Você está na página 1de 10

Dynamic analysis of wind turbines including soil-structure interaction

M. Harte
a
, B. Basu
a,
, S.R.K. Nielsen
b
a
Department of Civil, Structural & Environmental Engineering, Trinity College Dublin, Ireland
b
Department of Civil Engineering, Aalborg University, Denmark
a r t i c l e i n f o
Article history:
Received 20 March 2012
Revised 1 June 2012
Accepted 27 June 2012
Available online 18 August 2012
Keywords:
Soil-structure interaction
Wind turbines
Cone model
Blade element momentum theory
Kaimal spectrum
a b s t r a c t
This paper investigates the along-wind forced vibration response of an onshore wind turbine. The study
includes the dynamic interaction effects between the foundation and the underlying soil, as softer soils
can inuence the dynamic response of wind turbines. A Multi-Degree-of-Freedom (MDOF) horizontal
axes onshore wind turbine model is developed for dynamic analysis using an EulerLagrangian approach.
The model is comprised of a rotor blade system, a nacelle and a exible tower connected to a foundation
system using a substructuring approach. The rotor blade system consists of three rotating blades and
includes the effect of centrifugal stiffening due to rotation. The foundation of the structure is modeled
as a rigid gravity based foundation with two DOF whose movement is related to the surrounding soil
by means of complex impedance functions generated using cone model. Transfer functions for displace-
ment of the turbine system are obtained and the modal frequencies of the combined turbine-foundation
system are estimated. Simulations are presented for the MDOF turbine structure subjected to wind load-
ing for different soil stiffness conditions. Steady state and turbulent wind loading, developed using blade
element momentum theory and the Kaimal spectrum, have been considered. Soil stiffness and damping
properties acquired from DNV/Ris standards are used as a comparison. The soil-structure interaction is
shown to affect the response of the wind turbine. This is examined in terms of the turbine structural dis-
placement and also the base shear and bending moment in the tower and the foundation. The frequency
domain response of the bending moment and shear force in the foundation and the tower base, for stiffer
soil condition, is shown to be characterized by peaks at multiples of the wind turbine rotational speed (3P
effects). The effect of dynamic soil-structure interaction on the rotation of the foundation has also been
investigated.
2012 Elsevier Ltd. All rights reserved.
1. Introduction
It is widely recognized that the dynamic response of a structure
on soft soil may be different from the response of a similarly ex-
cited structure supported on rm soil [1]. This nding has been
conrmed in numerous studies on the effects of soil-structure
interaction (SSI), a full review of the development of SSI can be
found in [2].
Luco [3] in the study on the seismic response of tall chimneys
showed that SSI had an effect only for softer soils (shear wave
velocity lower than 750 m/s) and could lead to reductions or in-
creases in response, depending on the characteristics of the chim-
ney and the seismic excitation. This has been conrmed in a recent
study by Moghaddasi et al. [4] in which a Monte Carlo simulation
was carried out for a range of SDOF structures and soil conditions
excited by a series of seismic excitations.
The use of SSI in design is usually restricted to buildings in seis-
mic zones, however wind turbines contain moving parts and must
sustain continuous vibration-induced forces throughout their
operational life. Novak and Hifnawy [5,6] showed that the re-
sponse of a structure when subject to a dynamic wind loading
can be affected by SSI. The literature to date on the dynamic inter-
action of wind turbines taking into account SSI has been somewhat
sparse.
Wind turbines have increased tremendously in both size and
performance in recent years. In order to reduce the costs, the over-
all weight of components is minimized; this results in a more ex-
ible wind turbine sensitive to dynamic excitation even at low
frequencies [7]. Modern wind turbines are typically installed with
variable speed systems, so rotational speed of the rotor varies from
1020 rpm. Therefore the rst excitation frequency interval is
around 0.170.33 Hz and is referred to as the 1P frequency inter-
val. Since the rst resonance frequency of a modern three bladed
wind turbine is often placed between 1P and 3P, it is important
to be able to evaluate the resonance frequencies of the wind tur-
bine structure accurately [8].
0141-0296/$ - see front matter 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.engstruct.2012.06.041

