Renewable and Sustainable Energy Reviews xxx (xxxx) xxx

Contents lists available at ScienceDirect

Renewable and Sustainable Energy Reviews

journal homepage: http://www.elsevier.com/locate/rser

Considerations for the structural analysis and design of wind turbine

towers: A review
Edwin Hernandez-Estrada a, Orlando Lastres-Danguillecourt a, Jose B. Robles-Ocampo b, *,
Andres Lopez-Lopez a, Perla Y. Sevilla-Camacho b, Bianca Y. Perez-Sariñana b,
Jose R. Dorrego-Portela a
a
Instituto de Investigación e Innovación en Energías Removables, Universidad de Ciencias y Artes de Chiapas, Libramiento Norte Poniente 1150, Colonia Lajas Maciel C.
P, 29039, Tuxtla Gutiérrez, Chiapas, Mexico
b
Cuerpo Académico de Energía y Sustentabilidad, Universidad Politécnica de Chiapas, Carretera Tuxtla Gutiérrez. - Portillo Zaragoza Km 21+500, Col. Las Brisas,
Suchiapa, Chiapas, CP.29150, Mexico

A R T I C L E I N F O A B S T R A C T

Keywords: The use of wind generators has grown exponentially in recent decades to meet the increasing demand for
Wind turbine tower electricity. With both generator design and generation capability growing, the resulting increases in the size of
Types of loads generators require them to withstand multiple and intense dynamic loads. These loads cause greater stresses,
Structural design
fatigue, torsions, deflections, and vibrations, among others, leading to greater failures during a generator’s life
Foundations
Structural analysis
cycle. These issues are of great significance to the research and technological development involved in improving
Fluid-structure interaction the design, manufacturing process, and installation of wind turbine towers. This work presents a detailed review
of the most notable aspects involved in the analysis and design of towers. These aspects include loads and
actuating forces, types of structural analysis, used software, and types of experiments used for validating the
aspects themselves. In addition, different perspectives regarding the types of supports for onshore and offshore
wind turbines are discussed. Likewise, the proposals for new designs and construction materials are also
analyzed. The present review integrates the most relevant aspects and recent developments in the design,
manufacture, and installation of wind turbine towers. This has been carried out with the objective of providing a
contemporary frame of reference that will facilitate the future research and project development related to wind
turbine towers.

Because wind turbines (WTs) are used to convert energy from the
1. Introduction wind into electrical energy, the amount of generated electricity depends
mainly on the rotation speed of the wind turbine (WT), the wind
Electricity demand is increasing significantly due to both rapid resource and the aerodynamic design [4]. A WT comprises three main
economic development and urbanization. It is estimated that global parts, which are the rotor, nacelle and tower. The wind turbine tower
electricity consumption will reach 31 657 TWh by 2030 [1]. Currently, (WTT) elevates the rotor and the nacelle above ground level to a mini-
electricity is mainly produced via fossil fuels [2], which is a major mum height, which corresponds to the diameter of the rotor. This en-
concern as energy crises and environmental problems are expected to sures that the blades do not collide with the ground. The maximum
occur due this increased energy demand. With fossil fuels being rapidly height is limited by cost, as well as by challenges of installation [4].
depleted, the use of renewable energies has attracted much interest, as it However, these constraints are constantly evolving along with scientific
may be used to meet current and future energy needs [3]. Among and technological advances and innovations.
renewable energy options, wind energy is believed to be one of the most The first WTTs had a lattice design, whereas cylindrical or conical
promising, due to the cost-effectiveness, sustainability, and low envi- tubular WTTs built out of steel or concrete are now the most used de-
ronmental impact of the use of wind generators. signs [5]. The criteria for selecting a design depends often on the natural

* Corresponding author.
E-mail addresses: [email protected] (E. Hernandez-Estrada), [email protected] (O. Lastres-Danguillecourt), [email protected]
(J.B. Robles-Ocampo), [email protected] (A. Lopez-Lopez), [email protected] (P.Y. Sevilla-Camacho), [email protected]
(B.Y. Perez-Sariñana), [email protected] (J.R. Dorrego-Portela).

https://doi.org/10.1016/j.rser.2020.110447
Received 19 December 2019; Received in revised form 2 October 2020; Accepted 3 October 2020
1364-0321/© 2020 Published by Elsevier Ltd.

Please cite this article as: Edwin Hernandez-Estrada, Renewable and Sustainable Energy Reviews, https://doi.org/10.1016/j.rser.2020.110447
E. Hernandez-Estrada et al. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx

Nomenclature fy Elastic limit

g Acceleration of gravitation
α Power law exponent Hz Hertz
γ′ Effective unit weight of soil I Moment of inertia
Γ Function Gamma km Kilometer
Δσeq Equivalent stress range due to wind and wave loads KT Thickness exponent
Δσwind Stress range due to wind load k Shape parameter
Δσwave Stress range due to wave load kW Kilowatt
δr Tower deflection due to thrust force L Length of tower
δt Tower deflection due to aerodynamic wind load LPM Lumped-Parameter-Model
distributed along the tower M Bending moment
θ Rotation angle MW Megawatt
λ Wave length m Meter
ʋ kinematic viscosity of air mS-N Negative slope of S–N curve
ν Average speed measured a measurement interval of 10 min mg Mass of the nacelle
ρw Air density mr Rotor mass
ρwa Density of seawater mT Total mass
σmax Maximal stress mt Tower mass
Φ Friction angle m2 Square meter
ω Angular frequency m/s Meter per second
ω1 First natural frequency of tower N Number of data
A Swept area NREL National Renewable Energy Laboratory
AMD Active Mass Damper Neq Fatigue life
API American Petroleum Institute n Surface wave profile
Ab Cross sectional area of the tower nw Wave number
Aefe Area projected for a body on to a plane perpendicular to P Water pressure
the wind direction Pcr Critical buckling load
Aproj,B Area projected for a blade on to a plane perpendicular to P(VJ) Frequency at which wind speed occurs within the interval j
the wind direction p Spring force corresponding to lateral resistance
Aw Wave amplitude Q Spring force corresponding to bearing resistance
a Structure radius Re Reynolds number
B Number of blades SSI Soil Structure Interaction
BEM Blade Element Momentum Theory Su Undrained shear strength
BET Blade Element Theory TWh Terawatt hour
CAsT Computer-Aided Aerodynamic and Aeroelastic Technology t Time
CFD Computational Fluid Dynamics tf Thickness through which the potential fatigue crack will
Ca Hydrodynamic drag coefficient grow
CD Drag coefficient tref Reference thickness
Ce Aerodynamic drag coefficient of the tower Vave Average wind speed
Cm Hydrodynamic inertia coefficient Vc Marine current speed
CT Thrust coefficient Ve1 Extreme wind speed with a recurrence period of 1 year
c Scale parameter Ve50 Extreme wind speed with a recurrence period of 50 years
cm Centimeter Vr Wind speed at hub height
D Structure diameter vi Average wind speed recorded in a time interval i
DNV Det Norske Veritas vj Central value of the wind speed interval j
Dcil Diameter of tower VNom Nominal wind speed
d Water depth Vref Reference wind speed
E Modulus of elasticity Vw Wind speed
F External forcing V(z) Wind speed as a function of height
FAST Fatigue, Aerodynamics, Structures and Turbulence W Watt
FEM Finite Element Method WT Wind Turbine
Faero Aerodynamic forces WTT Wind Turbine Tower
Feqk Seismic load forces ẋ Induced wave speed
FD Drag force of sea currents ẍ Water acceleration
Fg Gravity force y Distance to the neutral axis of the cross section of the tower
FH Hydrodynamic forces z Height on the z axis
FR Thrust force Zr Wind velocity measuring height
FT Aerodynamic wind load on the tower Z0 Terrain roughness
Fwave Wave loads [M] Mass matrix
F(z) Aerodynamic wind load distributed along the tower [C] Damping matrix
f(v) Weibull distribution function [K] Stiffness matrix

2
E. Hernandez-Estrada et al. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx

frequency of vibration, the wind resource, aesthetic considerations and designs and the methods by which these can be optimized to reduce
the proposed installation site. Currently, WTTs are subjected to great costs. Finally, the last section concludes by discussing the future direc-
stress, due to the various dynamic loads that these are required to sus- tion of research in the field of WTT structure.
tain during operation. The main dynamic loads are the interaction of the
wind with the rotor and tower [6] and the weight of both the nacelle and 2. Types of wind turbine loads
rotor. The weight of the latter, which is located at the top of the tower,
causes instabilities in the structure, and its natural frequency is greater 2.1. Wind load
than that of a building of the same height [3]. The effect of earthquakes
on WTs should also be considered. A WT located at sea will also be Although wind loads acting on a WT mainly comprise two types,
subjected to wave and sea current loads on its supports and WTTs. The those acting on the turbine and those acting on the tower, wind is un-
effects of the wind, which become less turbulent but more intense as predictable by nature and exerts thrust on the structure in various di-
WTT height increases, must also be considered. rections and at different intensities.
As wind power continues to develop globally, it is important to un-
derstand and reliably predict the structural response of the tower due to 2.1.1. Wind modelling
various intense external loads. Therefore, detailed and comprehensive Due to the disperse and random characteristics of the wind, the only
structural analytical models must be developed in order to ensure the way of accurately estimating the adequate area of land for the instal-
operational serviceability and safety of WTTs. These models should lation of a wind farm is to analyze wind measurements recorded by
consider both several degrees of freedom and non-linear parameters. weather stations via statistical methods. For this purpose, wind esti-
Also, a detailed structural analysis of installations at sea or on land de- mation methods have been developed. The oldest of these is the Fisher-
mands that the effect of the soil and its interaction with the structure be Tippet method (1928) [11], which was modified by Jenkinson in 1955 b
considered. Furthermore, simulations and experiments are required to y generalized Fisher-Tippet formulas into a single equation. This equa-
validate the analytical results obtained [7–9]. With regard to the above, tion is known as Generalized Distribution of Extreme Values [12]. Fer-
selection and optimization remain very important issues in the struc- nandez [13] made an important contribution by writing a historical
tural and economic development of the tower’s structural systems, as review of the evolution of wind estimation methods.
the cost of the WTT constitutes approximately 20–30% of the total cost There are three types of distribution for the above-mentioned anal-
of a typical WT project [10]. ysis: Type I or Gumbel distribution; Type II or Frechet distribution; and,
Fig. 1 presents a classification of the main parameters involved in the Type III or Weibull distribution. The selection of some of these distri-
design of WTTs. The objective of this study is to contribute ideas and butions has been the subject of intense debate in the scientific commu-
alternatives for future research projects. The information gathered nity. Today, several researchers prefer an adjustment towards a Type I
herein presents current research and development advances for WTs, distribution [7–9]. However, Mayne [14] states that this type of distri-
taking into account the different types of loads, analytical methodolo- bution is adjusted satisfactorily in ‘well-behaved’ climates, namely
gies, foundations and simulations. places with little variability in events that produce extreme winds. A
The second section of the present study describes the most common Type III distribution minimizes asymptotic behavior errors in the dis-
load types considered in WTT designs, such as wind modelling, the force tribution, and although it does not work well for tropical areas, the data
exerted by the wind on the different parts of the WT, the force exerted by can be conditioned to use this method in those areas. It is clear that a
the sea on the structure and its foundations, and the gravitational and general and detailed method for estimating the dynamic behavior of the
other less common loads, such as the seismic load and the load caused by wind is not yet available.
the accumulation of snow. The third section describes the supports used
for WTs and the types of soils in which these are installed. For its part, 2.1.1.1. Average wind speed. In the development of wind energy pro-
the fourth section describes the most common types of structural ana- jects, a database obtained from weather stations and compilation of at
lyses conducted on WTTs, namely: vibration modes, stiffness, de- least one year of wind behavior records is indispensable [3]. This data is
flections, fatigue, and wind force. The fifth section presents the methods obtained from weather stations located in the selected area and then
for solving and verifying the proposed models via instrumental analysis processed and analyzed using various statistical methods. One of the
or the use of specialized software. Afterward, the sixth section presents most outstanding methods is the Weibull distribution, which is recom-
proposals for new building materials for WTTs, as well as hybrid tower mended by the International Electrotechnical Commission’s (IEC)

Fig. 1. Classification of the main parameters involved in the design of WTTs.

