Soil Dynamics and Earthquake Engineering 144 (2021) 106674

Contents lists available at ScienceDirect

Soil Dynamics and Earthquake Engineering

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

Multi-body dynamic analysis of offshore wind turbine considering

soil-structure interaction for fatigue design of monopile
M. Mohamed Sajeer, Arka Mitra, Arunasis Chakraborty *
Department of Civil Engineering, Indian Institute of Technology Guwahati, Assam, 781039, India

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

Keywords: Fatigue design of monopile foundation for an offshore wind turbine using multi-body dynamic analysis is the
Monopile design theme of this paper. For this purpose, a benchmark turbine is modelled in Kane’s approach considering soil-
Offshore wind turbine structure interaction for a site at the North Sea. Each soil layer is modelled using the p-y curve, and the dy­
Fatigue life estimation
namic response of the combined system is simulated using aero-hydro-servo-elastic analysis. The fatigue life is
Serviceability analysis
Soil-pile interaction
estimated following IEC 61400-3 guidelines. The numerical analysis presented in this study shows stress-based
Bi-axial stress fatigue analysis using rainflow algorithm in the light of soil-structure interaction for short-term damage accu­
Palmgren-Miners summation mulation. Besides short-term effects, the study also investigates serviceability criteria over the lifespan of the
structure, i.e. long-term effects. Finally, design curves are developed for a given power rating. Overall, the study
highlights the importance of soil-structure interaction and its impact on the fatigue life of offshore wind turbines.

1. Introduction affect the structure. Bisoi and Haldar [4] studied soft-soft and soft-stiff
monopile design and checked the overall safety against serviceability
Offshore wind farms have gained popularity in the recent past due to and fatigue limit states. They observed that the length of monopile
its various advantages, e.g. steady wind flow over an extended period below its critical depth had a negligible impact on its design.
and vast space in the marine environment. Many of these turbines in Rong et al. [5] derived an analytical formula for the natural fre­
shallow water (depth around 30∼50 m) are supported by monopiles, quency of monopile supported wind turbine using Euler-Bernoulli beam
whose dimensions depend on the size and power rating of the turbines, model with p-y curves for foundation stiffness. In this context, different
besides other environmental factors. The design of these structures is aspects of offshore foundation design may be found in Bhattacharya [6].
challenging and requires a high-fidelity model of the complete multi- This book explains the importance of soil-structure interaction and
body rotating system, where soil-structure interaction (SSI) plays a multiple methods to evaluate the foundation stiffness. It also provides a
vital role. comprehensive review of different methods used for long-term behav­
Various methods are proposed in the literature to model the soil- iour estimation. Banerjee et al. [7] modelled the effect of foundation
structure interaction. For example, a closed-form expression for flexibility using an equivalent spring-dashpot model for monopile below
finding the fundamental frequency of a wind turbine tower under SSI the mudline. They carried out dynamic analysis using four different soil
was developed and experimentally validated by Adhikari and Bhatta­ conditions and observed that the total displacement at the tower-top
charya [1]. Finite element analysis of monopile supported offshore wind increased with reduced shear wave velocity, especially for lower soil
turbine is a popular approach, where soil properties are modelled by stiffness. Kementzetzidis et al. [8] studied different geotechnical aspects
linear or non-linear springs [2]. The dynamic interaction results in an of monopile foundation using 3D non-linear finite element analysis with
increased response of the combined system. Zuo et al. [3] modelled a significant emphasis on the robustness of soft-stiff design, particularly
NREL 5 MW benchmark wind turbine in the finite element framework in the extreme weather. In this context, impact load test on pile [9,10] is
using ABAQUS and observed a significant reduction of the tower natural a standard tool to estimate its stiffness and in-situ frequency response
frequencies due to SSI. They also noticed that blade and tower responses function for finite element modelling.
increased with rotor speed. Thus, monopile design without considering Over the last decade, offshore wind turbines witnessed significant
blade rotation led to gross underestimation that could subsequently growth in its size to meet the power demand. Due to this reason, the

* Corresponding author.
E-mail address: [email protected] (A. Chakraborty).