Corresponding author.
E-mail addresses: hartemi@tcd.ie (M. Harte), basub@tcd.ie (B. Basu), srkn@civil.
aau.dk (S.R.K. Nielsen).
Engineering Structures 45 (2012) 509518
Contents lists available at SciVerse ScienceDirect
Engineering Structures
j our nal homepage: www. el sevi er. com/ l ocat e/ engst r uct
Veletsos and Verbic [9] showed the presence of exible soil
underneath the foundation of a structure increases the damping
capacity of the foundation and reduces the structures natural fre-
quency. The same was also observed in a report by Stewart et al.
[10], that SSI usually has the effect of lengthening the vibration
periods of structures, which may be of concern in wind engineering
due to the localization of wind-induced spectral energy at low
frequencies.
Currently wind turbine foundations are modeled simply by
beam elements or static springs which means that the foundation
stiffness is frequency independent. These idealized assumptions of
xity at the base of the tower are conservative and could poten-
tially lead to overestimation of stiffness and thus the systems nat-
ural frequency. Consequently, unless the separation between the
operational and natural frequencies is large, these assumptions of
xity should not be used and SSI needs to be considered in design
[11]. Therefore it is key to know the structures overall natural fre-
quency to allow sufcient separation of the structural systems
natural frequency from the turbine operational frequencies, to
avoid potentially catastrophic failures [12].
The soil-structure model adopted in this study is composed of a
multi-degree-of-freedom (MDOF) wind turbine model located on
an embedded rigid gravity based foundation with two DOF, sized
using DNV/Ris design practice [13]. The foundation is assumed to
be resting on a homogeneous linear elastic half-space. The idealized
soil-foundation-structure is modeled using cone model. The origi-
nal cone model was developed by Ehlers [14] to represent a surface
disk under translational motions. An improved model, developed
by Wolf and Deeks [15], is used here to generate the complex
impedance functions. The impedance functions for the embedded
foundation was used as boundary conditions for the foundation-soil
domain in a complete MDOF wind turbine model.
A wind turbine model composed of exible elements is devel-
oped deriving the equations of motion from the EulerLagrange
stationary conditions of the action integral in Hamiltonian
mechanics. The horizontal axis wind turbine model is comprised
of a rotor blade system, a nacelle and a exible tower connected
to a rigid gravity based foundation using a substructuring ap-
proach. The rotor blade system consists of three rotating blades
and includes the effects of centrifugal stiffening due to rotation.
Only vibrations in the apwise directions are considered and the
local deformations of each blade is modeled by a truncated modal
expansion. Model parameters for the wind turbine are taken from
the NREL [16] for their standard 1.5MW wind turbine. The model is
exposed to both a steady state and a random wind loading excita-
tion, generated using the Kaimal spectrum [17] to model a homo-
geneous isotropic turbulence eld. The wind load has been
developed in accordance with the Beam Element Momentum The-
ory [18] and therefore depends on the blade geometry, rotational
speed and wind speed.
Simulations are presented for the MDOF wind turbine model
subjected to wind loading resting on a layered soil prole for range
of soil stiffness values. Transfer functions for displacement of the
nacelle and foundation are obtained and the modal frequencies of
the systemestimated. Fully xed conditions are used as a reference
and soil modeling parameters from DNV/Ris standards are used to
validate cone model for a simple uniform soil prole. The response
in terms of base shear, bending moments, structural displacement
and rotation of foundation are examined with regard to SSI.
2. MDOF wind turbine structure
Using a sub-structure methodology, the structure and the soil
stratum can be modeled separately and then combined. The MDOF
superstructure model is formulated using the EulerLagrange
equations, as expressed below
d
dt
dT
d
_
q
i
_ _

dT
dq
i

dV
dq
i
F
i
1
where T is the kinetic energy and V is the potential energy of the
conservative forces in the system, q
i
is the displacement and F
i
is
the generalized loading for degree of freedom i. Fig. 1 shows the
model along with all the generalized degrees of freedom.
The model was developed for horizontal motion corresponding
to apwise (longitudinal) blade vibrations, apwise vibrations oc-
cur out of the rotor plane. Thus the nacelle is free to translate in
this out-of-plane direction. The foundation, resting on exible soil,
was modeled by the introduction of two DOF allowing it to trans-
late and rotate.
The absolute motion of the rotor blade number i can be de-
scribed by a truncated modal expansion,
u
b;i
t; z

N
n1
q
n
b;i
tU
n
z q
nac
t q
F
t hh
F
t 2
where q
n
b;i
is the modal blade co-ordinates describing the motion of
blade i for the n
th
mode relative to the hub. The notation N indicates
the number of modes, U
n
(z) is the mode shape function and z is a
position vector along the blade. The term q
nac
represents the dis-
placement of the nacelle relative to the foundation, q
F
and h
F
are
the horizontal translation and rotation of the foundation.
The kinetic and potential energies of the model can be written
as,
T
1
2

3
i1
_
R
0
m
b
zv
b;i

2
dz
1
2
M
nac

_
q
nac

_
q
F
h
_
h
F

1
2
M
F
_
q
2
F

1
2
I
F
_
h
2
F
3
V
1
2

3
i1
_
R
0
EIz
d
2
u
b;i
dz
2
_ _
2
dz V
cs ;i
_
_
_
_

1
2
K
nac
q
2
nac
4
Fig. 1. Wind turbine model.
510 M. Harte et al. / Engineering Structures 45 (2012) 509518
where,
d
2
u
b;i
dz
2
_ _

N
n1
U
00
n
zq
n
b;i
t 5
In Eqs. (3) and (4), m
b
(z) is the mass per unit length of the entire
blade-root-hub assembly, R is the length of the blade, EI(z) is the
exural rigidity of the rotor blades, h is the tower height, v
b,i
is
the velocity of the rotor blade and u
b;i
is the relative motion of
the blade. The terms M
nac
and M
F
are the total mass of the nacelle
and the foundation respectively. The term K
nac
is the stiffness of
the nacelle and I
F
is the mass moment of inertias of the foundation
about the tilt axes of rotation. The overdots represent differentia-
tion with respect to time and the primes represent differentiation
with respect to position.
For accuracy, a centrifugal stiffening term, V
cs
, was added into
the blade stiffness matrix, resulting in the stiffness of the blades
increasing with the rotational speed, X
b
. Considering innitesimal
elements along the blade of length dz and integrating over the
length, the tensile force in the blade due to centrifugal body forces
on blade i, acting at the point f is expressed as,
F
ci
X
2
b
_
R
f
m
b
zzdz 6
The resulting potential energy due to centrifugal stiffening of blade i
is given by
V
cs ;i

1
2
X
2
b
_
R
0

N
n1
q
n
b;i
dU
n
dz
_ _
2
_ _
_
R
f
m
b
zzdz
_ _
7
The effect of gravity on the vibration of the blades in the apwise
direction is negligible and therefore not included in the study.
The reaction forces due to the foundation-soil interaction H(t)
and M(t) are non-conservative and cannot be included in Eq. (4),
as a potential energy does not exist. The effect of the reaction
forces is therefore introduced in the formulation as non-conserva-
tive forces in the time domain and can be written in terms of the
following convolution integral,
Ht
Mt
_ _