3
E. Hernandez-Estrada et al. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx

International Standard 61 400–1 [15,16]. Because of its two adjustment for application in wind energy problems. While the logarithmic model
parameters, the distribution function is better coupled with the data with adiabatic correction is more realistic, it requires more parameters,
obtained than with a Rayleigh distribution. Although this latter distri- such as friction speed, turbulence, and atmospheric stability, and this
bution is also recommended by IEC 61400–1, it has only one adjustment data can only be obtained at sites where specialized measurements are
parameter and is a specific form of the Weibull distribution with a form available, such as wind farms with three-dimensional sonic anemome-
factor of k = 2. ters [25].
The Weibull probability density function is an equation character- On the other hand, empirical models depend on experimental wind
ized by two parameters, known as c scale and k shape [17], The first data obtained over the course of at least one year and based on a
defines how disperse the distribution is and the latter defines the shape comprehensive knowledge of the geography of the site [26]. Moreover,
of the distribution, as shown in Eq. (1). empirical models are less complex and are recommended by IEC
( )k 61400–1,2 [27,28]. Consequently, the logarithmic profile is the most
suitable for the wind profile calculations required for WTs with heights
( )( )
k ν k−1 νc
f (v) = e (1)
c c close to or greater than 100 m [29], due to the curve of the graph of
Equation (2), in which is rapidly stabilized.
where ν is the average wind speed, considering a measuring interval of
10 min. There are various methods for determining the shape and scale
⎛ ⎞
z
ln
parameters, as presented in Table 1. The method for determining the (2)
⎜ z⎟
V(z) = Vr ⎝ z0r ⎠
ln z0
Weibull parameters will depend on the computing and mathematical
resources available to the analyst.
Moreover, if precision is required in order to identify the shape and where Vr is the wind speed measured at a known height (which may
scale parameters, numerical methods are used. These methods use either be the height of the hub or the height of the measuring tower at
mathematical iterations that provide a better fit to the Weibull distri- which the wind speed sensor is located), z is the height at which the
bution and are known as maximum probability [18] and maximum wind speeds are to be obtained, and zr is the height at which the mea-
modified probability [19]. surements are obtained. Therefore, flat sites for the installation of WTs
If rapid calculations are required, the graphical method or the are usually preferred. Potential WT installation sites are as flat as
method of least squares [20] can be used. The latter uses the linear possible, as irregularities in the terrain cause turbulence in wind flow.
equation, interpolating the cumulative distribution function. By using These irregularities may include trees, shrubs, and houses, and their
the standard deviation of wind speed and average speed at the site in value is designated as zo [30].
order to generate the k shape and c scale values, the moment method As the power law (Eq. (3)) presents stable behavior at low/medium
[21] is another acceptable method and requires little mathematical heights, it is used to perform wind speed calculations for low power WTs
knowledge. or for structures that do not exceed 100 m in height.
( )α
z
2.1.1.2. Wind profile. A vertical wind profile is used to obtain the V(z) = Vr (3)
zr
variation in wind speed as the distance from the ground increases [22].
It is necessary to consider this phenomenon, given that the wind speed at where α is the wind shear exponent. Although some studies consider a
the base of the tower will not be the same as that at the top of the return period kr for obtaining the power law [22], this is not considered
structure, which is a variation that is more pronounced if tower are close in Eq. (3) because it has a value close to 1. The selection of any of the
to or over 100 m in height [23]. Although the wind profile can be above equations depends on the data pertaining to the geographical
modelled either quasi-statically or stochastically, the latter is not dis- parameters of the wind farm [31]. In addition, the exponent α is not as
cussed in the present paper. However, it should be noted that constant as expected, depending on other variables, such as zo, wind
quasi-static models have been developed either empirically or with speed, and height. Hsu [32] made observations over the course of a year
theoretical foundations. Based on Monin-Obukhov theory [24], a loga- of winds at altitudes of 10 m and 33 m off the coast of the United States
rithmic model without adiabatic correction and a logarithmic model Virgin Islands, whereby the α coefficients for different atmospheric
with adiabatic correction are the models with the most widely used conditions in any tropical coastal zone were determined. On the other
theoretical foundations. Thus, these models are preferred by researchers

Table 1
The degree of precision of some statistical methods used to obtain the shape and scale factors [17].
Method k C Precision

Maximum probability method N )1

∑N
i=1 vi ln(vi ) i=1 ln(vi )
]−1
k ✓✓✓
[∑
k = 1 ∑N
k
(
−
N
∑N k
i=1 vi
k
C = v
N i=1 i
Modified maximum probability method ]1
[∑N ∑N
j=1 ln(vj )P(vj ) j=1 ln(vj )P(vj )
]−1 ✓✓✓✓
k = 1 N
k
[
−
P(ν ≥ 0)
∑N
j=1 vj P(vj )
k
C = vkj P(vj )
∑
P(ν ≥ 0) j=1

Moment method 2 1 ν
( ) ( )
C = (
✓
Γ 1+ − Γ2 1 +
s2 k k 1
)
= Γ 1+
ν2 1 k
( )
Γ2 1 +
k
Method of least squares or graphical method ln[ − ln(1 − F(ν))] = k ln ν − k ln c ✓✓

Christofferson Method )1
]0.5
N(N − 1)
✓
[
π
k = √̅̅̅ 1 ∑N
k
(
2
6 N( C = vki
∑N 2 ∑N
i=1 ln vi ) − ( i=1 lnvi ) N i=1

4
E. Hernandez-Estrada et al. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx

Table 2 above has been reviewed and critiqued by Wang [42]. Also, other the-
Comparison of the parameters for roughness and the power exponent [33]. ories available in the literature that are modifications of or corrections of
Type of terrain Roughness z0 Power exponent α these five theories.
Open sea with waves 0.0003 [34,35] 0.110 [35,38–40]
Arable land 0.0500 0.160 [37] 2.1.2.1. Rotor in movement. The rotation of the rotor causes a circular
Forest 0.5000 0.210 [36] sweep effect at the top of the tower, while the wind pressure causes force
Suburbs 1.5000 0.250 in the direction of the wind. Momentum theory establishes that the rotor
City centers 3.0000 0.290
swept area is not completely solid, and is, instead, a porous disc through
which air can pass [35]. This concept enables the simplification of the
hand, IEC 61400–1 [27] and Det Norske Veritas (DNV) [33] have pro- calculations and is thus very frequently. With its adjustment iterations,
duced roughness and power exponent tables with average values the BEM model is more accurate, although more detailed information on
(Table 2), which are mainly used in northern Europe. While many re- the geometry of the blades and the dynamics of the air flow is required
searchers use these values for calculations regarding wind turbine de- [42].
signs, studies should be conducted in specific geographical areas, to The momentum model stipulates that, if the disc were completely
obtain superior data for the design of WTTs. solid, the thrust coefficient CT may be up to 1.5 at low speeds [28];
therefore, in WT analysis, this coefficient is below this value (between
2.1.1.3. Extreme winds. Extreme wind conditions are used to determine 0.1 and 0.9) [22]. The thrust calculations exerted by the wind on the
extreme loads on WTs and, therefore, are considered in structural design rotor, FR, can be determined by means of Eq. (6), which is recommended
calculations. These conditions are often caused by storms. under IEC 61 400–1.2.
The procedure by which extreme wind values are obtained begins 1
with wind resource analysis, which is undertaken via the Weibull
2
FR = ρw ACT VNom (6)
2
probability density function and provides a more recurrent annual wind
value, denoted by Vave. The average wind speed values are increased where ρw is the air density, A is the blade-swept area, VNom is the nominal
fivefold (when k > 1.7 or flat terrain). This fivefold increase is obtained speed of the WT and CT is the thrust coefficient. In the design of tower
via standard Gumbel-type analysis of wind speed measurements for structures, the thrust coefficient is the most important and difficult
extended periods of more than 20 years. The reference winds Vref, which parameter to obtain. Finding the right value for this parameter provides
are extreme wind speed values taken at a measurement interval of 10 a more realistic indication of the behavior of the fluid dynamics of any
min at the height of the wind turbine rotor, with a return period of 50 structure, especially that of a WTT. In obtaining the rotor load, the
years, are then obtained. These values are used to calculate the survival design speed generally plays an important role, due to the fact that it
wind speed, or Ve50 [27]. In IEC 61400–1, the ratio between the refer- increases quadratically. Controlling the angle of attack of each blade as a
ence speed (Vref) and the annual average speed (Vave) is constant for the braking mechanism is indispensable. For its part, the maximum force
different WT classes, a relationship based on values typical of both exerted on the rotor axis is obtained when the angle of attack is optimal
British and Dutch building codes [41]. Depending on the values for the and working at nominal velocity. If the angle of attack of the blade
average speed Vave, winds are classified into three classes [27]: Class I, changes from the optimal value, this also changes the thrust coefficient,
which are winds of 10 m/s; Class II, which are winds of 8.5 m/s; and, decreasing its value until reaching maximum wind and, therefore, the
Class III, which are winds of 7.5 m/s. There is a fourth class, known as WT will brake.
the Special Class, which has specific design values. Table 3 Presents the Generally, a low power WT does not have any control for regulating
values for these classes. the angle of attack of the blades and presents the highest thrust force on
Extreme wind speeds Ve50 and Ve1 are used to calculate the maximum the rotor axis at wind speeds between 15 m/s and 25 m/s [43]. Eq. (6)
wind speeds that may occur. Ve50 is used to calculate wind speeds in the can be used to obtain the forces on low power WTs, by substituting the
event of an extraordinary natural phenomenon. This is not to say that nominal speed VNom for 2.5 times the average speed [28].
this will occur every 50 years, but this is a statistical theme which
considers a certain probability that strong winds will occur within at 2.1.2.2. Parked condition. In the event of winds above wind shear, the
least a certain number of years (50 or fewer) with some probability WT should brake [2] by means of its brake controls, which control yaw,
curve, as described in Eqs. (4) and (5). the wind vane, pitch control, and the mechanical brake etc. Eq. (7) is
( )0.11 used to calculate the thrust force exerted on the rotor under these con-
Ve50 (z) = 1.4 Vref
z
(4) ditions (extreme winds and parked condition). However, there are two
zr main load considerations, denoted below in sub-sections A and B.

