Vous êtes sur la page 1sur 12

Structures Congress 2013 © ASCE 2013 1391

Response of Wind Turbine Towers to Seismic Loading


At Different Damping Ratios

S. Jerath1 and S. Austin2


1
Professor of Civil Engineering, University of North Dakota, 234 Centennial drive
Stop 8115, Grand Forks, ND 58202-8115; PH (701) 777-3564; FAX (701) 777-3782;
email: sukhvarsh.jerath@engr.und.edu
2
Graduate Student, Department of Civil Engineering, University of North Dakota,
234 Centennial Drive Stop 8115, Grand Forks, ND 58202-8115; PH (701) 777-3562;
FAX (701) 777-3782; email: saman.montazeri@my.und.edu

ABSTRACT
First the paper compares experimentally obtained values in the literature to those
obtained from the time history analysis by the finite element method using ANSYS
computer program for a 65 kW wind turbine when the turbine is subjected to the
Landers earthquake. It is shown that the experimental and numerical values compare
well at the damping ratio of 0.86 establishing the validity of the numerical method.
The effect of wind turbine sizes (65 kW, 1 MW, 2 MW, and 5 MW turbines),
damping ratios, base acceleration directions, and different earthquakes (Landers,
Imperial Valley, and Northridge) on the wind turbine responses are studied. The
responses studied are the peak accelerations and deformations on the top of the
nacelle, and the maximum von Mises stresses at the base of the wind turbine towers.

INTRODUCTION

The role of wind turbines in the production of alternate source of energy


continues to grow rapidly throughout the world as the electric power generated by
wind is increasing. The wind energy power grew by 20 GW in 2007 to 100 GW in the
world as shown by the annual report of the U.S. Department of Energy (DOE 2008).
The United States remains one of the world’s largest and fastest growing wind
markets in 2011 with wind power representing a 32 percent of all new electric
capacity additions in the United States accounting for $14 billion in new investments.
Roughly 6800 MW of new wind power capacity was added to the U.S. grid and the
total U.S. wind power capacity reached 47 GW by the end of 2011, enough electricity
to power 13 million homes annually. This growth is accompanied by technical
innovations allowing for the construction of larger and taller wind turbines with
longer and lighter wind blades. These advances have increased the efficiency of
power generation from wind energy.

Now, a large number of wind turbines are installed in regions that are more
prone to large earthquakes which led to increased interest in the seismic design of
wind towers. Traditionally, wind turbines have been analyzed by modal methods
Structures Congress 2013 © ASCE 2013 1392

(Clough and Penzien 1993) used in the design of buildings, but these methods are not
adequate for the design of wind turbines. The behavior of wind turbines is different
from ordinary structures under earthquake load because of the presence of a rotating
mass at the top of a slender tower and the effect of wind on its damping properties
(Hodges and Pierce 2002). It is necessary to analyze the wind turbine structures by
time history analysis. . The wind tower structure consists of a tower fixed at the base,
a nacelle that sits at the top of the tower and contains the rotor, and a hub attached to
the front of the nacelle to which wind turbine blades are attached. First the paper will
compare experimentally obtained values in the literature (Prowell et al. 2009) to those
obtained from the time history analysis by finite element method using
ANSYS(2007) computer program for the 23 m (75.5 ft) high, 65 kW wind turbine.
The experimental model is shown in Figure 1, and the finite element model is shown
in Figure 2. The paper will also present the effect of wind turbine size, damping, base
acceleration direction and different earthquakes on the response of wind turbines
capable of producing 65 kW, 1 MW, and 5 MW of electric power. The investigation
should be of interest to the design professionals as a guidance to pick the damping
ratio and earthquake load direction for designing and analyzing forces in the wind
turbines.