_
t
1
kt s
q
F
t
h
F
t
_ _
ds 8
where the impulse matrix function k(t) represents the memory af-
fect and the coupling between the horizontal translation and the
rotational DOF of the foundation. Eqs. (3) and (4) can be substituted
back into Eq. (1) and the equations of motion derived for apwise
motion, expressed in standard form as,
Mf

qtg Cf
_
qtg Kfqtg fFtg 9
where [M], [C] and [K] are the mass, damping and stiffness matrices
of the system respectively, the stiffness matrix includes stiffness
components from the three blades and the nacelle. The terms
f qtg; f_ qtg and {q(t)} are the acceleration, velocity and displace-
ment vectors and {F(t)} is the generalized loading derived in a fol-
lowing section. The Lagrangian formulation does not account for
damping, thus structural and aeroelastic damping are added
separately.
The effect of the dynamic interaction with the foundation given
in Eq. (8) will be accounted for in the analysis by using the coupled
horizontal translation and rotation impedance matrix. This will be
achieved by carrying out the analysis in frequency domain (i.e. by
Fourier transforming the equations of motion and incorporating
the frequency dependent impedance matrix). The impedance ma-
trix is obtained by a Fourier transform of Eq. (8). The response in
the time domain will then be obtained by applying an inverse
Fourier transform.
3. Impedance functions
The foundation of the wind turbine is modeled as an embedded,
rigid circular foundation resting on homogeneous elastic half space
whose movement is related to the surrounding soil by means of
complex impedance functions obtained from cone model. Rigorous
procedures for calculating foundation dynamic stiffness exist,
including nite element and boundary elements methods. How-
ever, these methods require signicant computational time and
experience, while cone model [15] is simple to use and provides
physical insight with conceptual clarity, all within acceptable engi-
neering accuracy. The only approximations made are that of the 1-
dimensional strength of materials based on wave propagation in
cones. The use of cone models does lead to some loss of precision
compared to rigorous methods based on 3-D elastodynamics. How-
ever the accuracy of any analysis is limited anyway because of the
many uncertainties some of which can never be eliminated (soil
properties).
In the frequency domain, the dynamic behavior of the founda-
tion can be described by the dynamic-impedance matrix,
K
i;j
x K
0
i;j
k
i;j
a
o
ia
o
c
i;j
a
o
10
where a
o
is the dimensionless frequency coefcient, K
0
i;j
is the static
stiffness component, k
i,j
(a
o
) is the dimensionless spring coefcient
which governs the force that is in phase or counter phase with
the displacement and c
i,j
(a
o
) is the dimensionless damping or dash-
pot coefcient that describes the force that is 90
0
out of phase.
In cone model the effective damping is a summation of two
sources: radiation damping (transmitted by the structure to the
soil) and hysteretic (material) damping of the soil. For each degree
of freedom, at frequency x, the force amplitude can be related to
the displacement amplitude through the equation (Fourier trans-
form of Eq. (8)),

Hx

Mx
_ _
K
i;j
x
^
q
F
x
^
h
F
x
_ _
11
where,
K
i;j
x
K
HH
x K
HM
x
K
MH
x K
MM
x
_ _
12
The terms K
HH
and K
MM
are the horizontal translational and rota-
tional impedance functions, K
HM
and K
MH
are the off diagonal cou-
pling terms, such that the impedance matrix satises
K
i;j
x K
T
i;j
x 13
and
K
i;j
x K

i;j
x
_ _
14
where the superscripts T and represent the matrix operations
transpose and conjugate respectively.
The vertical vibration of embedded footings is not signicant for
wind-induced structural responses and therefore are not consid-
ered in this paper.
4. Bending moment and shear forces
The shear force and bending moment in the foundation and in
the base of the wind turbine tower including SSI effects can be
evaluated by isolating the foundation [19]. Once the response
quantities ^ q
F
x and
^
h
F
x are solved for in the frequency domain,
the Fourier transformed shear force and bending moment in the
foundation may be evaluated by solving Eq. (11). Using these, the
M. Harte et al. / Engineering Structures 45 (2012) 509518 511
Fourier transformed shear force and bending moment in the wind
turbine tower (at the base) can be given as,

H
T
x x
2
m
F
^
q
F
x

Hx 15
and

M
T
x x
2
I
F
^
h
F
x

Mx 16
Solving and inverting back into the time domain the time history re-
sponse of shear force and bending moment can be evaluated.
5. Aerodynamic load calculation
Modal wind drag force time-histories are simulated in accor-
dance with the classical Blade Element Momentum (BEM) theory
which couples the momentum theory with local events taking
place at the actual blades. BEM is currently the most popular tool
for the determination of aerodynamic loads on a rotating rotor
[20], as satisfactory results can be obtained given good aerofoil
data. Here a corrected BEM model, accounting for Glauert correc-
tion and Prandtls tip loss factor, is used. The dynamic loads are cal-
culated by a quasi-static aerodynamic assumption so changes in
the effective angle of attack a are instantly felt in the aerodynamic
loads. This means that the time scale for adjustment of the non-
stationary ow is assumed to be small compared to the fundamen-
tal eigen-period of the blade. With the BEM method it is possible to
calculate steady wind loads and thus the thrust and power with re-
spect to wind speed, rotational speed, pitch angle, number of
blades and actual geometry of the aerofoil (in terms of blade pro-
le, twist and chord distribution). The wind loads are calculated
following an approach given by Hansen [18], which assumes all
sections are independent along the rotor, so that the aerofoil can
be divided into several elements and ow at each element calcu-
lated separately.
Due to changes in the mean wind velocity with height, as the
blade rotates the mean velocity component on the blades experi-
ences a sinusoidal variation in magnitude, the frequency of this
sinusoid is equal to the rotational frequency. Therefore the instan-
taneous wind speed may be expressed as
V
i
t vH Dv
r
R
_ _
cosX
b
t h
i
17
where H is the nacelle height, vH is the mean wind speed at hub
height, r is the position along the blade and Dv is the change in
wind speed between the hub and the top of the blade in an upright
position. The scaling factor (r/R) is used to calculate the required
amplitude at each point along the blade in order to represent the
sinusoidally varying wind velocity above and below the hub. The
term t is the time and h
i
is the phase difference between the tur-
bines blades. The relative wind speed, with respect to each blade
element for each blade, V
r,i
(r,t), is given as,
V
r;i
r; t