Ve1 (z) = 0.8Ve50 (z) (5) 1

FR = ρw BCD V2e50 Aproj,B (7)
2
2.1.2. Force caused by wind thrust on the rotor
where B is the number of blades, Ve50 is the extreme wind speed with a
When the air flow encounters certain resistance caused by bodies
obstructing their movement, this obstruction is known as a thrust. For a
WT, the thrust force is exerted by the wind on the rotor and the tower.
The thrust force exerted on the rotor can be calculated by five widely-
studied theories, which are momentum theory; blade element theory
(BET); blade element momentum theory (BEM); vortex theory; and
computational fluid dynamics (CFD) theory. Each theory mentioned

Table 3
Wind speeds for WT design by class [27].
Class I II III Special

Vave 10 8.5 7.5 Specific design values

Vref 50 42.5 37.5
Fig. 2. Blades pitch angles to 90◦ .

5
E. Hernandez-Estrada et al. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx

50-year recurrence, CD is the drag coefficient of a blade, and Aproj,B is the the drag coefficient for the tower, Ce, for each height differential, as this
area of a blade projected perpendicular to the wind. is a more complex task in the development and solution of the models
2.1.2.2.1. Blades pitch angles to 90◦ . As seen in Fig. 2, as an addi- [56,57].
tional safety measure, a WT places its blades in feathered position when To date, research is yet to be carried out on the approximations of
it is parked to ensure that blades are not in angle of attack position, each method for obtaining the force of thrust exerted on the tower. As a
which would cause the turbine to stall [44]. Under this condition, the result, possible errors that could occur in tower design using each
drag coefficient of the blade tends to decrease drastically, whereas the method remain unknow. Maximum thrust force is caused by extreme
level of thrust force on the rotor axis is very low compared to the force wind speeds due to the fact that the tower is static and has to sustain
exerted were the WT to be in operation. On the other hand, IEC 61400–1 storm gusts. Eq. (8) calculates the thrust force of the wind on the tower
recommends that the calculations for the design and resistance of the and is adapted to each of the models discussed above.
structure take into account a situation in which there are extreme winds
1
and the blades are in feathered position. Some studies have applied these 2
F(z) = ρw Aefe Ve50 Ce (8)
2
recommendations and ensured that their calculations take extreme
winds into account, as set out below: Vref (37.5, 42.5 and 50 m/s) [15, where Aefe is the effective area of the tower impacted directly by the
45]; Ve50 (52.5, 59.5, 70 m/s) [37,46]; and real wind speeds caused by wind, and Ce is the drag coefficient of the tower. The drag coefficient for
hurricanes (between 30 and 50 m/s) [47]. a cylinder is calculated according to the graph 2.1 shown in Ref. [139],
2.1.2.2.2. Blades in angle of attack. If an error in the orientation which establishes that the drag coefficient depends on the Reynolds
mechanism of the blades occurs, a WT remains in angle of attack mode, number calculated via Eq. (9), the parameters of which are wind speed
which means that the drag coefficient CD maintains a higher value Vw; diameter of the cylinder Dcil; and the coefficient of the kinematic
compared to the value of CD at 90◦ . As studies on this extreme case have viscosity of the air ʋ. The literature shows that drag coefficient values
not been found, it is recommended that the calculations required to between 0.6 and 0.9 have been obtained for cylindrical towers [33].
analyze the behavior of the structure are carried out in order to compare
the results with data from WTs that have collapsed after exposure to
Vw Dcil
Re = (9)
hurricanes. Thus, It is highly important to take additional design con- υ
siderations into account.
Standard IEC 61400-2 specifies that a situation causing the turbine to 2.2. Wave forces on offshore structures
rotate at high speed, such as extreme winds and all braking systems
disabled, qualifies as an extreme case for the analysis of low power WTs The loads that act on WT supports at high seas are caused by the
without variable pitch regulation. These conditions cause a very high weight borne by the tower, nacelle and rotor; moreover, there are sig-
thrust force on the rotor axis that can collapse the entire system. The nificant horizontal forces caused by the waves and ocean currents [36].
skill of the designers is revealed in their ability to ensure that the WT Waves driven by the wind are the biggest source of force that can be
does not operate under these design conditions, opting instead for exerted on the supports. Furthermore, the swell is irregular, varies in
various braking systems. height and length and can impact on the structure from various di-
rections [58].
2.1.3. Thrust force of the wind on the tower
The simplest model for wind force on a tubular tower considers the 2.2.1. Wave force
diameter of the tower as constant from base to top, and the vertical The force exerted by waves on the WTT supports are mainly calcu-
profile of the wind is deemed as uniformly distributed [48,49], as lated via two methods, the theory of diffraction [59–61] and the Morison
observed in Fig. 3(a). equation, with their respective variants [62–64]. The theory of diffrac-
Another model used to facilitate calculations is to divide the tower tion calculates force of the water pressure acting on a body, and can be
into sections and obtain the thrust force via the sum for each section used when the size of the foundations is sufficiently large in relation to
[50–52], as observed in Fig. 3(b). If the curvature of the vertical profile the amplitude of the movement of the water. This method presents a
of the wind is considered, the mathematical model is approximated to more solid theoretical base, which includes the effects of diffraction.
the real behavior of the structure [53–55], as seen in Fig. 3(c), obtaining However, Rahman [60] states that the expression does not consider the
effect of the properties of the fluid on the structure. Eq. (10) is the total
horizontal force exerted by the waves on the structure.
∫2π ∫n
Fwave = [P](−a cosθ)dzdθ (10)
0 d

where P is the water pressure, a is the radius of the structure, d is the

depth of the water, n is the water surface, and θ is the angle.
Abhinav [64] applies the Morison equation to slender structures,
namely when the length of the wave λ is five to ten times greater than the
diameter of the structure (λ > 5D); therefore, the effects of diffraction
are negligible. Eq. (11) is the basic form of the Morison equation used to
calculate the force of the waves via the speed and relative acceleration of
the particles, which impact against the structural elements.
∫n ∫n
D2 D
Fwave = Cm ρwa π ẍdz + Ca ρwa ẋ|ẋ|dz (11)
4 2
d d

where D is the diameter of the structure, Cm is the inertia coefficient, Ca

is the drag coefficient, and ẋ and ẍ are the induced speed of the wave and
Fig. 3. Wind load models for a WTT. the acceleration of the water, respectively. Begum [59] consider that the

6
E. Hernandez-Estrada et al. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx

induced speed and acceleration of the fluid can be calculated via the areas, obtaining results that highlight the importance of redesigning
first-order linear wave theory and are presented via Eq. (12) and (13). WTTs to avoid the collapse caused by torsion when applying seismic
loads. Chen [71] proposes the use of a spherical tuned liquid damper to
cosh(nw (z + d))
ẋ = Aw ω sin(ωt) (12) reduce the vibration undergone by WTs during earthquakes, designing a
1/20 scale model to study structural vibration, and concluding that
sinh(nw d)
dynamic responses could be reduced by more than 40%. Huo [72]
cosh(nw (z + d))
ẍ = Aw ω2 cos(ωt) (13) analyzed the response of a 1.25 MW WT, with a particular focus on the
sinh(nw d)
interaction of the soil with the structure when subject to seismic
where Aw is the amplitude of the wave, ω is the angular frequency of the movement, highlighting the importance of incorporating the effect of
wave, nw is the number of waves, and t is time. The theory of diffraction soil interaction when analyzing the dynamic response of WTs, especially
and the Morison theory are of extended and corrective form, with those located in soft soil regions. In conclusion, this is not a new field of
objective of arriving at more precise models. For example, research study. Research regarding seismic loads has not always been included in
conducted by MacCamy and Fuchs [65] has extended the diffraction the analyses, due to the complexity of this phenomenon. However, due
model applied for surface water, while the research conducted by to increased energy demands, WTs are being installed in seismic regions.
Havelok [66] has been applied for deep water. It should thus be clear that this type of load must be included in
calculations.
2.2.2. Force exerted by ocean currents Fig. 4 presents a summary of the load types that are most used in the
Drag force caused by ocean currents on the tower supports is pro- structural analysis of towers, where the wind, waves, ocean currents,
portional to the square of the speed of the current, while the drag force and the weight of the structure are the most important factors. It is
acts in a normal direction on the axis of the structure [36] and can be necessary to include the highest number of loads possible to predict with
calculated via Eq. (14): certainty the behavior of the WT. However, as the model becomes more
detailed, the solution becomes more complex, although this does bring
1
FD = ρwa Ca Vc2 (14) the benefits of greater accuracy and safety in the design of WTTs.
2

where FD is the drag force per normal unit of length exerted on the axis of
the structure, ρwa is the density of the seawater, and Vc is the speed of the
current.

2.3. Gravitational loads

The load corresponding to the mass of a WT comprises three main

elements, rotor, gondola and tower, each of which causes normal force
and momentum on the base. Eq. (15) is used to obtain the load, adding
the effect of gravity g on the mass of the rotor mr, tower mt and gondola
mg [67].
Fg = mr g + mt g + mg g (15)

2.4. Other loads

Ice is another important design criterion for high structures in cold

climates, with lattice structures particularly vulnerable to the formation
of ice [68]. The drag coefficient of the structure increases considerably
in the presence of accumulated ice. International standard ISO 12494
[69] presents a methodology for evaluating the ice load on structures.
However, for tubular towers, the ice load does not directly affect the
structure, instead accumulating on the nacelle, causing additional
weight to be borne by the structure. Lasse [68] used a 7.5 m high lattice
structure and a 127 m high lattice television tower, comparing ice
masses on both structures and then calculating the ice masses on them
according to the ISO 12494 method. As his results show that the ice
classes formed on the structure are comparable with those published in
the standard, this methodology is a useful tool for evaluating frost and
ice loads on the structures.
When being designed, WTTs generally take the effect of the wind
load into account, considering it as the sole source of dynamic envi-
ronmental perturbations. However, the damage that may be caused by
earthquakes are not generally taken into account in the design [53,59],
even in areas of high-risk seismic activity. Bazeos [67] sets out the need
to consider seismic loads, applying them to onshore WTs. Wang [35]
investigated the dynamic responses of an offshore WT supported by
monopile foundations and subject to wind, waves, and earthquakes,
using the OpenSees open source software platform and developing a
three-dimensional finite model of the system. Triantafyllos [70] uses
accelerograms of real earthquakes that have occurred in specific study Fig. 4. Types of loads acting on WTT.