https://doi.org/10.1016/j.soildyn.2021.106674
Received 19 June 2020; Received in revised form 28 January 2021; Accepted 15 February 2021
Available online 4 March 2021
0267-7261/© 2021 Elsevier Ltd. All rights reserved.
M.M. Sajeer et al. Soil Dynamics and Earthquake Engineering 144 (2021) 106674

load-carrying capacity of monopile has to increase to accommodate springs, where the turbines and blades are lumped at the tower-top and
more giant turbines. This requirement has led to developing the new (ii) dynamic analysis using software, e.g. FAST [28], Bladed [29],
methodologies to improve the bearing capacity using stiffener to avoid 3DFloat [30], HAWC2 [31]. Since monopile fatigue analysis depends on
the abnormal size of monopile [11]. In general, fatigue design focuses on the stress reversal, the first category models have inherent limitations as
the short-term effects, while long-term effects are equally crucial for they do not consider blade dynamics. Although blade and drivetrain
serviceability. Katsikogiannis et al. [12] studied the stress due to axial dynamics are modelled in the software, they have scope for further
force and bi-directional bending moments for damage estimation. They improvements for more accurate modelling with SSI in the multi-body
compared the sensitivity of three soil modelling techniques (viz. dynamic framework (e.g. the SubDyn module [32] in FAST uses
macro-element model, linear elastic stiffness model with soil damping soft-coupling). Thus, the objectives of this work are -
and p-y curves) under wind-wave misalignment. Their study concluded
that significant fatigue damage might occur for misalignment greater 1. Develop a multi-body dynamic model of a monopile supported
than 30◦ . offshore wind turbine considering soil-structure interaction, where
In general, soil type around the monopile affects its design life. all the states are integrated as a tightly coupled system. Multiple soil
Schafhirt et al. [13] investigated this influence on long-term perfor­ layers are proposed to be characterised by their respective non-linear
mance due to cyclic loading. They modelled soil-structure interaction by load-deformation curves. The dynamic equilibrium equations are
non-linear p-y curves to estimate the fatigue life using the mudline developed following Kane’s approach, where Winkler’s beam
bending moment, which was affected by the softening of the sur­ formulation models the soil-pile interaction.
rounding soil. Besides spring stiffness, aero-elastic and hydrodynamic 2. Investigate the impact of generalised stress-based formulation for
damping play a significant role in fatigue life and cost-effective design fatigue design using the 3D multi-body dynamic model and compare
[14,15]. In this context, current industry standards are often inadequate it with a moment-based approach. For this purpose, estimate the
[16] for the upcoming large turbines due to their simplified approach for longitudinal stress at different locations from the bi-directional
stiffness and damping characterisation and wind-wave modelling in the bending moment and axial force acting on the monopile foundation.
light of misalignment and soil-structure interaction. These shortcomings 3. Once the stress-based cycle counts are ready, the short-term damage
often lead to underestimation of fatigue damage accumulation and in each load cases and its accumulation over the design life are
subsequent design life of the foundation. Marino et al. [17] compared estimated to find the fatigue life. The damage-equivalent stress is
the linear and non-linear wave models for fatigue load estimation. They proposed to be evaluated, which is useful for the load test of
concluded that the linear wave model underestimated fatigue load in the monopiles.
non-operating (or parked) condition, where hydrodynamic loads pre­ 4. Finally, the long-term serviceability of the combined system at the
dominate. Simultaneously, the non-linear wave model was less critical mudline is proposed to be investigated, followed by the sensitivity