V
i
t1 a v
w
t
2
X
b
r1 a
0

2
_
18
where a and a
0
are the axial and tangential induction factors which
are calculated by means of the classical BEM method with correc-
tions. The instantaneous wind speed apparent at the blade, V
i
(t)(1 a), incorporates the mean wind speed which varies sinusoi-
dal due to the effect of the wind shear and blade rotation. Turbu-
lence is included only in the normal direction.
The wind loading on any structural member maybe decom-
posed into a quasi-static mean wind velocity V
i
(t), and a uctuating
turbulent component v
w
(t). The generation of this uctuating tur-
bulent component v
w
(t) on the rotor plane, is the basis of any aero-
dynamic simulation and may be obtained through the use of a
power spectral density function (PSDF). Nodal uctuating velocity
time-histories can be simulated by virtue of the fact that an
arbitrary time-history, with zero mean, may be represented by a
DFT with a discretized version of a continuous frequency content
as,
v
w
t

1
k1
a
k
cosx
k
t

1
k1
b
k
sinx
k
t 19
where a
k
and b
k
are the Fourier coefcients, x
k
is the discredited
circular frequency and t is the time instant. This uctuating velocity
time-history is generated in conjunction with a wind velocity PSDF.
Time-histories were simulated using Eq. (20), a modied version of
the spectrum offered by Kaimal et al. [17], expressed as,
S
v
H; x
r
2
v

100c
3x1 50c
5=3
20
where S
v
(H,x) is the one sided PSDF of the uctuating wind velocity
as a function of the hub elevation and circular frequency (x), r
v
2
is
the variance (related to the turbulence intensity) and c is known as
the Monin coordinate. The latter term may be obtained from the
expressions,
c
xH
2p vH
21
vH
1
k
v

ln
H
z
o
22
where k is Von Karmans constant (typically around 0.4), z
o
is the
roughness length and v

is the friction velocity. The above formula-


tion was used to generate an isotropic, homogeneous turbulence at
the hub height to represent the turbulence across the rotor eld.
Due to the rotation of the blades the turbulence spectra will be
non-homogeneous in nature leading to a rotation sampled spectra.
However the focus of this paper is the effect of SSI in wind turbines
and thus the assumed simplied turbulence eld is adequate.
The wind load is taken only in the normal/apping direction as
only apwise vibrations are considered. The normal force per unit
length of the blade [18] can be expressed as,
p
N;i

1
2
pV
2
r;i
r; tC
N
acr 23
where p is the density of air, c(r) is the chord length, C
N
(a) is the
normal coefcient which is dependent on the lift and drag coef-
cients (which are calculated from tables based on the blade aerofoil
properties and angle of attack). Integrating over the entire rotor
blade length R, the generalized wind load on blade i can be given by,
F
i
t
_
R
0
p
N;i
x; t/
n
xdx 24
6. Numerical examples
Some numerical examples are now presented in order to illus-
trate the effects of SSI on the response of the MDOF wind turbine
model. Transfer functions for the system are obtained and dis-
cussed, as this type of analysis is essential in understanding the re-
sponse of the structure due to more complex random excitations.
Aerodynamic wind load, developed form BEM theory with turbu-
lence generated using the Kaimal spectrum, is then applied to
the structure and the response computed. The analysis is carried
out for different soil stiffness conditions, while a fully xed condi-
tion is used as a reference. Soil damping and stiffness properties
obtained from the DNV/Ris design standards are used as a com-
parison to validate the cone model coefcients.
512 M. Harte et al. / Engineering Structures 45 (2012) 509518
6.1. Model properties
Model properties for the blades, nacelle and tower were taken
from an NREL 1.5MW 3-bladed wind turbine. Considering (in Eq.
(2)) the rst mode of blade vibration only, the equations of motion
were developed for the coupled foundation/nacelle/blade 6 DOF
wind turbine model. The foundation is assumed to be embedded
in rigid soil and designed for a constant vertical gravity load and
the maximum over turning moment experienced by the turbine
during normal operation at near cut-out wind speed. The size of
the foundation is designed based on the bearing capacity of the soil
medium. The vertical bearing capacity is calculated using a con-
stant soil density and friction angle. Details of the foundation
and the soil prole are given in the Table 1 and Fig. 2.
Two soil proles are modeled for a range of shear wave velocity
values. The soil loss factor (soil material damping) is assumed to be
constant for all frequencies i.e. hysteretic damping is assumed. Site
A is chosen due to the fact that it can be modeled by analytically
derived formulas given by DNV/Ris standards in the form of static
springs (stiffness) and damping coefcients, and therefore can be
used as a comparison to validate the frequency dependent stiffness
and damping coefcient given by cone model. Site B is chosen as a
more realistic soil prole (one with multiple layers) and is modeled
solely by cone model.
The wind load is generated in accordance with the parameters
shown in Table 2, note also that as BEM theory is applied blade
geometry and characteristics also inuence the wind load.
The Lagrangian formulation does not account for damping.
Modal damping was assumed for the structural damping in the
tower and in the blades. The modal damping ratio was taken as
5% and 20% for tower bending and ap blade bending (rst mode)
respectively. A high damping ratio for the blade bending mode was
assumed due to the known large effects of aerodynamic damping
in this mode [20].
6.2. Response calculation
The dynamic model must be solved in the frequency domain
due to the presence of complex frequency dependent impedance
functions in the equation of motions. Transforming Eq. (9) to the
frequency domain and incorporating the foundation impedance
matrix due to SSI gives,
x
2
Mf
^
qxg ixCf
^
qxg Ksxf
^
qxg f