7
E. Hernandez-Estrada et al. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx

Table 4
Types of supports for WT.
Type Description Fig. Reference

Onshore A concrete plate retained in the ground by 9-a [73,80,

gravity and reinforced with piles if the soil is 86]
soft.
Gravity Consists of a circular pile and a concrete slab 9-b [87]
which is buried as deep as possible and kept in
place by the weight of the structure. Only
applicable for superficial waters, since it
becomes very expensive for deep water
applications. It is very structurally safe.
Monopile This type of support is currently the most 9-c [88–90]
commonly used, especially in European waters
no deeper than 30 m. Consists of a circular pile
Fig. 5. Wind turbine foundations on different soils.
buried in the seabed at a depth similar to the
depth of the water. Installation requires less
3. Types of wind turbine supports drilling effort and is economical for sand and
gravel soils. However, the installation causes a
lot of noise, which would be a problem for
3.1. Onshore wind turbine foundations fishing and environmental activities.
Suction The bucket-type structure is inverted and 9-d [91]
The design of foundations for an onshore WT has been studied for bucket buried in the seabed, providing greater
decades, with two types of design being established, which depend resistance at a lower cost than other types of
anchors at depths of 5–60 m. Its installation is
mainly on the properties of the soil [73], as seen in Fig. 5. If the soil is
economical and low noise, since it does not
dense, gravity foundations are used, whereas softer are reinforced with need heavy machinery, compared to the
piles in order to prevent the structure from falling over. However, given monopile type that requires 50% less material.
that the demand for WTs continues to grow and that sites with favorable However, its behavior for deep waters is not
wind availability are becoming increasingly rare, ever-higher WTs are well defined. Therefore, it is necessary to carry
out exhaustive studies related to the suction
being developed to exploit wind resources in areas with low speeds. In cubes.
addition, the significant seismic activity in many of these areas requires Tripod With their load distributed uniformly across 9-e [36]
the rethinking of both the patterns of interaction of the structure with their three feet, these structures are used for
the soil (Soil-structure-interaction, or SSI) and the design of its deep waters. It is relatively low in weight.
Jacket Although this type of support is identical to a 9-f [92]
foundations.
lattice tower, the jacket support requires more
The first onshore SSI studies addressed the vibration undergone by complex structural analysis and, is, thus, used
WTT foundations in an elastic soil zone [74], and a full review of the in deep waters. It is economically viable due to
development of SSI over time can be found in Kausel’s work [75]. The its relatively low weight.
cone model presented by Wolf [76] is a simplified analytical approach to Floating When the waters are deeper, a system of 9-g [46,93]
floating supports is used, consisting of a base
simulating the effects of SSI. The Lumped-Parameter-Model (LPM) is designed to keep the structure afloat and
another simplified analytical approach, which enables efficient SSI mooring lines anchored to the seabed. Most of
calculations in the time domain for shallow and pile foundations [77]. the floating supports are prototype versions and
Bazeos [67] produced one of the first studies to describe the alternation are aimed at very deep areas (100 m and 200 m)
and with WTs with more than 5 MW. Upcoming
of the natural frequency of a WT installed in ‘well-behaved’ soils. Taddei
projects like the one to be installed in
[78] applied Finite Element Method (FEM) coupling and the LPM to Fukushima are expected to provide information
study the effects of SSI on a WT with foundations in a shallow layer of for future commercial farms.
soil. Andersen [77] used the LPM to show how the SSI alters the WTs
resonance frequencies, which depend on the stiffness and depth of the
soil layer. Santangelo [79] investigated the behavior of the turbine
tower under earthquake conditions and subject to wind load. While the
dynamic behavior of small pile groups is well known, the question re-
mains as to whether the results can be generalized to multiple pile
networks, which are widely used in the construction of onshore WTs.
Other study trends found in the literature relate to monitoring failures
and the sinking of turbine tower foundations [80–82].

3.2. Offshore wind turbine supports

The main advantages of building offshore WTs are the presence of

winds with less turbulence, the presence of winds with less shear effect,
and fewer civil complaints about noise and visual pollution [83]. One of
the disadvantages of offshore WT construction is the cost, which is 1.5–2
times higher than an onshore WT, due to the need supports, foundations,
and marine wiring [84]. While the design of an offshore WT support is
similar to the design of oil platforms, the design principles for the latter
Fig. 6. Types of support for WTs.
should not be applied directly for the former, requiring experimental
studies to be carried out in order to update the corresponding building
three classes: surface water (0–30 m); transition water (30–50 m); and,
codes.
The type of support is chosen according to the depth of the water and deep water (50–200 m) [83]. More than 90% of offshore generators are
located 10 km from the coast, at depths of less than 20 m [85], and
the properties of the soil. The depth of the water can be classified into

8
E. Hernandez-Estrada et al. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx

Table 5
Properties of soils characteristic of the east coast of India [64].
Depth [m] Type γ′ [kN/m3] Φ [◦ ] Su [kPa]

0.0–1.5 Sand 8.0 20 –

1.5–5.2 Clay 8.0 – 20
5.2–6.6 Sand 8.5 20 –
6.6–8.8 Clay 8.5 – 20
8.8–11.7 Sand 9.0 25 –
11.7–13.1 Sand 9.0 30 –
13.1–15.6 Clay 8.5 – 35
15.6–16.7 Sand 9.0 25 –
16.7–37.0 Sand 9.0 30 –
37.0–49.9 Clay 8.5 – 110

and clayey [64]. Both are deposited on the seabed, forming random
layers of different thickness. Abhinav [64] described the properties of
the sea floor on the east coast of India, as a reference for modelling
monopile-type supports. Table 5 presents the properties of this soil,
where γ′ refers to the weight per unit of volume, Φ is the angle of internal
friction and Su is the undrained shear resistance.
Dynamic SSI models are classified mainly into three types, which are
the equivalent linear spring model, the distributed non-linear spring
model, and the continuous element model. The equivalent linear spring
model is used for its simplicity and shorter calculation time. Adhikaria
[94] developed a WT model composed of two aggregated masses, which
correspond to the WTT and nacelle, where the former connected to the
ground with a vertical spring and an oscillating spring. Adhikaria found
that the first natural frequency changed significantly in relation to low
soil stiffness, suggesting that degradation of the soil modulus is
dangerous, especially when the soil is soft. The above described model
was improved by Harte [95], who included viscous damping between
the ground and the tower and between the tower and the nacelle,
whereas Sapountzakis [96] connected a spring with a rigidity equivalent
to the WTT and modelled with a finite element method. Authors found
that the structure with an equivalent stiffness had a first natural fre-
quency that was 6% lower than the natural frequency of the structure
with a fixed limit. In general, this model is valid for linear materials used
on homogeneous soils. In addition, the model is not applicable for
offshore WTs with monopile-type support, as it is very thin [33]. Also, it
should be noted that the bedrock must be twice as deep as the length of
the piles so that these equations can be applied. Therefore, this design is
not viable when the pile is located near the bedrock. While many of the
equations used in this model originated in the oil industry, the nature of
an offshore WT is very different to that of an oil rig. Therefore, in order
to continue using equations from the model, the recommended equa-
tions should be modified in code Det Norske Veritas [33] and updated
based on experimental data [83].
The distributed non-linear stiffness model is characterized by the
non-linear stiffness distribution between the soil and the structure
(Fig. 8), with the spring parameters characterized using load deflection
curves. One of the organizations that can provide data on soil is the
American Petroleum Institute (API) [97], which provides various types
Fig. 7. Diagram of a hybrid support for offshore WTs. of load deflection data, such as p-y curves corresponding to lateral
resistance curves, t-z curves corresponding to vertical resistance, and Q-z
mostly use monopile-type supports. In addition, different types of sup- curves corresponding to bearing resistance, where p, t, and Q are the
port designs have been developed for offshore WTs and are presented in spring forces. Studies have used this model to predict the dynamics of
Table 4 and Fig. 6. offshore WTs, where those by Bosoi [98] and Gerolymos [99] are the
In recent years, new prototypes have been developed, consisting of most noteworthy. Bosoi found that the displacement of the tower is
hybrid arrangements of those mentioned above, as seen in Fig. 7. This greater than that observed for the equivalent spring model, whose
new hybrid support is composed of a jacket and a gravity base and is modelling used a Euler-Bernoulli beam element, whereas the springs
reinforced with piles, this is a configuration that can provide stability to were characterized by the curves p-y, t-z, and Q-z, and aerodynamic and
the structure when unfavorable soil properties are present. hydrodynamic loads were applied. Gerolymos used translational and
rotational springs, taking the effect of soil-structure damping into ac-
3.2.1. Properties of the sea floor count, thus obtaining analytical expressions for both homogeneous and
While the properties of the sea floor vary from region to region linear soils. While studies on both non-linear and non-homogeneous
around the world, the properties comprise two main soil types, sandy soils were presented [100], Gerolymos found that it considering

9
E. Hernandez-Estrada et al. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx

displacement. While this type of SSI model is very accurate, predicting

long-term behavior using continuous models requires high performance
computing for these models.

4. Types of analysis conducted on wind turbine towers

4.1. Force analysis

The maximum resistance of the tower is determined via the static

load calculation method and is based on the weight of the tower, rotor
and nacelle. Moreover, the bending moment that the base of the tower
can withstand is considered, due to the aerodynamic thrust force exerted
on the rotor and tower [107]. Karpat [52] segmented wind loads to-
wards the tower in order to facilitate the optimization methodology,
while Hegseth [107] demonstrated that the most extreme thrust con-
ditions in WTs are produced when maximum rotor speed is reached with
nominal wind speed.
Axisa [46] applied the conditions stipulated in IEC 61 400–2 [28],
thus demonstrating that the highest thrust force will occur when the
rotor has its brakes engaged and when extreme wind velocities (Ve50)
take place. Also, Wood [108] demonstrated that, with a spinning rotor
and a cut velocity, a push force with a similar magnitude will arise.
The value of tension in the tower will be determined by the weight of
the structure and the bending moment exerted by the wind on the tower
and rotor. Thus, the maximum tension will be obtained at the base of the
tower. Moreover, the bending moment caused by the wind load also
reaches its maximum value at the base of the tower. Liu [2] states that
(Eq. (16)) is essential in tower design. His analysis of a 1.5 MW WT
revealing that contemporary designs including the use of steel at
thicknesses of 40 mm at the bottom of the tower, do meet the re-
quirements for maximum stresses.
Fig. 8. Distributed spring model for monopile-type supports.
Fg M y
σ max = ± (16)
stochastic wind loads rather than quasi-static loads would improve Ab I
precision. These studies reveal that the use of this model requires pre-
where Fg is the weight of the WT, Ab is the area of the cross section of the
vious experiments that determine the spring stiffness parameters [101].
tower, M is the bending moment, I is the moment of inertia, and γ is the
In addition, this model is mainly applicable to monopile-type supports.
distance of the neutral axis in the cross section of the tower.
It should be mentioned that difficulties may arise when applying this to
multiple foundations, such as jacket or tripod-type supports, due inter-
action among piles through soil [83]. 4.2. Deflection analysis
Continuous models predict more accurate nonlinear SSI behaviors
than equivalent and distributed spring models. In addition, the param- WTTs are designed to withstand various loads that cause buckling,
eters for continuous models depend less on the design of the structure, deflections, and vibrations in the system.
while the parameters of equivalent and distributed spring models vary
with the shape and size of the structures. Furthermore, the models can 4.2.1. Deflection due to wind
be applied to the study of multiple supports, such as jacket and tripod Deflection in the towers is determined by the lateral load values
supports. Various studies on this model have been undertaken and can caused by the wind and the weight of the structure. These loads will
be classified as linear elastic, nonlinear elastic, elasto-plastic, visco- cause the displacement of the tower, but this shift will be avoided by
elastic, and elasto-visco-plastic models [102]. Currently, one of the most fixing the tower to the foundations in the base. However, displacement
widely used models is the elasto-plastic model with the Mohr-Coulomb may occur at the top of the tower, causing curvature, as observed in
fault criterion. This model assumes an invariant yield surface over the Fig. 9.
plastic deformation [103]. When deflection is small, this curvature effect will not cause the
Haldar [104] used soil where shear strength varying over space with tower to break. However, deflections can cause more damage to the
normal distribution, predicting the response of a pile to side loads and structure if the load conditions surpass the elastic limit of the material
demonstrating that the load capacity of a pile improves when resistance [55]. The deflection caused by this load is given in Eq. (17). Feliciano
is distributed more evenly over space. Achmus [105] predicted the [48] developed an analytical model to calculate the maximum de-
behavior of piles in horizontal and vertical load combinations using the flections of a 5 MW WT characterized by the National Renewable Energy
Mohr-Coulomb fault-based elasto-plastic model, based on numerous Laboratory (NREL). The obtained results show a displacement of up to
load-displacement curves estimated from different load conditions. 80 cm at the top of the tower. Huo [67] considered seismic effects in
Achmus presented diagrams showing the displacement ratio between deflection analysis conducted on a 1.2 MW WT, obtaining up to 120 cm
the stack and the combined loads at different stack lengths and di- displacement. This data which is very critical to predicting possible
ameters. These diagrams are useful for estimating the equivalent stiff- failure via overturning or fracture.
ness of the piles if the effects of the moments are included.
FR L3
Abdel-Rahman [106] compared the results obtained using the δr = (17)
3EI
plastic-elastic model with Mohr-Colulomb fault criteria for monopile
bases. Their finding suggesting that p-y curves might underestimate pile where FR is the load on the rotor, L is the height of the tower and E is the