in normal operating conditions, where aerodynamic loads dictated the analysis for various parameters, e.g. diameter, thickness, and water
design. Risi et al. [18] investigated the performance of offshore wind depth. The study suggests the site-specific design curves that help the
turbine under extreme events, e.g. earthquake along with wind and planning and foundation design of similar systems.
wave loads. In their study, a finite element model of an offshore wind
turbine was developed for seismic fragility analysis to investigate its 3. Multi-body dynamics of an offshore wind turbine on
vulnerability, particularly under crustal earthquakes in soft soils. Loken monopile
and Kaynia [19] estimated the fatigue life of monopile and caisson
foundation of offshore wind turbine using the mudline bending moment, This section presents the modelling of a three-bladed horizontal axis
which concluded that fixed bottom boundary condition over-estimated offshore wind turbine, as shown in Fig. 1 supported over monopile
the design length. foundation.
Monopiles in the marine environment are often subjected to stress
reversal caused by the environmental loads. Thus, fatigue crack propa­ 3.1. Wind turbine model
gation in these structures is the subject of various studies [20,21], where
residual stress has a significant role in the performance of welds and In this section, Kane’s approach [33] is adopted to model a monopile
connections. In this context, structural loads in the marine environment supported turbine in a multi-body dynamic framework, which has sig­
are highly uncertain. Hence, stochastic analysis of monopile has nificant advantages in accuracy and computational efficiency. The
remained a significant area of research, where the overall reliability complete turbine is modelled as a 16◦ of freedom (DOFs) system,
index is often quantified by approximate models [22] or Monte Carlo including two flapwise and one edgewise modes for each blade and two
simulation [23]. Haldar et al. [24] studied the stochastic response of modes in fore-aft and side-to-side directions for monopile-tower. How­
monopile supported offshore wind turbine in clay using non-linear p-y ever, this formulation is general and can include as many modes
curves. They observed that the increase of embedded length reduced the required for a particular case. The details of these DOFs are given in
mean of maximum tilt at the mudline, while marginally influencing its Table 1, which takes the following vector form
fatigue life. Teixeira et al. [25] used kriging for stress-cycle analysis of ⎧ ⎫T
offshore wind turbine at a reduced computational cost without ⎪
⎨ ⎪
⎬
compromising its accuracy. Velarde et al. [26] conducted a sensitivity u = u1 …u9 , u10 …u13 , u14 , u15 , u16 (1)
analysis of fatigue loads due to uncertainty in environmental, structural ⎩⏟̅̅̅̅⏞⏞̅̅̅̅⏟ ⏟̅̅̅̅̅⏞⏞̅̅̅̅̅⏟
⎪ ⏟⏞⏞⏟ ⏟̅̅̅⏞⏞̅̅̅⏟⎪⎭
Blades Monopile− Tower Nacelle Drivetrain
and geotechnical parameters. They observed that the uncertainties in
turbulence intensity and wave load had a significant influence on fatigue Fig. 1 shows the global inertial reference frame marked by ̂ x along
loads. A comprehensive review of stochastic fatigue damage analysis for with all local coordinates, whose details are given in Table 1. In this
offshore wind turbine may be found in Jimenez-Martinez [27]. context, each coordinate system has three orthogonal components,
which are marked by subscripts 1, 2 and 3 in Fig. 1. The blade coordinate
2. Problem formulation systems have additional subscript j (e.g. ̂g c,j ), which denotes the blade
number, i.e. j = 1, 2, 3. The coordinate system ̃ ac,j in Table 1 is used for
The above literature review clearly shows the importance of fatigue aero-elastic load estimation, as shown in Fig. 3a. Appropriate trans­
design and long-term behaviour analysis of monopile supported offshore formation matrices are used to establish the relationships between each
wind turbines. These studies use models which can be broadly classified coordinate system [33].
into two major subgroups - (i) static or quasi-static analysis with soil Consequently, the governing equations of motion for n DOF system