Fxg 25
where Ks(x) is the frequency dependent complex stiffness matrix
(including the foundation impedance matrix). Rearranging Eq.
(25) the input-output relationship is given by,
f
^
qxg Hx:f

Fxg 26
The transfer matrix can be expanded as,
Hx x
2
M Ksx ixC
1
27
For numerical calculations using discrete frequencies the transfer
matrix is given as,
Hx
j

x
2
j
M Ksx
j
ix
j
C
1
for j 6 N=2
H

x
Nj
for j > N=2
_
28
where N is the number of sample points taken and x
j
(=jDx) is the
discrete circular frequency at each step, noting Dx= 2p/T
p
. An in-
verse Fourier transform can be employed to transform the response
f^ qxg back to the time domain, yielding the complete time history
response {q(t)}. For solving the problem, the time span must be se-
lected so that period T
p
is considerably longer than the duration of
excitation [21], thus resulting in an interval of zero excitation be-
fore and following the wind excitation load. This is a necessary
requirement so that the free vibration response during the intervals
of zero excitation will dampen out almost completely. Otherwise,
the assumed zero initial conditions at the start of the excitation will
not be sufciently satised.
6.3. Transfer functions
Transfer functions of different response quantities are deter-
mined for a series of soil stiffness conditions, a fully xed condition
is used as a reference. The transfer functions were obtained by con-
sidering the output response of the relevant DOF. To obtain the
transfer functions with respect to the input at the nacelle, the Fou-
rier spectrum of the loads applied to the different DOF were nor-
malized by the Fourier spectrum amplitude of the load at the
nacelle.
The fundamental modal frequency of the coupled wind turbine
system was determined by examination of the displacement trans-
fer function, as resonance of a structure when subjected to a har-
monic load can be observed as local peaks in the magnitude of
the nodal displacements [8]. Thus, fundamental modal frequencies
as a function of soil shear wave velocity were generated for site A.
This was done modeling the soil using DNV standards and cone
model, the results of which are plotted in Fig. 3.
Fig. 3 validates the cone model for the uniform soil prole. With
validation established cone model is used to model the more com-
plex layered site B prole, transfer functions for which are pre-
sented in Fig. 4. Only the low frequency range of the transfer
functions for Site B are shown as most of the energy is concen-
trated in this region. The rst peak range is between 13 rad/s
for the given soil conditions and contains the fundamental modal
frequency of the system. From Fig. 4 it is evident that as the shear
wave velocity of the soil decreases the fundamental modal fre-
quency of the system drops, while for higher values the fundamen-
tal modal frequency of the system tends towards a fully xed
structural response. Therefore SSI has the effect of lengthening
the vibration period of the wind turbine, a fact shown before in
other SSI studies [1,22]. When the ground consists of layered stra-
ta, as in Site B, the waves transmitted from the structure are re-
ected at the layer interfaces and are returned to the structure.
As a result the radiation damping normally decreases compared
to uniform site (Site A) with similar ground stiffness, however as
Site Bs underlying layers have a lower stiffness than the top layer
(Fig. 2) this effect cannot be seen here.
SSI generally has the effect of increasing damping in the system
as energy is dissipated by radiation into the soil and hence should
reduce the relative displacement response. However the relative
displacement of the nacelle response shows only a slight reduction
in peak response as shear wave velocity drops, this can be seen in
the transfer function Fig. 4a. This is due to the fact the wind turbine
has a relatively large foundation compared to the vertical load, as
the foundation is sized based on the large over turning moment.
Table 1
Foundation and soil properties.
Properties Values
Foundation radius (R) 5.1 m
Foundation embedment depth (D) 3 m
Depth to bedrock (H) 20 m
Soil loss factor (n
s
) 0.05
Shear wave velocity (V
s
) range 40500 m/s
Poissons ratio (v
s
) 0.3
Mass density (p
s
) 1750 kg/m
3
Friction angle (/
0
) 35
M. Harte et al. / Engineering Structures 45 (2012) 509518 513
Total nacelle displacement includes horizontal foundation dis-
placement, lateral displacement due to foundation rotation and
relative displacement of the nacelle. Fig. 4b shows the total dis-
placement of the nacelle increases as shear wave velocity drops,
especially for lower shear wave velocity values. This detrimental
effect of SSI has been shown before in previous studies for tall
slender structures [3,4] and is caused due to the rocking of the
foundation which for tall structures increases the acceleration of
the mass and the associated inertia force, this whipping effect
can lead to a corresponding increase in response [1]. However total
displacements compared to relative displacements do not have the
same effect on stresses and deformations within the superstruc-
ture [4] and therefore are not as critical. Total displacements are
still an important design criteria as the motion of the nacelle must
be limited for obvious reasons.
To avoid resonance the fundamental modal frequency of the
wind turbine system should not coincide with the rotor speed
range (1P) and blade passing speed range (3P for three blade tur-
bines). The 1P and 3P areas are plotted in Fig. 5 to visualize the
zones in which the structural natural frequency should not lie to
prevent dynamic interaction [23]. In Fig. 5 P
M
and P
R
represent
the rotor speed at cut in and rated conditions respectively for the
variable speed 1.5MW wind turbine. Overlaid upon this is the
range of wind turbine fundamental modal frequencies as a func-
tion of shear wave velocity, V
s
= 40500 m/s. The wind turbine
should be designed such that the fundamental modal period lies
in between the 1P and 3P range, and for stiff soil conditions this
is shown to be true. However for soft soil the fundamental modal
period slips down into the 1P range which may lead to undesirable
dynamic interaction.
6.4. Wind excitation response
To illustrate the effects of SSI on the wind turbine system some
time histories for the displacement response of the blades, nacelle
and foundation are shown in Fig. 6. These were generated under a
turbulent wind excitation for rated conditions of rotation and wind
speed.
The nacelle relative displacement is shown in Fig. 6a and illus-
trates the neutral effect of SSI on the relative vibration response.
Fig. 6b shows the total displacement response of the nacelle, and
as expected there is a large increase in response as the shear wave
velocity of the soil drops, caused by the rotation of the foundation
increasing the displacement of the nacelle. Fig. 6c shows the rela-
tive displacement of a blade. The effect of the exible soil on the
blade relative displacement is negligible which is to be expected
as the interaction with the foundation is dampened out (ltered)
through the superstructure. The foundation rotation is shown in
Fig. 6d for a soil with a shear wave velocity V
s
= 100 m/s. An aver-
age rotation of around 0.0014 rad is observed, for a 82 m hub
height and a foundation depth of 3 m. This corresponds to a hori-
zontal translation of around 119 mm at the nacelle.
Fig. 2. Soil proles.
Table 2
Turbine properties and operational data.
Properties Values
Minimum rotor speed 9.5 rpm
Rated rotational speed (X
b
) 20 rpm
Blade length 35 m
Blade natural frequency 8.23 rad/s
Tower height 82m
Tower natural frequency 2.58 rad/s
Rated wind speed, at hub ^ vH 12 m/s
Air density (p
w
) 1.225 kg/m
3
Roughness (z
o
) 0.005
Turbulence intensity 10%
0 50 100 150 200 250 300 350 400 450 500
1.4
1.6
1.8
2
2.2
2.4
2.6
2.8
Shear wave velocity (m/s)
F
u
n
d
a
m
e
n
t
a
l