10
E. Hernandez-Estrada et al. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx

elasticity modulus of the material. The wind load over the length of the
tower is a non-uniform distributed load. This will cause a lower level of
displacement compared the load on the rotor. Some designs have opted
to linearize the load of the vertical profile of the wind to an average
uniformly distributed load in order to simplify calculations, although at
the cost of precision in the design. The distributed linear load is con-
verted into a point load FT located at the tower’s centre of gravity [48,
52]. Therefore, the displacement of the tower due to wind load is given
in Eq. (18).

5FT L3
δt = (18)
48EI
Once both displacements have been determined, these are added and
their total value must be lower than 2% of the length of the tower, in
order to prevent the failure of the foundations and the tower from
tipping over [72].

4.2.2. Buckling analysis

The buckling of the tower is caused by the mass of the nacelle and
rotor, especially if the load is not found at its centre of gravity [48]. To
avoid the occurrence of this phenomenon, it is necessary to increase the
structure’s stiffness. Therefore, the load that can be borne by a cylin-
drical thin-section tower is given by Eq. (19).

π2 EI
Pcr = (19)
4L2

4.3. Fatigue analysis

A static calculation is not sufficient to determine the structural

behavior of the tower in the long term, as these static steel structures are
designed to last more than 100 years [109]. It is important that research
regarding fatigue is conducted to avoid structural failures in the tower.
The phenomenon of fatigue is of particular importance in structural
design, given that the number of load cycles sustained during the
structure is useful life is in the order of 106 cycles. This was show by Hsu
[5], where dynamic structures have been designed to last for more than
20 years, according to the building code for structures, ISO 2394 [109].
A material’s resistance to fatigue is determined based on its fatigue
curve, in relation to the number of load cycles required to cause the
breakage of the material [36], whose value is estimated in Eq. (20).
( ( ) KT )
tf
(20)
N
log10eq = loga10 − mS−N log10 Δσeq
tref

where Neq is the number of fatigue cycles, mS-N is the negative slope for
the S–N curve, tref is the tower reference thickness, KT is its thickness
exponent, and tf is the thickness by means of which a fracture can be
produced.
Fatigue can occur in a WTT due to different factors, such as turbu-
lence, cyclical loads and variation in wind direction. For these factors, a
range of equivalent forces Δσeq is used, comprising the sum of the forces
on the structure, mainly caused by the wind Δσwind and the action of
waves Δσ wave [36], and is denoted by Eq. (21).

(21)
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
Δσ eq = Δσ2wind + Δσ 2wave

4.4. Analysis of vibration and resonance

A structure is designed based on tests of its natural frequency, and is

made to not coincide with other frequencies causing resonance, such as
spinning of the rotor. Kim [110] implemented a frequency analysis for a
Fig. 9. Loads and forces exerted on WTT.
5 MW offshore WT, finding that the first frequency of the vibration mode
is 0.290 Hz and is located between the rotation frequencies of the rotor
and the vibration frequency of the blades. Kim concludes that, while
WTs with broad foundations have similar frequency behaviors, it is

11
E. Hernandez-Estrada et al. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx

necessary to perform vibration studies for each specific structure.

Ahmad [111] used accelerometers in a Nordex N43 WT, finding higher
vibration modes ranging from 0.701 to 17.96 Hz for the first and eighth
modes, respectively.
Resonance is an effect that tends to increase the amplitude of the
tower’s movements, due to the nature of wind or by spinning of the
rotor. The most important modes of vibration in the tower are the first
and second modes and lateral bending [112], given that these most
closely resemble the modes of vibration of the rotor and the movement
of the passing blades. For the first mode, the value for the natural fre-
quency of vibration is calculated in Eq. (22).
√̅̅̅̅̅̅
3.516 EI
ω1 = 2 (22)
L mT

where mT is the value for the total mass of the structure. The value for
the natural frequency of the vibration of the tower must not coincide
with another frequency close to the first natural frequency that could
cause resonance in the structure. For example, Table 6 presents the
natural frequencies of a 5 MW conical tower [113]. For this particular Fig. 10. Types of analysis most used for WTTs.
case, the tower was designed to ensure that its natural frequencies are
distanced from the other natural frequencies produced by the movement full dynamic simulation using the FEM code of CAsT (computer-aided
of the blades, and the rotation of the rotor and drivetrain. aerodynamic and aeroelastic technology). This model couples the WTs
Fig. 10 presents the types of structural analysis that have been most aerodynamic load and the tower’s elastic properties, concentrating on
studied over the last five years, where the most common research topic the mass of the blades, hub, and nacelle at the top of the WT. The results
involved performance of vibration modes under different environmental were compared with the FEM model, presenting a 5% error in the first
conditions in order to ascertain the dynamic behavior of the structure. natural frequencies due to the influence of equivalent mechanical
The second most common research topics have studied the deflections properties. The FEM-CAsT model was later used by Xu [115], who used a
exerted on the towers. Other topics include structure resonance and simplified model of the elastic tower and the rigid rotor from a com-
material fatigue, the limitations of which involve the electronic instru- mercial WT with a tubular tower. It was found that the model provides
mentation of the WT. Such instrumentation requires large economic aerodynamic damping variation, due to the variations of the yaw angle
investment, a high level of computing power, and access to a WT and a of the blade and the correction factors in the vibration modes. Among
wind farm. This last activity mentioned can be a difficult task, some- the advantages of CAsT code is the fact that it does not use a modal
times impossible. overlay solution for dynamic equations, such as common commercial
aeroelastic codes for WTs. Instead, it directly solves motion equations
5. Dynamic analysis techniques applied on the tower structure directly for an arbitrary set of forces at each step, based on an aeroelastic
calculation procedure. This has an effect on structure and on the forces
5.1. Dynamic modelling generated by the structure itself. This code has been verified by field
tests.
All the structural dynamics problems can be formulated based on the The main drawback to estimating the natural frequency of WT
equation of motion (Eq. (23)). structures by using the simplified modelling of a single degree of
→ freedom is that part of the mass of the structure is concentrated at the
[M]→
¨z + [C]→
z˙ + [K]→
z = F (23)
top of the tower. In practice, this concentration ratio causes simulation
where z is the vector of generalized coordinates, [M] is the mass matrix, errors, while the behavior of WT structures is influenced by the flexi-
and [C] and [K] are the damping and stiffness matrices, respectively. bility of the foundation. To reduce these errors, Yung [116] developed a
The external exciting forces are represented by F and can be caused by simple model of a single degree of freedom. Therein, the concentrated
wind, waves, ocean currents, snow, and earthquakes [71]. spring model was introduced to simulate the flexibility of the foundation
A WTT can be simplified as a tubular conical cantilever beam with and equivalent elastic and mass properties were proposed. The stiffness
the total mass of the rotor and nacelle at the top, a model which re- varies with the height of the tower, thus obtaining a good correlation of
sembles an inverted pendulum [54], as observed in Fig. 11-(a). As the the results.
degrees of freedom increase, the system moves closer to the real phe-
nomenon, as observed in Fig. 11-(b). Ishihara [114] analyzed the natural 5.2. Computing tools
frequencies of a 400 W WT and determined its damping ratios using a
The development of computational simulations uses the finite
element method to obtain the structural responses of WTs. Various
Table 6
platforms using this method have been developed, some of which can be
Natural frequencies of a 5 MW WT [113].
observed in Table 7. However, not all of them were produced exclusively
Description Frequency (Hz) for the analysis of WTs, as these are used for a great variety of structures.
1st Tower Fore-Aft 0.3240 The FAST (Fatigue, Aerodynamics, Structures and Turbulence) code
1st Tower side to side 0.3120 is an integrated aeroelastic simulator capable of predicting the extreme
1st Drivetrain torsion 0.6205
loads and fatigue loads of two and three-bladed horizontal axis WTs
1st Blade Asymmetric Flapwise yaw 0.6664
1st Blade Asymmetric Flapwise pitch 0.6675 [130]. With its operational diagram presented in Fig. 12, the most
2nd Tower Fore-Aft 2.9003 popular platform developed for the structural analysis of WTs can be
2nd Tower side to side 2.9361 seen. This platform was developed by NREL and provides multiple
2nd Blade Asymmetric Flapwise yaw 1.9337 analysis tools, which will now be mentioned. Firstly, TurbSim provides a
2nd Blade Asymmetric Flapwise pitch 1.9223
numerical simulation of a flow that contains turbulence structures. For

12
E. Hernandez-Estrada et al. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx

Fig. 11. Degrees of freedom for dynamic WT models.