COMPARISON OF EXPERIMENTAL TEST AND NUMERICAL ANALYSIS

The east-west component of the Landers earthquake (Prowell et al. 2009) was
applied as an axial horizontal excitation through a shake table in the experimental

Figure 1. 65 kW wind turbine experimental model

study. In the numerical analysis by the finite element method a horizontal base
excitation simulating the Lander’s earthquake identical to the experimental test was
applied. In both cases the rotor was parked with one blade oriented downward,
parallel to the tower. The uni-axial horizontal motion at the base of the tower was
applied in the direction perpendicular to the rotor’s axis of motion. The tower and the
blades are modeled by shell elements, the hub and the nacelle are modeled by the
solids elements in the finite element model. First the modal analysis is performed on
the finite element model to calculate the natural frequencies and the size of the time
Structures Congress 2013 © ASCE 2013 1393

step in the time history analysis of the finite element model. The first and second
mode natural frequencies are calculated as 1.7 Hz and 11.7-12.3 Hz respectively in
the experiments. The first and second mode frequencies are found to be 1.65 and 9.14
Hz respectively from the numerical analysis. The difference in the second mode
frequency may be due to the inaccuracies of material properties defined in the model.

Nacelle (SOLID186)

Hub (SOLID186)

16 m Blade (SHELL181)
23 m

Tower (SHELL181)

Z
Y
X

Figure 2. 65 kW wind turbine finite element model

The mode shapes of the 65 Kw turbines from experiments and the computer analysis
are shown in Figure 3. The equivalent viscous damping at the first natural frequency
is calculated from the experiments using the log decrement method (Chopra 2012),
and is found to be 0.86%. The effect of different modes on the behavior of a wind
turbine is found by calculating the ratio of the effective mass to the total mass for
each mode as follows:

={ } { }
(1)

Where, is the participating factor for the ith mode, ϕi is the eigen vector
representing the mode shape of the ith frequency, and { } is the structure mass
matrix for the ith mode. The cumulative mass fraction for the ith mode is given by
Structures Congress 2013 © ASCE 2013 1394

∩ ∑
M= (2)

Where, N is the total number of modes. It is seen that the first two modes have greater
effect on the turbine’s dynamic behavior, whereas the other modes do not have
significant effect. The peak acceleration response at the top of the nacelle was 0.28g
at a viscous damping of 0.86% from the experiments, whereas the value was
calculated to be 0.283g at the damping ratio of 0.86% from the numerical analysis.
Therefore, the viscous damping for the turbine model compares well with the
experiments. Thus the finite element model developed can be used for analysis and
design of the above wind turbine. After proving the validity of the numerical analysis
we will now study the responses of wind turbines of various capacities when
subjected to different earthquake loadings.

First mode Second mode First mode Second mode Third mode
1.7 Hz 11.7-12.3 Hz 1.65 Hz 9.14 Hz 37.65 Hz
(a) Experimental values (b) Computed fundamental modes

Figure 3. Fundamental modes of the 65 kW turbine

FINITE ELEMENT ANALYSIS OF WIND TURBINES

This paper presents the effect of wind turbine size, damping, direction of acceleration
(two horizontal and vertical),and different earthquakes on the response of wind
turbines capable of producing 65 kW, 1 MW, and 5 MW of electric power.

Properties. Two different materials are used in the construction of wind turbines. A
carbon fiber composite where an epoxy is reinforced with carbon fibers is used for
constructing the wind turbine blades. The structural steel is used for the construction
and manufacture of tower, nacelle, and the hub. The nacelle is a hollow steel box that
contains rotor, magnets, electric wiring and other parts for generating electric power.
Hence, its weight is much less than the solid steel box with the same volume. Hence,
an equivalent lower density is used for it in the finite element model. The material
Structures Congress 2013 © ASCE 2013 1395

and geometric properties of the wind turbines of different sizes and capacities are
given in Tables 1 and 2.
Table 1. Material Properties
Properties Blade Composite Material Steel
Density 1760 kg/m3 (3.41 slug /ft3) 7860 kg/m3 (15.25 slug/ft3)
Young’s modulus 235 GPa (34084 ksi) 200 GPa (29000 ksi)
Tensile yield strength 3920 MPa (569 ksi) 250 MPa (36 ksi)
Tensile ultimate
3920 MPa (569 ksi) 460 MPa (66.7 ksi)
strength