2
M.M. Sajeer et al. Soil Dynamics and Earthquake Engineering 144 (2021) 106674

Fig. 1. Schematic diagram of offshore HAWT and coordinate system of (a) wind turbine, (b) drivetrain, (c) hub, (d) airfoil, (e) blade element & (f) tower element.

(here n = 16) are derived from Newton’s law for holonomic multi-body system with the centre of mass point at Ck and reference frame at Xk for
system as each body, the generalised active force for each degree of freedom is
given by
Fi + Fi* = 0 i = 1, 2, …, n (2)
Nb
∑ [ Ck C ]
In the above equation, Fi and are the generalised active and inertia
Fi* Fi = vi .F k + ωXi k .M Xk (3)
forces for ith DOF, respectively. Considering an Nb (∕
= n) body dynamic k=1

Here, vCi k is the partial velocity at the centre of mass Ck , while FCk is
Table 1 the corresponding generalised active force on the kth body. Similarly,
Details of degrees of freedom and coordinate systems. MXk and ωXi k in the same equation represent the generalised active
Multi-body Components DOFs Coordinate Remarks moment and partial angular velocity of the kth body with respect to its
Names systems reference frame Xk . Generalised inertial force in Eq. (2) is given by
Blade 1 1st Flap mode u1 ̂f Global
c,1 ∑
Nb
[ ( ) ( )]
2nd Flap mode u2 ̂g c,1 , ̂
h c,1 , ̂i c,1 Cone, Pitch, Fi* = − vCi k . mk aCk + ωXi k . I k .αXk + ωXk × I k × ωXk (4)
Twist k=1
1 Edge mode
st u3 bc,1 , ̃
̃ ac,1 Element fixed
system where mk and Ik are the mass and generalised moment of inertia of the
Blade 2 1st Flap mode u4 ̂f Global
c,2 kth rigid body and aCk and αXk are the generalised linear and angular
2nd Flap mode u5 ̂g c,2 , ̂
h c,2 , ̂i c,2 Cone, Pitch, acceleration with respect to Ck . Total linear and angular velocities of the
Twist
1st Edge mode u6 Element fixed kth body are expressed as a summation of the partial velocities, which
bc,2 , ̃
̃ ac,2
system are given by
Blade 3 st
1 Flap mode u7 ̂f Global
c,3 ∑
n
2nd Flap mode u8 ̂g c,3 , ̂
h c,3 , ̂i c,3 Cone, Pitch, v Ck = vCi k .qi + vCt k (5a)
Twist i=1
1st Edge mode u9 bc,3 , ̃
̃ ac,3 Element fixed
system ∑
n

Monopile- 1st Fore-aft mode u10 ac

̂ Global ωXk = ωXi k .qi + ωXt k (5b)
tower i=1
1st Side-to-side u11 bc
̂ tower-top
mode In the above two equations, vCi k .qi and ωXi k .qi are the components of
2nd Fore-aft u12 bc Element fixed the total velocities, expressed in linear superposition form. At the same
mode system time, vCt k and ωXt k are the components of the respective total velocities
nd
2 Side-to-side u13 that do not follow the same format. Kane and Levinson [33] described qi
mode as the generalised speed, which in this study, is assumed to be the first
Nacelle Nacelle yaw u14 ̂c c Yaw differential of ui with respect to time, i.e. qi = u̇i . In this formulation, vCt k
Drivetrain Shaft torsional u15 Slow speed shaft
dc
̂
and ωXt k become zero, which lead to the total linear and angular accel­
flexibility tilt and skew
Generator u16 ̂e c rotating slow erations of the kth body in the following form
azimuth speed shaft

N.B.: subscript c stands for three orthogonal components, i.e. c = 1, 2, 3.