its part, AeroDyn is an aerodynamic WT modulus, which is able to

Table 7
couple with the FAST tool to enable the aeroelastic simulation of WTs
The most commonly used software for the structural analysis of WTTs.
with a horizontal axis. Finally, HydroDyn is a hydrodynamic modulus
Software Type of analysis Category Reference coupled to the FAST software so that hydrodynamic simulation of ma-
FAST Specialized for structural analysis Open source [89, rine WTs is facilitated. HydroDyn is applicable to both fixed and floating
of onshore and offshore WTs with 117–119] marine substructures.
a horizontal axis
ANSYS Simulation via finite element for Commercial [43,50,55,
structures and fluids 72,120]
ADAMS Dynamic simulation of Commercial [72,121]
5.3. Experimental studies
multibodies
ABAQUS Package of programs that apply a Commercial [91,111, Simple models for the elastic tower, rigid rotor, and FEM models of a
finite element for dynamic 117,122] complete WT have been validated experimentally in research that
structural, static, and thermal
required the electronic instrumentation of the WTTs. This has draw-
calculations, among others
OpenSees Applied for the analysis of Open source [35,117] backs, such as high cost, the need to implement a control, and instal-
structures subject to earthquakes lation difficulties. Ishihara [131] conducted excitation tests for a 2.4
HAWC2 Realistic wind turbine simulations Commercial [123–126] MW horizontal axis WT, using an active mass damper (AMD) to identify
in the time domain. Describes the structural parameters for the first and second vibration modes. He
models with external effects, the
application of loads, structural
proposed an empirical formula for estimating first-mode structural
dynamics and control systems damping for steel-tower WTs via the use of excitation test data. Zen-
Flex 5 Dynamic simulation of WTs Open source [90] dehbad [34] developed a portable opto-mechanical platform for
SHAKE2000 I-D Analysis of Geotechnical Commercial [127] measuring tower deviations which does not require modifications to the
Earthquake Engineering Problems
turbine structure, obtaining a maximum deflection of 21 cm at the tip of
I-D Analysis of Geotechnical
Earthquake Engineering Problems the tower. However, as these results must first be compared to those
MATLAB- The Comsol Multiphysics Commercial [128,129] obtained using other measurement instruments, the accuracy of this
Simulink platform, it is possible to model measurement method is yet to be ascertained.
dynamic systems for WTs Recent years have seen interest in a better understanding of dynamic
CAsT Developed to predict wind loads Open [114]
acting on WTs Source
structural behavior, with the literature focusing mainly on natural fre-
quencies, modes of vibration and damping relationships. Experiments

13
E. Hernandez-Estrada et al. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx

Fig. 12. Structural schematic of the FAST software.

have been developed in order to verify the proposed models or better computer programs that obtain results increasingly closer to the real
understand the behavior of the structure. This experimentation also phenomenon. At the end of the list were experimental analyses
enables the determination of the modal parameters of an invariant linear attempting to corroborate mathematical models (18.19%). In this last
system over time, as a simple way of understanding the phenomenon. An category, (12.98%) conducted scale experiments for cost reasons and
understanding of the modal parameters can serve various objectives, 5.21% conducted experiments in real conditions with the WTs in oper-
including structural modification, evaluation of structural integrity and ation. These conditions were limited due to difficulty experienced by
reliability, and the updating of theoretical models. researchers in gaining access to wind farms.
Literature shows that experimental studies are mainly conducted to
scale. Kim [132] developed a 1:300 scale model of a 3 MW WT located in 6. Geometric optimization and sustainable materials
Korea. The scale model was electronically instrumented with in-
clinometers and accelerometers, with which vibration modes were ob- Most WTTs are built out of steel or concrete, as these are highly
tained. Results were then compared with FEM obtaining a 3.2% error for resistant materials. The main types of steel towers are lattice and
the first vibration mode. These experiments involve low costs and tubular, whereas concrete towers have steel reinforcements, with either
simulate behaviours that resemble real conditions of operation [111]. construction conducted on site or the towers delivered in prefabricated
Thus, it is possible to correlate the dynamic models obtained to date and form. There are also hybrid solutions available which combine steel and
to make improvements design. reinforced concrete. Each type of design has a range of applications
Fig. 13 summarizes the research conducted on all dynamic analysis depending on its height. Table 8 presents the main construction alter-
techniques that have been used for the study of WTTs over the past five natives for wind towers. There is great demand for higher-capacity WTs
years, where the main research categories are specified. Up to 45.45% of to be installed at high altitudes or in places with high wind speeds.
the studies use computational solutions to test mathematical models or Hybrid structural solutions with tubular and lattice sections can be used
experimental results; followed by 36.36% developing simple and com- to meet the safety and durability regulations. According to recent
plex mathematical modelling, which contributed towards developing studies, these are economically and environmentally sustainable
solutions.
Gkantou [44] analyzed a hybrid tower consisting of a 60 m tube at
the top and a 60 m lattice structure at the bottom. The structural
behavior of the hybrid tower was compared with a tubular tower of
equal magnitude, the hybrid tower being twice as resistant. The axial
forces in the lattice section presented mean values at wind speeds of 11
m/s. Chen [134] worked with a similar model for optimizing the
structural system, proposing, based on his results, fewer joints in the
lattice section, which suggests a simpler configuration and a lighter
structure.
Another important aspect to consider is the cost, which will increase
along with the increase in the height of the tower, due to transportation,
assembly, construction, and maintenance costs. Therefore, it is impor-
tant to obtain a structural and economically reliable WTT design via
structural optimization. Negm [135] developed five strategies for opti-
mizing the structural design of WTTs. As a result, it was discovered that
the best results were achieved when a genetic algorithm was used.
Gencturk [4] studied the cost savings that could be achieved in the
design of lattice towers via optimization. Once the cost of the founda-
Fig. 13. Analysis methods for WTTs structure. tions and the connections was included, it was concluded that up to 50%

14
E. Hernandez-Estrada et al. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx

Table 8
Design features for WTTs [5].
Material Design type Application [m] Weigh [ton/m] Advantage Disadvantages

Steel Lattice 20–160 2–3 -Lower cost in construction and transport [10] -Vulnerable to fatigue
-Superior stiffness [133] -Not visually pleasing
-Less wind load to the tower [133] -Causes turbulence to the blades [133]
-Ice load in cold places [133]
Tubular 20–120 2–5 -Aesthetically pleasing [133] -High transport and assembly costs
-Predictable dynamic and fatigue properties [133] -Expensive for very high towers [133]
-Convenient maintenance
Concrete On-site 60–115 8–19 -High Durability -Weather conditions vulnerability
-Superior stiffness
Precast 80–120 8–19 -Easy installation -Vulnerable joints
-High transport and assembly costs
Hybrids Hybrids 80–150 3–15 -Low cost of construction and transport -In experimental stage
-Expected to solve weaknesses of previous alternatives

savings can be made in terms of the material used to construct tubular • Little research has been conducted on the foundations of onshore
towers. WTTs, which continue to be designed under the established con-
struction principles, albeit with some modifications. However, with
offshore WT supports of great contemporary relevance, studies are
6.1. Manufacturing materials being conducted on new types of supports, in search of improve-
ments to the stability of the structure in deeper waters.
Small WTs can offer a cost-effective option for electricity generation • The mathematical models that are used for modelling the system
in remote and off-grid regions of developing countries. Small WTs are present similar characteristics, with each author contributing data on
classified by the rotor area, which should be less than 200 m2 [28]. Most the types of external exciting forces. The verification of the numer-
of these turbines are installed in monopole steel towers, which are ical models is based on the FAST tool and/or via a computer-
generally difficult to transport to remote locations. Clifton-Smith and generated finite element. Although experimental studies would
Wood [136] reported that the manufacturing cost of monopole towers provide a better schematic for the real behavior of the structure, this
can correspond to 30–40% of the overall wind turbine costs. Moreover, is very expensive and, generally, the access to wind farms that is
the cost of transportation to remote areas can be very high, due to road required for undertaking the experiments is not given. However,
access problems. Research has also investigated the feasibility of using experimenting with to-scale models can provide very realistic
natural sustainable materials and timber materials, such as bamboo, for approximations.
the construction of lattice towers, or the recycling of aero-water pumps, • Tubular steel towers formed by steel plates and connected by flanges
to minimize manufacturing and transportation costs. To date, very few are widely used and popular in the industry, due to their aestheti-
studies have focused on the design aspects involved in small WTTs. cally pleasing appearance, their good dynamic behavior, and for
Wood [108] analyzed monopole and lattice towers based on the safety having the same rigidity to bending in all directions. However, as the
requirements stipulated in IEC 61400–3. Clausen [137] studied the size of the WT increases, the thicknesses and diameters of the steel
design of triangular and square lattice towers using FEM. Adhikari [138] tubes must be increased to withstand stresses, increasing
developed a design procedure for triangular and rectangular steel lattice manufacturing costs as a result. The eminent alternative is the design
towers, demonstrating that preventing leg buckling in direction of the of hybrid WTTs.
wind is a crucial design requirement. The aforementioned studies have
suggested that lattice towers are cheaper than monopole, while lattice
towers can be manufactured using simple technology and with minimal Declaration of competing interest
labor.
The authors declare that they have no known competing financial
7. Conclusions interests or personal relationships that could have appeared to influence
the work reported in this paper.
After the considerations included in this review, the following points
can be inferred regarding analysis and structural design of WTTs: Acknowledgements

• The variables and parameters that most influence the design of a The authors would like to acknowledge the Mexican Government
WTT are determined mainly by the action of the wind and water and foundation CONACYT for the financial support for the postgraduate
the weight of the WT. The literature review reveals that the authors students.
present an adequate set of parameters for their particular objective,
be this fatigue analysis, modes of vibration, buckling of the structure References
or forces. It is up to the designer to arrive to the number of param-
eters required to obtain a more reliable design. [1] The European offshore wind industry – key trends and statistics. 2014. Available
from: http://www.ewea.org/fileadmin/files/library/publications/statistics/EW
• To date, little static analysis has been conducted on towers. The few EA-European-Offshore%20Statistics-2014.pdf.
studies that have been conducted focus on the resistance of the [2] Wenyi L. Design and kinetic analysis of wind turbine blade-hub-tower coupled
analyzed materials. Dynamic analysis has tended to focus on the system. Renew Energy 2016;94:547–57.
[3] Yao H, Wen-Fang W, Yung-Chang C. Reability analysis of wind turbine towers.
modes of vibration in order to understand the dynamic response of
Procedia Engineering 2018;79:218–24.
structures under the various loads that it supports. On the other [4] Gencturk B, Attar A, Tort C. Optimal design of lattice wind turbine towers. WCEE;
hand, non-linear modelling in the design would be of great help 2012.
increasing the understanding of the structural phenomena of the [5] Oriorl P. Towars the sustainability in the design of wind towers. Energy Procedia
2017;115:41–9.
towers. However, few studies have undertaken this due to the [6] Mildford R. Annual maximum wind speeds from parent distribution function.
complexity involved. J Wind Eng Aerodyn 1987;25:163–78.