Table 2. Geometric Properties


Part 65 kW Turbine 1 MW Turbine 5 MW Turbine
Blade radius (R) 16 m (52.5 ft) 30.31 m (99.4 ft) 63 m (206.7 ft)
Tower length (L) 21.9 m (71.8 ft) 57.19 m (187.6 ft) 88.5 m (290.3 ft)
Hub height (H) 22.6 m (74.1 ft) 61.14 m (200.6 ft) 90 m (295.3 ft)
Outer diameter of
tower at the 2.02 m (6.6 ft) 3.88 m (12.7 ft) 6 m ( 19.7 ft)
bottom (Db)
Outer diameter of
tower at the top 1.06 m (3.5 ft) 2.45 m (8 ft) 3.87 m (12.7 ft)
(Dt)
Tower thickness
5.3 mm (0.21 in.) 18 mm (0.71 in.) 27 mm (1.06 in.)
(t)
2,400 kg (164 240x103 kg
Nacelle mass (mn) 53,700 kg (3680 slug)
slug) (16445 slug)
78,600 kg ( 5386 347,460 kg
Tower mass (mT) 6400 kg (439 slug)
slug) (23,809 slug)
Rotor and hub 42,000 kg (2878 110,000 kg
1900 kg (130 slug)
mass (mR) slug) (7537 slug)
1 slug = 32..17 lbm

Loading. Earthquake loads are applied as uniaxial excitations at the tower base. In all
models X direction is parallel to the rotor axis and Z direction is the vertical direction.
In this paper six input accelerations representing three earthquakes (Chopra 2012)
are:

• East-west and vertical components of the landers earthquake (June 28, 1992) with
the peak ground acceleration (PGA) of 0.15g and 0.17g respectively.
• North-south and vertical components of Imperial Valley earthquake (May 19,
1940) with the PGA of 0.3g and 0.21g respectively.
• North-south and vertical components of Northridge earthquake ( January 17,
1994) with the PGA of 0.34g and 0.55g respectively.

The earthquake accelerations are shown in Figure 4 at time interval of 0.02 seconds.
The input earthquake accelerations are shown in time and frequency domains.
Structures Congress 2013 © ASCE 2013 1396

Finite element modeling. For the wind turbines 3D finite element models are built
consisting of shell elements to form tower and blades, whereas the solid elements are

0.3 Im perial Valley NS 25 Imperial Valley NS


0.2 20

Amplitude
0.1 15
Acc. (g)

0 10
-0.1 0 20 40 60 80 5
-0.2 0
-0.3 0 5 10 15 20 25
-0.4 Tim e (s) Frequency (Hz)

0.2 Imperial Valley Up 9 Imperial Valley Up

Amplitude
0.1 6
Acc. (g)

0
3
0 20 40 60 80
-0.1
0
-0.2
0 5 10 15 20 25
-0.3 Time (s) Frequency (Hz)

0.15 Landers EW 12 Landers EW


0.1
Amplitude 9
0.05
Acc. (g)

6
0
-0.05 0 10 20 30 40 50 3
-0.1 0
-0.15 0 5 10 15 20 25
-0.2 Time (s) Frequency (Hz)

0.15 Landers Up 6 Landers Up


0.1
Amplitude

0.05 4
Acc. (g)

0
2
-0.05 0 10 20 30 40 50
-0.1 0
-0.15 0 5 10 15 20 25
-0.2 Time (s) Frequency (Hz)

0.4 Northridge NS 12 Northridge NS


0.3
Amplitude

9
0.2
6
Acc. (g)

0.1
0 3
-0.1 0 10 20 30 40 0
-0.2 0 5 10 15 20 25
-0.3 Time (s) Frequency (Hz)

0.6 Northridge Up 10 Northridge Up


0.4 8
Amplitude

0.2 6
Acc. (g)

4
0
2
-0.2 0 10 20 30 40
0
-0.4 0 5 10 15 20 25
-0.6 Time (s) Frequency (Hz)

Figure 4. Earthquake input accelerations


Structures Congress 2013 © ASCE 2013 1397

used to mesh the nacelle and the hub. The computer program ANSYS and its SHELL
181 and SOLID186 elements are used in this study.

