)
95
The overspeed curves comparison of one-cup weighted and three-cup weighted
96
The overspeed curves comparison of two-cup weighted and one-cup weighted (heavy)
The overspeed curves comparison of three-cup weighted and one-cup weighted (heavy)
97
APPENDIX C:
In 1952, Dr. Michael James Lighthill published a paper, “On Sound Generated
from a given fluctuating fluid flow with two major assumptions. The first assumption is
that the acoustic propagation of fluctuations in the flow is not considered. The second
assumption is the preclusion of the back-reaction of the sound produced on the flow
The Continuity and momentum equations are a fluid can be expressed as:
( ui )
0 (C.1)
t xi
( ui ) ( ui u j ) p ij
g i (C.2)
t x j xi x j
p is the statistic pressure of the flow field, ij stands for the stress tensor referred to as
the momentum j transported by fluid particles in the direction i per unit time and unit
ui u j 2
ij ij u k (C.3)
x j xi 3 xk
(C.1) multiplied by ui
( ui )
uj uj 0 (C.4)
t xi
98
Adding (C.4) to the momentum equation (C.2) leads to
u j ( ui ) u j p ij
uj uj ui
t t xi xi x j xi
(C.5)
( u j ) ( ui u j )
t xi
( u j ) Pij
(C.6)
t xi
0
Add c 2 to (C.6) then subtract
xi
dp
c 2 (C.7)
d is
Substitute p with either c2(ρ-ρ0) or (p-p0) and the molecular momentum transport
ij 0 (C.8)
~
P ij ( p p0 ) ij (C.9)
~
P ij c 2 ( 0 ) ij (C.10)
The Lighthill tensor, Lij, is the external force acting on the fluid
Lij ui u j p p0 c 2 0 ij ij (C.11)
99
0
The acoustic wave equation is derived from (C.6). So add c 2 to both sides of
xi
u j 0 Lij
c2 (C.12)
t xi xi
2 2 ui
0 (C.13)
t 2 xi t
2 u j 2 0 2 Lij
c2 (C.14)
x j t xi x j xi x j
2 0
2 2
2 Lij
c (C.15)
t 2 xi x j xi x j
1 2 p 2 ( p po ) 2 Lij
(C.16)
c02 t 2 xi2 xi x j
The right hand side represents the source term, where Lighthill’s stress tensor, Lij, is. The
left hand side described the propagation of the acoustic wave in both the spatial and
temporal domains. The equation was used to predict aircraft jet noise originally.
100
APPENDIX D:
Ffowcs Williams and Hawkings extended Lighthill’s equation and applied the motion of a
moving surface into the theory. The Ffowcs Williams-Hawkings (FW-H) equation does a
very good job of calculating the airfoil self-noise of a moving body. There are three
source terms in FW-H equation that represent, quadrupole, monopole and dipole
sources.
2 ( H ) 2 ( H )
2 ( HLij )
Pij ( f ) f 0 u i ( f ) f
2
c
t 2
xi
2
xi x j xi x j t xi
Quadrupole Dipole Monopole
Where
0, 𝑓 < 0
𝐻(𝑓) = { Heaviside function
1, 𝑓 ≥ 0
1, 𝑓 = 0
𝛿(𝑓) = { Dirac delta function
0, 𝑒𝑙𝑠𝑒𝑤ℎ𝑒𝑟𝑒
u u j 2uk
Pij p ij i ij
x j xi 3xk
101
APPENDIX E:
102
2. Twist convention is positive towards feather. Values listed are relative to zero twist at the 3.772-
m station [75% span on a 5.029-m blade]. Twist is 2.5 degree toward stall at the tip [on a 5.532-m
blade].
3. Each blade attaches to the hub at a point 0.508 m from the center of rotation.
4. There is a cylindrical section at the root that extends from 0.508 to 0.883 m. The airfoil transition
5. There is a transition from the cylindrical section to the S809 airfoil along the 0.883- to 1.257-m
region. The transition ends with a 0.737-m chord S809 airfoil at the 1.257-m span station.
103
BIBLIOGRAPHY
[2] I. Capellan-Perez, M. Mediavilla, O. Carpintero and L. J. Miguel, "Fossil Fuel Depletion and
[3] S. Shafiee and E. Topal, "When Will Fossi Fuel Reserves Be Diminished?," Energy Policy, pp.
181-189, 2009.
455-464, 2015.
[5] S. Pfenninger and J. Keirstead, "Renewables, Nuclear, or Fossil Fuels? Scenarios for Great
Britain's Power System Considering Costs, Emissions and Energy Security," Applied Energy,
[7] "Renewable Energy Capacity Statistics 2015," International Renewable Energy Agency
Report, 2015.
[8] L. Fried, L. Qiao, S. Sawyer and S. Shukla, "Global Wind Report, Annual Market Update
[9] Global Wind Energy Council, "Global Wind Report 2013 - Annual market update".
[10] "Annual Energy Outlook 2015 with Projections to 2040," United States Department of
104
[11] R. Wiser and M. Bolinger, "Wind Technologies Market Report," Department of Energy,
DOE/GO-102015-470271522014, 2015.
[12] M. Wolsink, "Wind Power and the NIMBY-myth: Institutional Capacity and the Limited
[13] "http://www.usatoday.com/money/industries/energy/2008-11-03-windturbines_N.htm".
[14] N. Pinder, "Mechanical Noise from Wind Turbines," Wind Energy, pp. 158-167, 1992.