m
o
d
a
l

f
r
e
q
u
e
n
c
y

(
r
a
d
/
s
)
Cone model
DNV standards
Fig. 3. Structural fundamental modal frequency versus shear wave velocity, Site A.
514 M. Harte et al. / Engineering Structures 45 (2012) 509518
To examine the effects of SSI on the base shear and bending
moment in the wind turbine, the model was exposed to a steady
wind load for rated conditions of rotational and wind speed. Re-
sults are presented in the frequency domain and are shown in
Fig. 7. It is observed from Fig. 7 that the major peaks occur around
13 rad/s and at around 3P depending on soil stiffness. Other
peaks occur roughly at multiples of 3P. These peaks are dampened
out for soft soil conditions and are only prominent in stiffer soil
conditions. The peaks in the range 13 rad/s are prevalent for soft-
er soils and are affected by SSI, as the shear wave velocity de-
creases there is a corresponding drop in frequency at which the
peaks occurs.
Sample time histories for base shear and bending moment are
presented in Fig. 8. The time histories show that although the peak
shear/moment are very similar regardless of soil stiffness, the fre-
quencies of vibration present are highly dependent on soil stiff-
ness. For a shear wave velocity of 70 m/s compared to 500 m/s
the frequency of a peak occurrence is quartered. This will have a
signicant effect on the fatigue analysis of the tower.
The rotation of a wind turbine foundation must be minimized in
order to prevent failure and other serviceability issues concerning
wind turbine operations. The maximum rotation yielded from the
proposed model for different soil stiffness conditions is shown in
Fig. 9. DNV/Ris standards specify, for serviceability limit state,
the maximum allowable tilt of the foundation as 0.0087 rad off
the vertical. It can be seen from Fig. 9 that at very low soil stiffness
this condition is violated and a maximum tilt of 0.015 rad off the
vertical is reached.
0 2 4 6 8
10
8
10
7
10
6
10
5
10
4
Frequency (rad/s)
D
i
s
p
l
a
c
e
m
e
n
t

t
r
a
n
s
f
e
r

f
u
n
c
t
i
o
n
(a)
0 2 4 6 8
10
12
10
10
10
8
10
6
Frequency (rad/s)
D
i
s
p
l
a
c
e
m
e
n
t

t
r
a
n
s
f
e
r

f
u
n
c
t
i
o
n
(c)
0 2 4 6 8
10
12
10
10
10
8
10
6
Frequency (rad/s)
R
o
t
a
t
i
o
n
a
l

t
r
a
n
s
f
e
r

f
u
n
c
t
i
o
n
(d)
Vs = 40 m/s
Vs = 100 m/s
Vs = 500 m/s
0 2 4 6 8
10
8
10
7
10
6
10
5
10
4
Frequency (rad/s)
D
i
s
p
l
a
c
e
m
e
n
t