The finite element model is shown in Figure 2. In this study the towers of the wind
turbines are considered as fixed at the base. Shell 181 is suitable for analyzing thin to
moderately thick shell structures. It is a four node element with six degrees of
freedom at each node, translations in the x, y, and z directions, and rotations about the
x, y, and z axes. If the membrane option is used, the element has translational degrees
of freedom only. The degenerate triangular option is used as filler elements in mesh
generation. The geometry and shape functions associated with the Shell 181are given
in the ANSYS manual. Solid 186 is a higher 3-D twenty node solid element with
three degrees of freedom per node, translations in the nodal x, y, and z directions. This
element is well suited for modeling irregular meshes. The geometry and the shape
functions associated with the Solid 186 element are given in the ANSYS manual.

The element numbers in the models of different turbines are specified by comparing
first fundamental frequencies for different mesh sizes. It is important to keep the node
numbers low enough not to increase the analysis time. Table 3 gives the summary of
mesh sizes for models of the various turbines.

Table 3. Meshing Summary


Parts 65 kW Turbine 1 MW Turbine 5 MW Turbine
Element
Elements Nodes Elements Nodes Elements Nodes
Type
Tower 1760 12430 988 6981 1472 10384 Shell 181
Nacelle 579 2073 593 2232 568 1910 Solid 186
Hub 28 199 44 287 21 180 Solid 186
Blades 278 387 458 649 558 800 Shell 181
Total 2645 14,926 2083 10,007 2619 13,206

NUMERICAL RESULTS

Modal analysis. The modal analysis is performed using Block Lancoz method in
ANSYS. The eigenvector solver uses the Lancoz algorithm. This method is
recommended to find many modes of large models and can handle poorly shaped
solid and shell elements. First 50 modes were found for each wind turbine model that
included 90% of the effective mass. It is seen that the first two modes have the
predominant effect on the turbine’s dynamic behavior, whereas the other modes have
much smaller effect shown in Table 4.

Transient dynamic analysis. The method sometimes also called time history
analysis is used to calculate the structural response to arbitrary time dependent loads.
The basic equation of motion to be solved is written as

{ }+ { }+ { }={ } (3)
Structures Congress 2013 © ASCE 2013 1398

Where, , , and are the mass, damping, and stiffness matrices respectively
of the system under consideration. { }, { }, { }, and { } are the nodal acceleration,
nodal velocity, nodal displacement, and applied load vectors respectively. At any
Table 4. Fundamental Modes of Towers
X direction Y direction Z direction
Ratio of Ratio of Ratio of
effective effective effective
Freq. Freq. Freq.
Tower Mode mass to mass to mass to
(Hz) (Hz) (Hz)
total total total
mass mass mass
1 1.65 0.62 1.65 0.61 31.19 0.73
65 kW 2 9.48 0.15 9.14 0.14 21.84 0.02
3 21.84 0.07 20.02 0.08 9.48 0.00
1 0.43 0.73 0.43 0.73 11.67 0.70
1 MW 2 3.18 0.09 3.36 0.11 9.25 0.10
3 9.25 0.03 7.49 0.05 5.42 0.02
1 0.23 0.71 0.23 0.71 6.15 0.58
5 MW 2 2.20 0.10 1.51 0.11 4.81 0.13
3 4.81 0.02 7.17 0.04 5.49 0.07

given time, these equations can be thought of as a set of ‘static’ equilibrium equations
that also take into account inertia forces { } and damping forces { } . The
ANSYS program uses the Newmark time integration method to solve Eqs. (3) at a
discrete time point. The Newmark method uses finite difference expansions in the
time interval Δt, in which it is assumed that

{ }={ }+ 1− { }+ { }∆ (4)

{ }={ }+{ }∆ + −∝ { }+∝ { } ∆ (5)