3
M.M. Sajeer et al. Soil Dynamics and Earthquake Engineering 144 (2021) 106674

( )
kin Hs
p = AL pu tanh y (11)
AL p u

where AL is the factor for accounting different loading cases, i.e. AL =(3 −
0.8Hs D−p 1 )≥ 0.9 for static loading and AL = 0.9 for cyclic loading. The
parameter kin denotes the initial modulus of the subgrade reaction, and y
Once the displacement, velocity and acceleration for each compo­ is the lateral deformation of the monopile. The angle of internal friction
nent are defined, the governing equation of motion can be derived in the (φ) of the sand dictates the initial modulus of subgrade reaction. The
global reference frame, which is described as parameter pu in Eq. (11) is the ultimate bearing capacity of soil at a
depth Hs , which is estimated from the failure mechanism of soil, i.e.
∑
3
wedge failure and/or flow failure. The minimum of ultimate bearing
Fi* = Fi,tw
* *
+ Fi,na *
+ Fi,hu *
+ Fi,gn + *
Fi,b (7)
j capacity of the soil is evaluated, which is given by
j=1
{( )
In the above equation, the first subscript i correspond to the DOF, pu =
C1 Hs + C2 Dp γHs wedge failure
(12)
while second subscripts tw, na, hu, gn and bj represent monopile-tower, C3 Dp γHs flow failure
nacelle, hub, generator and jth blade, respectively. Each component of In these equations, Dp and Hs denote the diameter of monopile and
external conservative and non-conservative forces, as well as the forces height of the soil strata, while γ represents the effective soil weight. The
arising from the interaction of the multi-body systems and their con­ coefficients C1 , C2 and C3 are the functions of internal friction of soil (φ),
strained relations, contribute to the generalised active force, which can which are evaluated using the following expressions [37].
be expressed as
C1 = C0 Ca− 1/2 tan φ sin βcos− 1 α + Ca− 1/2 tan2 β tan α
∑
3 ∑
3
Fi = Fi,tw,hy + Fi,tw,ae + Fi,bj ,ae + Fi,tw,gv + Fi,na,gv + Fi,hu,gv + Fi,bj ,gv + Fi,ys + C0 tanβ [tan φ sinβ − tanα] (13a)
j=1 j=1
∑
3 ∑
3
C2 = Ca− 1/2 tanβ − Ca (13b)
+Fi,gn +Fi,br +Fi,tw,el + Fi,bj ,el +Fi,dt,el +Fi,tw,d + Fi,bj ,d +Fi,dt,d
j=1 j=1 ( )
C3 = 0.4 tan φtan4 β + Ca tan8 β − 1 (13c)
(8)
In the above equations, the subscripts hy, ae, gv, ys, br, el, dt and where α = 0.5φ and β = 45o + α. C0 is the coefficient of earth pressure at
d represent the hydrodynamic, aerodynamic, gravitational, yaw, brake, rest, which is constant (C0 = 0.4), whereas Ca is the coefficient of active
elastic, drivetrain and damping, respectively. Substituting Eq. (7) and earth pressure [Ca = tan2 (45∘ − α)]. Similarly, the soil resistance
Eq. (8) into Eq. (2) and rearranging them in matrix form lead to the against the axial deformation due to skin friction and soil end bearing is
following governing equation of motion modelled using t-z and Q-z springs [36], respectively. The soil stiffness is
{ ( )} evaluated by computing the slope of the line between the two consec­
[M(u, t)]ü + f u, u̇, va , vf , g, t = 0 (9)
utive points on the p-y curve. For the non-linear region, the stiffness is
Eq. (9) reveals that M(u, t) is the time-dependent system matrix and obtained from the deformation after the convergence analysis using a
f(u, u̇, va , vf , g, t) contains stiffness and damping forces besides other modified Newton-Raphson technique at every time instant [38]. The
externally applied loads. The vectors va and vf represent wind and wave equivalent monopile properties are estimated by considering Kx* =
velocities, respectively, while g is the gravitational acceleration. The Keq x* , which is used for the multi-body dynamic analysis, as shown in
complete derivation of the matrices in Eq. (9) for monopile-tower as­ Fig. 2. The modulus of subgrade reaction for the next iteration is eval­
sembly is provided in the Appendix. Derivation for the remaining bodies uated from this soil stiffness and is fed back to the finite element model
(i.e. blade and drivetrain) follow the same procedure and hence, are at every time step.
omitted in this paper to avoid repetition. Here, it is worth mentioning that the p-y curve in the API [36] is also
recommended by DNV [39], and it has gained popularity over the past
decades among the designers. However, the use of API p-y curve for the
3.2. Soil-monopile model offshore wind turbine has been questioned by many researchers. The
reason behind the critics of API p-y curve is due to its formulation based
In this study, the monopile foundation is modelled using finite ele­ cyclic load test on a smaller diameter pile (Dp < 1 m) in the controlled
ments, where Winkler’s approach models the soil layers. For this pur­ environment of the laboratory, which differs from the actual field con­
pose, two noded Euler-Bernoulli beam elements [34] and soil springs dition [40]. Several experiments using centrifuge models have been
[35] are used to quantify the monopile deformation for equivalent proposed in the literature [41,42] to model this issue. Lee et al. [40]
stiffness in multi-body dynamic analysis. Thus, the combined conducted centrifuge model tests on monopiles in dry dense sand and
soil-monopile element stiffness matrix can be expressed as follows proposed a cyclic p-y backbone curve
⎡ ⎤ ⎡ ⎤ ( )− 1
1 y
⎢
12 − 6le − 12 − 6le
⎥ ⎢
156 − 22le 54 13le
⎥ p=y + (14)
kin pu
Ep I p ⎢
⎢ − 6le 4l2e 6le 2l2e ⎥ E l ⎢ − 22le 4l2e −
⎥ s e⎢
13le − 3l2e ⎥
⎥
e
KN = 3 ⎢ ⎥+ ⎢ ⎥ (10)
le ⎢ − 12
⎣
6le 12 6le ⎥ 420 ⎢ 54 − 13le
⎦ ⎣
156 22le ⎥
⎦ where pu is the ultimate soil resistance, which is defined as
− 6le 2l2e 6le 4l2e 13le − 3l2e 22le 4l2e