t
r
a
n
s
f
e
r

f
u
n
c
t
i
o
n
(b)
Vs = 40 m/s
Vs = 100 m/s
Vs = inf
Vs = 40 m/s
Vs = 100 m/s
Vs = inf
Vs = 40 m/s
Vs = 100 m/s
Vs = 500 m/s
Fig. 4. Transfer functions: (a) Nacelle relative displacement, (b) Nacelle total displacement, (c) foundation horizontal translation and (d) foundation rotation.
Fig. 5. Frequency range of 1P and 3P, with structural fundamental modal frequency range due to SSI.
M. Harte et al. / Engineering Structures 45 (2012) 509518 515
7. Conclusions
An along-wind forced vibration analysis of an onshore wind tur-
bine has been carried out including the effects of soil exibility
underneath the foundation. A MDOF wind turbine model was
developed using an EulerLagrangian approach. The soil-founda-
tion interaction was modeled by complex impedance functions
generated using cone model and included in the overall model
using a substructuring approach. The equations of motion were
solved in the frequency domain due to the frequency dependent
impedance matrix.
Two soil proles were examined in this study, a uniform prole
used to validate the cone model by comparison with the DNV/Ris
standards and a more complex soil prole with multiple soil layers
0 50 100 150
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
(b)
Time (s)
D
i
s
p
l
a
c
e
m
e
n
t

(
m
)
Vs = 40 m/s
Vs = 100 m/s
Vs = 500 m/s
0 50 100 150
0.26
0.28
0.3
0.32
0.34
0.36
0.38
0.4
(a)
Time (s)
D
i
s
p
l
a
c
e
m
e
n
t

(
m
)
0 50 100 150
1.5
1.6
1.7
1.8
1.9
(c)
Time (s)
D
i
s
p
l
a
c
e
m
e
n
t

(
m
)
0 50 100 150
1.1
1.2
1.3
1.4
1.5
1.6
1.7
x 10
3
(d)
Time (s)
R
o
t
a
t
i
o
n

(
r
a
d
s
)
Vs = 100 m/s
Vs = 40 m/s
Vs = 500 m/s
Vs = 100 m/s
Fig. 6. Response time-histories: (a) Nacelle relative displacement response, (b) Nacelle total displacement response, (c) blade relative displacement response and (d)
foundation rotation response.
0 5 10 15 20
10
4
10
6
10
8
Frequency (rad/s)
T
r
a
n
s
f
e
r

f
u
n
c
t
i
o
n

a
m
p
l
i
t
u
d
e

(a)
0 5 10 15 20
10
0
10
2
10
4
10
6
Frequency (rad/s)
T
r
a
n
s
f
e
r

f
u
n
c
t
i
o
n

a
m
p
l
i
t
u
d
e

(b)
Vs = 40 m/s
Vs = 100 m/s
Vs = 500 m/s
0 5 10 15 20
10
4
10
6
10
8
Frequency (rad/s)
T
r
a
n
s
f
e
r

f
u
n
c
t
i
o
n

a
m
p
l
i
t
u
d
e

(c)
0 5 10 15 20
10
0
10
2
10
4
10
6
Frequency (rad/s)
T
r
a
n
s
f
e
r

f
u
n
c
t
i
o
n

a
m
p
l
i
t
u
d
e

(d)
Vs = 40 m/s
Vs = 100 m/s
Vs = 500 m/s
Vs = 40 m/s
Vs = 100 m/s
Vs = 500 m/s
Vs = 40 m/s
Vs = 100 m/s
Vs = 500 m/s
Fig. 7. Transfer functions: (a) foundation shear force, (b) foundation bending moment, (c) tower base shear force and (d) tower base bending moment.
516 M. Harte et al. / Engineering Structures 45 (2012) 509518
of different stiffness. Fundamental modal frequencies of the wind
turbine system were calculated and SSI was shown to lengthen
the vibration period of the wind turbine. The fundamental modal
frequency of the wind turbine system, located on the multi-layered
Site B, was shown to dip into the 1P range for softer soil conditions,
which should be avoided as it may lead to dangerous dynamic
interaction.
SSI is generally considered to have a positive effect on structural
vibrations as it adds damping to the system. However the relative
displacement of the nacelle showed only a slight reduction in re-
sponse for the soil conditions considered, while SSI was found to
have a detrimental effect on the total displacement of the nacelle,
especially for softer soils. This detrimental effect is due to the tall
slender nature of the wind turbine structure. The base shear and
bending moment at the base of the tower and in the foundation
were computed. No signicant difference between the shear and
moment in the foundation and tower base was found, as the foun-
dation inertia was found to be negligible. Peaks in the frequency
response were found to occur at multiples of 3P (three times the
rotation speed) for stiffer soil conditions. SSI was shown to have lit-
tle effect on the peak bending moment or the peak shear force in
the foundation. However the frequency content in the response
time history was signicantly affected due to SSI. This could have
a signicant inuence on the fatigue of the wind turbine founda-
tion system. The rotation of the foundation was shown to increase
signicantly with decreasing soil stiffness and violated prescribed
limit state of DNV/Ris standards for lower soil stiffness
conditions.
Acknowledgements
This work was supported by the Irish Research Council for Sci-
ence Engineering and Technology (IRCSET) and the EU under FP7
Marie Curie ITN project SYSWIND (Grant No. 238325).
References
[1] Veletsos AS, Meek JW. Dynamic behaviour of building-foundation systems.
Earthquake Eng Struct Dynam 1974;3:12138.
[2] Kausel E. Early history of soilstructure interaction, Soil Dynam Earthquake Eng
2010;30:82232 [Special Issue in honour of Prof. Anestis Veletsos].
[3] Luco J. Soil-structure interaction effects on the seismic response of tall
chimneys. Soil Dynam Earthquake Eng 1986;5:1707.
[4] Moghaddasi M, Cubrinovski M, Chase J, Pampanin S, Carr A. Effects of soil-
foundation-structure interaction on seismic structural response via robust
Monte Carlo simulation. Eng Struct 2011;33:133847.
[5] Novak M, Hifnawy LE. Damping of structures due to soil-structure interaction. J
Wind Eng Ind Aerodynam 1983;11:295306.
[6] Novak M, Hifnawy LE. Structural response to wind with soil-structure
interaction. J Wind Eng Ind Aerodynam 1988;28:32938.
[7] Liingaard M, Andersen L, Ibsen LB. Impedance of exible suction caissons.
Earthquake Eng Struct Dynam 2007;36:224971.
0 5 10 15 20
2.784
2.7845
2.785
2.7855
2.786
2.7865
2.787
2.7875
2.788
2.7885
x 10
5
Time (s)
F
o
r
c
e