Where α, δ are the Newmark integration parameter; { }, { } , and are the nodal
displacement, velocity, and acceleration respectively at time tn. Similarly, { },
{ }, and { } are the nodal displacement, velocity, and acceleration at the time
tn+1. Δt = tn+1 - tn . The governing Eq. (3) is written at time tn+1 to calculate { } as
follows:

{ } + { }+ { }={ } (6)

The { } is calculated by rearranging Eqs. (4 and 5) as follows:

{ }= { }−{ } − { }− { } (7)

{ }={ }+ { }+ { } (8)
Structures Congress 2013 © ASCE 2013 1399

Where = ∝∆ , = ∆
, = ∆
, = −1, = −1,


= −2 , =∆ 1− , and = ∆ .

{ } in Eq. (4) can be substituted in Eq. (5), the equations for { }and { }are
thus expressed in terms of unknown { } and the known displacements { },
velocities { }, and accelerations { } at the time tn. The equations for { } and
{ } are then substituted in Eq.(6) to get

+ + { }={ }+ { }+ { }+ { } +

{ }+ { }+ { } (9)

The displacements{ } are obtained from Eq. (9). Eqs. (7 and 8) are used to update
the velocities and accelerations. The Newmark parameters are related to the input as
follows:

∝= 1+ , and = + (10)

Where, γ is the amplitude decay factor. The solutions of Eq. (6) are stable if γ ≥ 0
(Zienkiewicz 1977).

The time step size of 0.02 seconds is used in the transient analysis because time
period of the 50th mode for the 65 kW turbine is 0.021 seconds. This enables the
model to simulate vibrations with frequencies under 50 Hz, responsible for more than
90% of the effective mass. The comparison of acceleration in the X and Y horizontal
directions at the top of the nacelle for three damping ratios of 0.5%, 1.0%, and 2% of
critical damping is shown in Fig. 5. The naming 1S0.5-LX in the Figure 5 means: the
first letter stands for 1 MW turbine ( 1,2, or 5 MW), the second letter S designates
steel as the material used in the tower, the third letter 0.5 indicates the damping ratio
(0.5, 1, 2 percent), the fourth letter represents the Landers (Landers, Imperial Valley,

0.3 1S0.5-LX
1S0.5-LY
0.2
Acceleration (g)

0.1

0
0 10 20 30 40 50 60
-0.1

-0.2 Time (s)


Structures Congress 2013 © ASCE 2013 1400

0.3 1S1.0-LX
1S1.0-LY
0.2
Acceleration (g)

0.1

0
0 10 20 30 40 50 60
-0.1

-0.2 Time (s)

0.3 1S2.0-LX
1S2.0-LY
0.2
Acceleration (g)

0.1

0
0 10 20 30 40 50 60
-0.1

-0.2 Time (s)

0.2 5S1.0-LX
5S1.0-LY
0.1
Acceleration (g)

0
0 10 20 30 40 50 60
-0.1

-0.2

-0.3 Time (s)

Figure 5. Comparison of acceleration response in x and y directions for different


turbines and damping ratios at the top of the nacelle

or Northridge) earthquake loading, and the last letter gives the direction of input
loading and response (X, Y, or Z). In all analyses, the measured response is in the
direction of the earthquake component.

The comparison of the acceleration response in the Z direction is shown in Figure 6


for three damping ratios. It can be seen that the peak acceleration response values are
very close for X and Y directions for different damping ratios. It is seen that the
Structures Congress 2013 © ASCE 2013 1401

damping ratio has little effect on peak acceleration response in the X and Y
directions, but it influences the response in the Z direction. The reason is that the
fundamental frequency of the turbine in the Z direction is close to the earthquake
frequency, hence the acceleration response factor, Ra, is sensitive to damping. In the
X and Y directions, the turbine frequency is smaller than the load frequency, hence

0.15 1S0.5-LZ

0.10 1S1.0-LZ
1S2.0-LZ
Acceleration (g)

0.05

0.00

-0.05 0 5 10 15 20 25 30 35 40 45 50

-0.10

-0.15
-0.20 Time (s)

Figure 6. Comparison of acceleration response in the Z direction


for different damping ratios

the response is almost independent of damping. The summary of peak acceleration


and deformation responses at the top of the nacelle, and σmax, the maximum von Mises
stress at the base of the tower are given in Table 5.