⏟̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅⏞⏞̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅⏟ ⏟̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅⏞⏞̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅⏟ pu = Ar Kp γHsnr Dp (15)

In the above equation, Ar and nr are the constants obtained from

Monopile stiffness Soil stiffness

where le is the element length, and the subscript N is the element regression analysis, whereas Kp is the Rankine passive earth pressure
number. Ep and Ip are the Young’s modulus and second moment of area coefficient.
of the monopile element, while Es is the soil modulus based on the p-y
curve. The lateral characteristic of sandy soil is modelled as recom­
mended in the API guideline [36], i.e.

4
M.M. Sajeer et al. Soil Dynamics and Earthquake Engineering 144 (2021) 106674

Fig. 2. Flowchart of MBD-SSI model.

Fig. 3. Schematic diagram of (a) 2D airfoil with the velocity and force components & (b) gravitational load acting on the blade.

4. Forces acting on offshore wind turbine Kaimal or von-Karman) [44,45]. In this study, the turbulence is gener­
ated using Kaimal spectrum in TurbSim [46]. The aerodynamic loads are
Offshore wind turbines are exposed to aerodynamic and hydrody­ evaluated by blade element momentum (BEM) theory [47] using this
namic loads in addition to the gravitational field, as shown in Fig. 1. IEC simulated 3D wind field. It is a combination of blade element and mo­
61400–3 [43] recommends different wind flow conditions, including mentum theories. The first one assumes that the force acting on each
operating and non-operating regions for fatigue analysis with six simu­ airfoil solely depends on its aerodynamic coefficients. The second part is
lated time-histories of 10 min each or a single 1-h response time-history. based on the momentum balance of the airflow on either side of the rotor
The details of these loads are briefly described with only the relevant plane. A set of equations for thrust and torque are developed using these
formulations in the following subsections. two theories independently, which are solved iteratively at every time
step.
The relative velocity of the wind flow has two components, as shown
4.1. Aerodynamic load
in Fig. 3 (a), i.e. one along with the flow and another orthogonal to it.
Therefore, the relative flow velocity vrel can be obtained as
The wind load acting on a blade has a mean and a fluctuating
component depending upon the position and orientation. The fluctu­ [ ′
vrel (r, t) = ((1 − a)v∞ )2 + ((1 + a )Ωr)2
]1/2
(16)
ating part is modelled using rotationally sampled spectrum (RSS), which
is derived from the 1D power spectral density of the wind flow (e.g. The relative velocity acts at an angle θa with the rotor plane, which is

5
M.M. Sajeer et al. Soil Dynamics and Earthquake Engineering 144 (2021) 106674

given by water density and CA , CM and CDr are the coefficients for added mass,
[ ] fluid inertia and viscous drag force, respectively. Dp (z) represents the
v∞ (1 − a)
θa = tan− 1
(17) x 2 . In Eq. (20), vfi (z, t) and
diameter of the monopile at a depth z along ̂
Ωr(1 + a′ )
afi (z, t) are the instantaneous fluid velocity and acceleration along ith
In the above equation, parameters a and a are the axial and
′