(
N
)
(a)
0 5 10 15 20
2.3775
2.378
2.3785
2.379
2.3795
2.38
2.3805
2.381
x 10
7
Time (s)
M
o
m
e
n
t

(
N
.
m
)
(b)
Vs = 70 m/s
Vs = 500 m/s
0 5 10 15 20
2.784
2.7845
2.785
2.7855
2.786
2.7865
2.787
2.7875
2.788
2.7885
x 10
5
Time (s)
F
o
r
c
e

(
N
)
(c)
0 5 10 15 20
2.3775
2.378
2.3785
2.379
2.3795
2.38
2.3805
2.381
x 10
7
Time (s)
M
o
m
e
n
t

(
N
.
m
)
(d)
Vs = 70 m/s
Vs = 500 m/s
Vs = 70 m/s
Vs = 500 m/s
Vs = 70 m/s
Vs = 500 m/s
Fig. 8. Response time-histories: (a) foundation shear force, (b) foundation bending moment, (c) tower base shear force and (d) tower base bending moment.
0 50 100 150 200 250 300 350 400 450 500
0
0.002
0.004
0.006
0.008
0.01
0.012
0.014
0.016
Shear wave velocity (m/s)
R
o
t
a
t
i
o
n

(
r
a
d
)
Cone model
DNV allowable tilt
Fig. 9. Rotation of the foundation versus shear wave velocity.
M. Harte et al. / Engineering Structures 45 (2012) 509518 517
[8] Liingaard M. Dynamic behaviour of suction caissons. Ph.D. thesis, Department
of Civil Engineering, Aalborg University, Denmark; 2006.
[9] Veletsos AS, Verbic B. Vibration of viscoelastic foundations. Earthquake Eng
Struct Dynam 1973;2:87102.
[10] Stewart JP, Seed RB, Fenves GL. Empirical evaluation of inertial soil-structure
interaction effects. 11 ed. California, USA: Dunod, Berkeley; 1998.
[11] Satari M, Hussain S. Vibration based wind turbine tower foundation design
utilizing soil-foundation-structure interaction. In: Santini A, Moraci N, editors.
2008 Seismic engineering conference commemorating the 1908 Messina and
Reggio Calabria earthquake. American Institute of physics conference series,
vol. 1020. p. 57784.
[12] Hamaydeh MA, Hussain S. Optimized frequency-based foundation design for
wind turbine towers utilizing soil-structure interaction. J Franklin Inst
2011;348:147087 [The third international conference on modeling,
simulation, and applied optimization].
[13] DNV. Guidelines for design of wind turbines. Technical report ISBN 87-550-
2870-5, DNV/Ris in technical co-operation, Det Norske Veritas, Copenhagen
and Wind Energy Department, Ris National Laboratory; 2001.
[14] Ehlers G. The effect of soil exibility on vibration systems. Beton Ehlers
1942;41(21/22):197203.
[15] Wolf JP, Deeks AJ. Foundation vibration analysis: a strength-of-materials
approach. 1st ed. Elsevier; 2004.
[16] Jonkman JM, Buhl Jr. ML. FAST users guide. Technical report NREL/EL-500-
38230, NREL, Golden, CO: National Renewable Energy Laboratory; 2005.
[17] Kaimal JC, Wyngaard JC, Izumi Y, Cote OR. Spectral characteristics of surface-
layer turbulence. Quart J Royal Meteorol Soc 1972;98:56389.
[18] Hansen M. Aerodynamics of wind turbines. 1st ed. Earthscan Publications Ltd;
2001.
[19] Veletsos AS, Tang Y. Soil-structure interaction effects for laterally excited
liquid storage tanks. Earthquake Eng Struct Dynam 1990;19:47396.
[20] Hansen M, Srensen J, Voutsinas S, Srensen N, Madsen H. State of the art in
wind turbine aerodynamics and aeroelasticity. Prog Aerosp Sci
2006;42:285330.
[21] Clough RW, Penzien J. Dynamics of structures. 3rd ed. USA: Computers &
Structures Inc.; 1995.
[22] Jennings PC, Bielak J. Dynamics of building-soil interaction. Bull Seismol Soc
Am 1973;63:948.
[23] Tong W, editor. Wind power generation and wind turbine design. Boston,
USA: WIT Press Southampton; 2010.
518 M. Harte et al. / Engineering Structures 45 (2012) 509518

Você também pode gostar