Table 5. Peak Acceleration and Deformation Responses at the Top of the


Nacelle, and Maximum von Mises Stress at the Tower Base
amax (g) δmax (mm) σmax
Model X Y Z X Y Z X Y Z
0.65S0.5-L 0.35 0.30 - 35.75 32.05 - 51.41 46.98 -
0.65S1.0-L 0.28 0.28 - 28.57 29.18 - 42.35 42.60 -
0.65S2.0-L 0.22 0.28 - 25.03 26.09 - 37.35 38.27 -
0.65S1.0-I - 1.10 - - 72.12 - - 107.15 -
0.65S1.0-N - 0.70 - - 67.29 - - 95.61 -
1S0.5-L 0.21 0.20 0.16 127.76 128.18 0.94 53.26 52.66 4.75
1S1.0-L 0.20 0.20 0.11 118.91 119.43 0.73 47.39 48.09 3.42
1S2.0-L 0.20 0.20 0.07 114.18 114.71 0.50 44.78 45.36 2.32
1S1.0-N - 0.52 - - 234.92 -- - 100.17 -
1S1.0-L0.5 - 0.65 - - 387.31 - - 156.89 -
5S0.5-L 0.19 0.23 0.30 450.38 446.90 2.29 154.12 156.82 10.67
5S1.0-L 0.19 0.22 0.24 409.58 405.99 1.73 135.88 137.71 8.44
5S2.0-L 0.19 0.21 01.8 341.77 338.77 1.32 112.06 113.69 6.89
5S1.0-N - 0.41 - - 330.93 - - 108.28 -
Structures Congress 2013 © ASCE 2013 1402

CONCLUSIONS

1. Finite element model results obtained in this project compared well with the
experimental results obtained from the literature for the 65 kW turbine. Hence,
numerical analysis by the transient dynamics approach is a valid tool for the
analysis of wind turbines.
2. The peak acceleration response at the top of the nacelle does not differ much in
the horizontal directions of X and Y with the variation in the damping ratios for
1 MW and 5 MW turbines for the same earthquake. There was some difference
in the peak acceleration response of the 65 kW wind turbine in the X and Y
directions when the damping ratios were varied.
3. The peak accelerations responses in the Y direction are larger than in the X
direction, Y direction being parallel to the rotor axis.
4. The damping ratios have definite effect on the peak acceleration responses in the
Z, the vertical direction. The peak acceleration responses reduce in the Z
direction as the damping ratios are increased.
5. There can be different behavior for different earthquakes because of frequency
effect. The peak ground acceleration (PGA) of the Northridge earthquake is 10%
larger than the PGA of the Imperial Valley earthquake, whereas the response is
more critical when the wind turbines are subjected to the Imperial Valley
earthquake.

REFERENCES

ANSYS Reference Documentations, ANSYS, Inc., 2007.


Chopra, A. K. (2012) Dynamics of Structures – Theory and Applications to
Earthquake Engineering, Prentice Hall, New York.
Clough, R. W., and Penzien, J. (1993) Dynamics of Structures, McGraw-Hill, New
York.
DOE (2008) Annual Report on U.S. Wind power Installation, Costs, and Performance
Trends: 2007, Department of Energy, Washington, D.C.
Hodges, D. H., and Pierce, G. A. (2011) Introduction to Structural Dynamics and
Aero-elasticity, Cambridge University Press, Cambridge, U.K.
Prowell, I., Veletzos, M., and Elgamal, A. (2009) Journal of Earthquake Engineering,
13, 1172-1190.
Zienkiewicz, O. C. (1977 The Finite Element Method, McGraw-Hill, New York.

Vous aimerez peut-être aussi