[15] J. I. D. Alexander, A. Ameri, M. Kaltschmit and K. Ranft, Acoustic Analysis of the NREL Phase
[16] J. Patterson, "The Cup Anemometer," Trans.Roy. Soc. Canada, pp. 1-54, 1926.
[17] P. Coppin, "Cup Anemometer Overspeeding," Meteorol Rdsch, pp. 1-11, 1982.
[20] T. Robinson, "On the Theory of the Rob Anemometer, and the Determination of its
2008.
[22] T. Burton, D. Sharpe, N. Kenkins and E. Bossanyi, Wind Energy Handbook, Chichester: John
[23] J. F. Manwell, J. G. McGowan and A. I. Rogers, Wind Energy Explained, John Wiley & Sons
Ltd., 2009.
105
[24] "Basics of Turbulent Flow," 2005. [Online]. Available:
http://www.mit.edu/course/1/1.061/OldFiles/www/dream/SEVEN/SEVENTHEORY.PDF.
Rotor Geometry on Cup Anemometer Response," The Scientific World Journal, 2014.
[26] I. Strangeways, Back to basics: The ‘met. enclosure’: Part 6 — Wind. Weather, Blackwell
[27] O. Schrenk, "Errors due to inertia with cup anemometers in fluctuating winds," NZ Tech.
and A Procedure of Measureing the Wind Velocity," Physical Oceanography, pp. 173-186,
2004.
[29] E. Kaganov and A. Yaglom, "Errors in wind speed measurements by rotation anemometers,"
[30] T. Lutz and A. Sayers, "The Effect of the Boundary Layer Present in Wind Tunnels on the
Aerodynamic Drag of a Model Truck," University of Cape Town, South Africa, 1997.
106
[34] "MEASNET. Anemometer Calibration Procedure, Version 2," MEASNET, Madrid, Spain,
2009.
[35] "Standard Test Method for Determining the Performance of a Cup Anemometer or
2002.
[36] IEC, "Wind Turbine Generator Systems, Part 12: Wind Turbine Power Perfrmance Testing.
Performance: The Stationary Harmonic Response," The Scientific World Journal, 2013.
calibration results from cup and propeller anemometers. Influence on wind turbine Annual
[40] K. Gharali and D. A. Johnson, "Numerical modeling of an S809 airfoil under dynamic stall,
[41] C. Niezrecki, P. Poozesh, K. Aizawa and G. Heilmann, "Wind Turbine Blade Health
[42] J. Ffowcs-Williams and M. J. Lighthill, Film Notes for the Aerodynamic Generation of Sound,
[43] United States Environmental Protection Agency, "Noise Control Act," 1972.
107
[44] S. A. Stansfeld and M. P. Matheson, "Noise Pollution: Non-Auditory Effects on Health,"
[45] D. Halperin, "Environmental Noise and Sleep Disturbances: A Threat to Health?," Sleep
[46] F. K. Yuen, "A Vision of the Environmental and Occupational Noise Pollution in Malaysia,"
[47] H. Moller and C. Pedersen, "Low-Frequency Noise from Large Wind Turbines," Journal of
[48] R. Tonin, "Sources of Wind Turbine NOise and Sound Propagation," Acoustics Australia, vol.
[49] S. Oerlemans and J. Schepers, "Prediction of Wind Turbine Noise and Validation Against
[50] H. Klug, "Noise from Wind Turbines: Standards and Noise Reduction Procedure," Sevilla,
Spain, 2002.
[51] I. Romero-Sanz and A. Matesanz, "Noise management n odern wind turbines," Wind
[52] M. J. Lighthill, "On Sound Generated Aerodynamically. I. General Theory," Proceeding of the
Royal Society A: Mathematical, Physical and Engineering Sciences, vol. 211, pp. 564-587,
1952.
[53] M. J. Lighthill, "On Sound Generated Aerodynamically. II. Turbuence as a Source of Sound,"
Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences, vol.
108
[54] S. Wagner, R. Bareib and G. Guidati, Wind Turbine Noise, Berlin: Springer, 1996.
Larwood, "Unsteady Aerodynamics Experiment Phase VI: Wind Tunnel Test Configurations
Press, 1972.
[59] Y. L. Chang, S. L. Yang and O. Arici, "Flow Field Computation of the NREL S809 Airfoil using
Various Turbulence Model," ASME Energy Wee-96 Book VIII, pp. 172-178, 1996.
[60] R. Camussi, Noise Sources in Turbulent Shear Flows: Fundamentals and Applications,
Springer, 2013.
Turbulence," NASA Contract Report 93-74, NASA Langley Research Center, Hampton, VA,
1993.
[62] G. M. Lilley, "The radiated noise from isotropic turbuence revisited," NASA Langley
[63] A. Huskey and J. v. Dam, "Wind Turbine Generator System Acoustic Noise Test Report for
the ARE 442 Wind Turbine," National Renewable Energy Laboratory, 2010.
[64] A. L. Rogers, J. F. Manwell and S. Wright, "Wind Turbine Acoustic Noise," Renewable Energy
109
[65] A. Huskey, "Wind Turbine Generator System Acoustic Noise Test Report for the Gaia Wind
[66] J. F. Williams and D. Hawkings, "Sound Generation by Turbuence and Surface in Arbitrary
[67] J. Wyngaard, J. Bauman and R. Lynch, "Cup Anemometer Dynamics," Proc. Flow, Its
163-172, 2002.
110