15
E. Hernandez-Estrada et al. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx

[7] Cook NJ. Towards better estimation of extreme winds. J Wind Eng Ind Aerod [44] Gkantou M, Vazquez P, Baniotopoulos C. On the structural response of a tall
1982;9:295–323. hybrid onshore wind turbine tower. Procedia Engineering 2017;199:3200–5.
[8] Harris RI. Extreme value analysis of epoch maxima convergence, and choice of [45] Enzo M, Alessandro G, Lance M. Offshore wind turbine fatigue loads: the
asymptote. J Wind Eng Ind Aerod 2004;92:897–918. influence of alternative wave modeling for different turbulent and mean winds.
[9] Chiodi R, Ricciardelli F. Three issues concerning the statistics of mean and Renew Energy 2016:1–38.
extreme wind speeds. J Wind Eng Ind Aerod 2014;125:156–67. [46] Axisa R, Muscat M, Sant T, Farrugia R. Structural assessment of a lattice tower for
[10] Zihua Z, Junhua L, Ping Z. Failure analysis of large-scale wind power structure a small, multi-bladed wind turbine. Int J Energy Environ Eng 2017;8:343–58.
under simulated typhoon. Math Probl Eng 2014:1–10. [47] Xiaofeng D, Jijian L, Haijun W, Tongshun Y, Yue Z. Structural vibration
[11] Fisher RA, Tippet LHC. Limiting forms of the frequency distribution of the largest monitoring and operational modal analysis of offshore wind turbine structure.
or smallest members of a sample. Proc Camb Phil Soc 1928;24:180–90. Ocean Eng 2018;150:280–97.
[12] Palutikof JP. A review of methods to calculate extreme wind speeds. Meteorol [48] Feliciano J, Cortina G, Spear A, Calaf M. Generalized analytical displacement
Appl 1999;6:119–32. model for wind turbine towers under aerodynamic loading. J Wind Eng Ind Aerod
[13] Ingrid FL, Vivian BP. Análisis de métodos de vientos extremos para calcular las 2018;176:120–30.
velocidades básicas. RCI 2016;7:15–25. [49] Adhikari RC, Wood DH, Sudak L. Low-cost bamboo lattice towers for small wind
[14] Mayne JR. The estimation of extreme winds. J Wind Eng Ind Aerod 1979;5: turbines. Energy for Sustain Dev 2015;28:21–8.
109–37. [50] Ying-di L, Meng-yuan G, Na W. Dynamic responses of K-type and inverted-K-type
[15] Nilanjan S, Zhen G, Torgeir M, Arvid N. Short-term extreme response analysis of a jacket support structures for offshore wind-turbines. J. Cent. South Univ. 2017;
jacket supporting an offshore wind turbine. Wind Energy 2014;17:87–104. 24:947–56.
[16] Anup KC, Jonathan W, Tania U. Urban wind conditions and small wind turbines [51] Ma HW, Meng R. Optimization design of prestressed concrete wind-turbine tower.
in the built environment: a review. Renew Energy 2019;131:268–83. Technol Sci 2014;57:414–22.
[17] Juan S. Comparison of methods for determining Weibull parameters for wind [52] Fatih K. A virtual tool for minimum cost design of a wind turbine tower with ring
energy generation. Sci Tech 2013;18:315–20. stiffeners. Energies 2013;6(8):3822–40.
[18] Ranganathan CR, Ramanathan M, Swaminathan KR. Estimation of wind power [53] Gustavo O, Filipe M, Alvaro C, Elsa C. Dynamic monitoring system for utility-
availability in Tamil Nadu. Renew Energy 1991;1:429–34. scale wind turbines: damage detection and fatigue assessment. J Civil Struct
[19] Seguro JV, Lambert TW. Modern estimation of the parameters of the Weibull Health Monit 2017;7:657–68.
wind speed distribution for wind energy analysis. J Wind Eng Ind Aerod 2000;85: [54] Jun-Ling C, Christos TG. Spherical tuned liquid damper for vibration control in
75–84. wind turbines. J Vib Contr 2013:1–11.
[20] Costa PA, Costa R, Coelho de Sousa C. Comparison of seven numerical methods [55] Begum YD, Yesim T, Umit G. Evaluation of offshore wind turbine tower dynamics
for determining Weibull parameters for wind energy generation in the northeast with numerical analysis. Adv Civ Eng 2018:1–11.
region of Brazil. Appl Energy 2012;89:395–400. [56] Hu Yinlong, Michael ZQ, Chanying L. Active structural control for load mitigation
[21] Dorvlo Atsu SS. Estimating wind speed distribution. Energy Convers Manag 2002; of wind turbines via adaptive sliding-mode approach. J Franklin Inst 2017;354:
43:2311–8. 4311–30.
[22] Feyzollahzadeh M, Mahmoodi MJ, Yadavar-Nikravesh SM, Jamali J. Wind load [57] Ali K, Sepehr S. Optimal design of jacket supporting structures for offshore wind
response of offshore wind turbine towers with fixed monopile platform. J Wind turbines using CBO and ECBO algorithms. Period Polytech Civ Eng 2017:1–10.
Eng Ind Aerod 2016;158:122–38. [58] Alberto Á, Rodolfo S, Georges G. Evaluación de fuerzas inducidas por oleaje en
[23] Robert F, Bettina H, Frank K, Markus F. Aerodynamic propeties of wind turbine sistemas de pilas cilíndricas. Ing Hidraul Mex 2004:83–94.
towers based on wind tunnel experiments. Procedia Engineering 2017;199: [59] Begüm Y, Muhammed E, Ümit G. Behaviour of large cylindrical offshore
3121–6. structures subjected to wave loads. TEM J 2017;6:550–7.
[24] Monin A, Obukhov A. Basic laws of turbulent mixing in the ground layer of the [60] Rahman M, Heaps HS. Wave forces on the structures: No linear wave diffractions
atmosphere. Trud. Geofiz. Inst. Akad. Nauk. SSSR 1954;24:163–87. by large cylinders. J Phis Oceanogr 1983;13:2225–35.
[25] Sathe A, Erik G, Alfredo P. Comparison of the atmospheric stability and wind [61] Rahman M. Wave diffraction by large offshore structures: an exact second-order
profiles at two wind farm sites over a long marine fetch in the North Sea. Wind theory. Appl Ocean Res 1984;6:90–100.
Energy 2012;14. [62] Shane C, Biswajit B. Tuned liquid column dampers in offshore wind turbines for
[26] Rozenn W, Hans EJ, Uwe SP. Remote sensing used for power curves. IOP Conf Ser structural control. Eng Struct 2009;31:358–68.
Earth Environ Sci 2008;1:1–9. [63] Xiong L, Cheng L, Gangqiang L, Ajit G, Yan C. Effects of aerodynamic damping on
[27] International standard IEC 61400-1 wind turbines – Part 1: design requirements. the tower load of offshore horizontal axis wind turbines. Appl Energy 2017;204:
[28] International standard IEC 61400-2 wind turbines – Part 2: small wind turbines. 1101–14.
[29] Reinhard H, Gideon PAG. Structural stability of concrete wind turbines and solar [64] Abhinav KA, Saha N. Dynamic analysis of monopile supported offshore wind
chimney towers exposed to dynamic wind action. J Wind Eng Ind Aerod 2007: turbines. Proc Inst Civ Eng 2017:1–17.
1079–96. [65] MacCamy RC, Fuchs RA. Wave forces on piles: a diffraction theory. Tech Memo
[30] Nikos DL, Matthew GK. Life-cycle cost structural design optimization of steel 1954;69:17.
wind towers. Comput Struct 2015:1–11. [66] Havelock TH. The pressure of water waves on fixed obstacle. Proc Roy Soc Lond
[31] José Manuel GD. Quantification of the profile wind up 100m from surface and its 1940;175:409–21.
incidence in air climatology. Terra 2013;29. [67] Bazeos N, Hatzigeorgiou GD, Hondros ID, Karamaneas H, Karabalis D, Beskos D.
[32] Hsu S. Determination of the power law wind profile exponent on a tropical cost. Static, seismic and stability analyses of a prototype wind turbine steel tower. Eng
J Appl Meteorol 1982;21. Struct 2002;24:1015–25.
[33] Det Norske Veritas. Environmental conditions and environmental loads. October [68] Lasse M, Pertti L, Mika H. Determining ice loads for tower structure design. Eng
2010. Struct 2014;74:229–32.
[34] Zendehbad M, Chokani N, Abhari RS. Measurements of tower deflections on full- [69] International Standard ISO 12494:2017-Atmospheric icing of structures.
scale wind turbines using an opto-mechanical platform. J Wind Eng Ind Aerod [70] Triantafyllos KM, Evangelos E. On the torsional–translational response of wind
2017;168:72–80. turbine structures. Arabian J Sci Eng 2016;41:1241–9.
[35] Piguang W, Mi Z, Xiuli D, Jingbo L, Chengshun X. Wind, wave and earthquake [71] Junling C, Guanyuan Z, Yong Z. Application of spherical tuned liquid damper in
responses of offshore wind turbine on monopile foundation in clay. Soil Dynam vibration control of wind turbine due to earthquake excitations. Struct Des Tall
Earthq Eng 2018;113:47–57. Special Build 2016;25:431–43.
[36] Hongwang M, Jun Y, Longzhu C. Effect of scour on the structural response of an [72] Tao H, Lewei T, Yunfeng Z. Dynamic response analysis of wind turbine tubular
offshore wind turbinesupported on tripod foundation. Appl Ocean Res 2018;73: towers under long-period ground motions with the consideration of soil-structure
179–89. interaction. Adv Steel Construct 2018;14:227–50.
[37] Jui-Sheng C, Yu-Chen O, Kuan-Yu L, Zhi-Jia W. Structural failure simulation of [73] Philipp M, Christoph B, Sven K. Pile-grid foundations of onshore wind turbines
onshore wind turbines impacted by strong winds. Eng Struct 2018;162:257–69. considering soil-structure interaction under seismic loading. Soil Dynam Earthq
[38] Gholamreza A, Sungmoon J. Along-wind buffeting responses of wind turbines Eng 2018;109:299–311.
subjected to hurricanes considering unsteady aerodynamics of the tower. Eng [74] Gazetas G. Analysis of machine foundation vibrations: state of the art. Soil Dynam
Struct 2017;138:337–50. Earthq Eng 1983;2(1):2–42.
[39] Gholamreza A, Sungmoon J. Buffeting response analysis of offshore wind turbines [75] Kausel E. Early history of soil-structure interaction. Soil Dynam Earthq Eng 2010;
subjected to hurricanes. Ocean Eng 2017;141:1–11. 30(9):822–32.
[40] Zhenyu W, Wei T, Hui H. A Comparative study on the aeromechanic [76] Wolf JP. Foundation VIbration analyses using simple physical models. New
performances of upwind and downwind horizontal-axis wind turbines. Energy Jersey: PTR Prentice Hall; 1994.
Convers Manag 2018;163:100–10. [77] Andersen L. Assessment of lumped-parameter models for rigid footings. Comput
[41] Jorge LL, Claudia P. Vientos extremos en la provincia de Neuquén. Univ. Nacl. Struct 2010;88(23):1333–47.
Comahue 2011;36:83–93. [78] Taddei F, Butenweg C, Klinkel S. Parametric investigation of the soil-structure
[42] Lin W, Xiongwei L, Athanasios K. State of the art in the aeroelasticity of wind interaction effects on the dynamic behavior of a wind turbine considering a
turbine blades: aeroelastic modelling. Renew Sustain Energy Rev 2016;64: layered soil. Hoboken: Wiley-Blackwell; 2014.
195–210. [79] Santangelo F, Giuseppe F, Santini A, Arena F. Time-domain uncoupled analyses
[43] Redeemer A, Martin M, Tonio S, Robert NF. Structural assessment of a lattice for seismic assessment of land-based wind turbines. Eng Struct 2016;123:275–99.
tower for a small, multi-bladed wind turbine. Int J Energy Environ Eng 2017;8: [80] Magnus C, Mohamed S, Christos T, Francis Q. Structural integrity monitoring of
343–58. onshore wind turbine concrete foundations. Renew Energy 2015;83:1131–8.

16
E. Hernandez-Estrada et al. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx

[81] Gustavo O, Filipe M, Álvaro C, Elsa C. Continuous dynamic monitoring of an [111] Kamel Sayed AK, Ghada NS, Boshra AA, Mostafa M. Seismic response of a full-
onshore wind turbine. Eng Struct 2018;164:22–39. scale wind turbine tower using experimental and numerical modal analysis. Int J
[82] Jack M, Marcus P, Grzegorz F, Pawel N, Iain M, Tim R. Deterioration of cracks in Adv Struct Eng 2016;8:337–49.
onshore wind turbine foundations. Eng Struct 2018;167:121–31. [112] Bum-Joon K, Pasin P, Ki-Du K, Hyun-Gi K. Three-dimensional analysis of
[83] Ki-Yong O, Woochul N, Moo SR, Ji-Young K, Bogdan IE. A review of foundations prestressed concrete offshore wind turbine structure under environmental and 5-
of offshore wind energy convertors: current status and future perspectives. Renew MW turbine loads. J Mar Sci Appl 2018;17:625–37.
Sustain Energy Rev 2018;88:16–36. [113] Jonkman J, Butterfield S, Musial W, Scott G. Definition of a 5-MW Reference
[84] Musial W, Ram B. Large-scale offshore wind power in the United States – Wind Turbine for Offshore System Development. National Renewable Energy
assessment of opportunities and barriers. NREL; 2010. Laboratory. 2009. Technical Report NREL/TP-500-38060.
[85] Xiaoni W, Yu H, Ye Li. Foundations of offshore wind turbines: a review. Renew [114] Ishihara T, Phuc PV, Fujino Y, Takahara K, Mekaru T. A field test and full dynamic
Sustain Energy Rev 2019;104:379–93. simulation on a stall regulated wind turbine. Proceedings of 6th asia-pacific
[86] Atul P, Sungmoon J, Oh-Sung K. Structural performance of a parked wind turbine conference on wind engineering. 2005. p. 599–612.
tower subjected to strong ground motions. Eng Struct 2016;120:92–102. [115] Nan X, Takeshi I. Analytical formulae for wind turbine tower loading in the
[87] Patrick M, Nathalie G. Gravity base foundations for the wind turbines on the parked condition by using quasi-steady analysis. Wind Eng 2014;38(3):291–309.
thorntonbank – Belgium. Innov Geotech 2008:1–18. [116] Yung-Yen K. A simplified structural model for monopile-supported offshore wind
[88] Sungmoon J, Sung-Ryul K, Atul P, Le Chi H. Effect of monopile foundation turbines with tapered towers. Renew Energy 2020;156:777–90.
modeling on the structural response of a 5-MW offshore wind turbine tower. [117] Mohammad-Amin A, William S, Jeffery V. Fragility analysis of a 5-MW NREL
Ocean Eng 2015;109:479–88. wind turbine considering aero-elastic and seismic interaction using finite element
[89] Haoran L, Zhiqiang H, Jin W, Xiangyin M. Short-term fatigue analysis for tower method. Finite Elem Anal Des 2016;120:57–67.
base of a spar-type wind turbine under stochastic wind-wave loads. Int J of Naval [118] Matthew AL, Mario AR. Passive structural control of offshore wind turbines. Wind
Arch and Ocean Eng 2018;10:9–20. Energy 2011;14:373–88.
[90] Signe S, Henrik B, Harry BB. The influence of fully nonlinear wave forces on aero- [119] Xin T, Xiaowei Z, Shi Z. Load reduction of a monopile wind turbine tower using
hydro-elastic calculations of monopile wind turbines. Mar Struct 2016;50: optimal tuned mass dampers. Int J Contr 2015:1–16.
162–88. [120] Umesh KN, Bharath P, Mohamed F. Design and analysis of 2 MW wind turbine
[91] Yun WC, Ji-hoon S, Young-Nam K, Jeong-min G, Young-ho K. Numerical studies tower. Int J Mech Prod Eng 2016:13–7.
on piled gravity base foundation for offshore wind turbine. Mar Georesour [121] Trung Q, John W, Hussam M. Fatigue life fragilities and performance-based
Geotechnol 2015:1–12. design of wind turbine tower base connections. J Struct Eng 2015:1–13.
[92] Yeon-Seung L, Jose AG, Ji Hyun L, Young Il K, K CP, Soonhung H. Structural [122] Saleh Y, Ahmad S. Joint slip investigation based on finite element modelling
topology optimization of the transition piece for an offshore wind turbine with verified by experimental results on wind turbine lattice towers. Front Struct Civ
jacket foundation. Renew Energy 2016;85:1214–25. Eng 2018;12(3):341–51.
[93] John MH, Erin EB. A semi-analytical frequency domain model for efficient design [123] Zhiyu J. The impact of a passive tuned mass damper on offshore single-blade
evaluation of spar floating wind turbines. Mar Struct 2019;64:186–210. installation. J Wind Eng Ind Aerod 2018;176:65–77.
[94] Adhikaria S, Bhattacharya S. Vibration of wind-turbines considering soil-structure [124] Mark LB, Ann-Sofie B, Jan H. Active tuned mass damper for damping of offshore
interaction. Wind Struct 2011;14(2):85–112. wind turbine vibrations. Wind Energy 2016:1–14.
[95] Harte M, Basu B, Nielsen SRK. Dynamic analysis of wind turbines including [125] Matthew SA, Michael WS, Shashank C, Morten HH. Output-only modal analysis of
soilstructure interaction. Eng Struct 2012;45:509–18. linear time-periodic systems with application towind turbine simulation data.
[96] Sapountzakis EJ, Dikaros IC, Kampitsis AE, Koroneou AD. Nonlinear response of Mech Syst Signal Process 2011;25:1174–91.
wind turbines under wind and seismic excitations with soil-structure interaction. [126] Mark LB, Jan H. Hybrid damper with stroke amplification for damping of offshore
J Comput Nonlinear Dynam 2016;10. wind turbines. Wind Energy 2016:1–16.
[97] Recommended Practice for Planning. Designing and constructing fixed offshore [127] Amir RG, Oya M, Reza K. Seismic soil–structure interaction analysis of wind
platforms-working stress design. API; 2003. turbines in frequency domain. Wind Energy 2016:1–18.
[98] Bisoi S, Haldar S. Dynamic analysis of offshore wind turbine in clay considering [128] Paweł M. Vibration control of wind turbine tower-nacelle model with
soil-monopile-tower interaction. Soil Dynam Earthq Eng 2014;63:19–35. magnetorheological. J Vib Contr 2015:1–22.
[99] Gerolymos N, Gazetas G. Winkler model for lateral response of rigid caisson [129] Jianyun C, Chenyang Y, Jing L, Qiang X. Semi-active fuzzy control of edge wise
foundations in linear soil. Soil Dynam Earthq Eng 2006;26:347–61. vibrations in wind turbine blades under extreme wind. J Wind Eng Ind Aerod
[100] Gerolymos N, Gazetas G. Development of Winkler model for static and dynamic 2015;147:251–61.
response of caisson foundations with soil and interface nonlinearities. Soil Dynam [130] Jonkman J, Buhl ML. FAST user’s guide. Natl Renew Energy Lab 2005. Technical
Earthq Eng 2006;26:363–76. Report NREL/TP-500-38230.
[101] Pando MA. Analyses of lateral loaded piles with p-y curves – observations on the [131] Ishihara Takeshi, Wang Lilin. A study of modal damping for offshore wind
effect of pile flexural stiffness and cyclic loading. Proceedings of the geo 3 T2 turbines considering soil properties and foundation types. Wind Energy 2019;22
conference. 2013. 3B-1-A49. (12):1760–78.
[102] Ti KS, Huat BBK, Noorzaei J, Jaafar MS, Sew GS. A review of basic soil [132] Kim W, Yi J, Kim J, Park J. Vibration-based structural health assessment of wind
constitutive models for geotechnical application. Electron J Geotech Eng 2009;14: turbine tower using a wind turbine model. Procedia Engineering 2017;188:333–9.
1–18. [133] Saleh Y, Ahmad S. Joint slip investigation based on finite element modelling
[103] Schanz T, Vermeer PA, Bonnier PG. The hardening soil model: formulation and verified by experimental results on wind turbine lattice towers. Front Struct Civ
verification. Beyond 2000 Comput Geotech 1999:281–96. Eng 2018;12(3):341–51.
[104] Haldar S, Babu GLS. Effect of soil spatial variability on the response of laterally [134] Junling C, Rongchang Y, Renle M, Jinwei L. Design optimization of wind turbine
loaded pile in undrained clay. Comput Geotech 2008;35(4):537–47. tower with lattice-tubular hybrid structure using particle swarm algorithm Struct.
[105] Acumus M, Thieken K. On the behavior of piles in non-cohesive soil under Des Tall Spec. Build. 2016;25:743–58.
combined horizontal and vertical loading. Acta Geotech 2010;5:199–210. [135] Negm HM, Maalawi KY. Structural design optimization of wind turbine towers.
[106] Abdel-Rahman K, Achmus M. Finite element modelling of horizontally loaded Comput Struct 2000;74(6):649–66.
monopile foundations for offshore wind energy converters in Germany. [136] Clifton-Smith MJ, Wood DH. Optimisation of self-supporting towers for small
Proceedings of the international symposium on frontiers in offshore geotechnics. wind turbines. Wind Eng 2010;34(5):561–78.
Perth: Taylor and Francis; 2005. p. 391–6. [137] Clausen PD, Peterson P, Wilson SVR, Wood DH. Designing an easily-made lattice
[107] John MH, Erin EB. A semi-analytical frequency domain model for efficient design tower for a small wind turbine. Nepal: International Workshop on Small Scale
evaluation of spar floating wind turbines. Mar Struct 2019;64:186–210. Wind Energy for Developing Countries; 2011.
[108] Wood DH. Small wind turbines: analysis, design and application. London: [138] Adhikari RC, Wood DH, Sudak L. Design procedures for tubular lattice towers for
Springer; 2011. small wind turbines. Wind Eng 2014;38:359–76.
[109] International Standard ISO 2394:2015 - general principles on reliability for [139] James AB, Seamus DG. Analysis of force-capping for large wind turbine rotors.
structures. Wind Eng 2015;39:213–28.
[110] Umesh KN, Bharath P, Mohamed F. Design and analysis of 2 MW wind turbine
tower. Int J Mech Prod Eng 2016:13–7.

17