DOF at a depth z, which are evaluated from the surface velocity (i.e. Ḣw )
tangential flow induction factors, respectively, whose values lie between
and acceleration (i.e. Ḧw ) using linear wave theory, while vpi is the ve­
0 and 1 and v∞ is the instantaneous upstream wind velocity. The
locity of the pile and dz is the differential height objecting the flow. The
traditional solution strategy in iterative framework tries to find a and a
′

fluid diffraction and radiation damping forces obtained from the po­
before estimating aerodynamic loads and often faces numerical insta­
tential flow theory are nominal for the current scenario as the monopile
bility. The algorithm proposed by Ning [48] is used in this study to avoid
is a slender cylindrical object undergoing very small rotations. The
this issue. It is more accurate and computationally efficient. The solution
diameter of the monopile is less compared to the wave-length of deep-
domain is segmented into three regions (i.e. momentum, empirical and
sea waves, and hence the fluid diffraction and radiation damping
propeller break) for estimating the optimal flow angle θa to enhance the
forces are not included in this mathematical modelling.
convergence. In general, BEM theory does not account for vortex
shedding at the tip of the blade. Therefore, it demands tip and hub loss
4.3. Gravitational load
corrections (i.e. Prandtl’s correction factor) along with Glaurt and Buhl
corrections for higher flow induction factors. The normal and tangential
The gravitational force on a blade changes with its rotation, which
forces acting on an annular ring of an Nb bladed turbine are evaluated
can be resolved into two components, as shown in Fig. 3b. Thus, the in-
using the optimal flow angle θa , which are given by
plane bending moment due to gravity changes with its position and
1 reaches its maximum, when the blade is horizontal. The force acting in
δpn = Nb ρa v2rel c(r)[Cl (α)cosθa + Cd (α)sinθa ]δr (18a)
2 the axial direction affects the stiffness of the blade. The gravitational
load acts vertically downward for different structural components (i.e.
1
δpt = Nb ρa v2rel c(r)[Cl (α)sinθa − Cd (α)cosθa ]δr (18b) nacelle, hub, generator and monopile-tower system). For the case of
2 monopile-tower, the generalised active force arising from the gravity is
discussed in Eq. A.17.
where ρa is the density of air and c(r) is the chord length at a radial
distance r from the centre of rotation. The lift and drag coefficients Cl
5. Fatigue analysis of monopile
and Cd in the above equations depend on the angle of attack (α) of the
respective airfoil [49].
Fatigue damage due to cyclic loading often starts at the surface (i.e.
extreme fibre) and eventually grows to the failure. It may occur even
4.2. Hydrodynamic load when the stresses are below the yield strength of the material. Thus,
fatigue damage assessment is essential while designing the structures
Besides aerodynamic loads, wind turbines in the offshore environ­ that are subjected to load reversal. In this study, the fatigue life of
ment are also exposed to hydrodynamic loads. Ocean wave profiles are monopile is estimated using stress at the extreme fibre. Fig. 4 shows the
generated either from the Pierson-Moskowitz (PM) spectrum [43,50] or axial force and bending moment acting on a monopile, where the total
from the Joint-North-Sea-Wave-Project (JONSWAP) spectrum [51] axial stress at any point on the cross-section can be evaluated as follows
using critical wave parameters, i.e. effective depth, fetch length. The Vz Mx xc My yc
details of these spectra are not discussed here as they are available in the σL = + − (21)
Acr Iy Ix
literature. Using the directional spectrum, i.e. S(fi , θj ), the wave
time-history can be evaluated by linear superposition of sinusoids [52], In the above expression, Vz is the axial force in the z-direction, Acr is
which is given by the cross-sectional area. Simultaneously, Mx and My are the bending
moments in the x and y directions, respectively, where the second
∑ moment of areas are Ix and Iy . The total shear stress is evaluated using
∞
Hw (x, y, t) =
j=1 the following expression
∞ √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
∑ ( ) [ ] √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
× 2S fj , θk Δfj Δθk .cos kj x cos(θk ) + kj y sin(θk ) − fj t − εj,k (19) 2 V 2x + V 2y sinβτ Mz Dp
k=1 τ= + (22)
Acr 2J
Hw (x, y, t) is the wave amplitude time-history at a location (x, y) on the
surface in ̂ x coordinate, where εj,k is the uniformly distributed phase
angle between 0 to 2π . Here, the distance x and y are measured along ̂ x1
and ̂ x 3 , respectively. In the above equation, the parameter kj represents
the wave number corresponding to the frequency fj while θk denotes the
direction of wave propagation. Once the wave is simulated, Morisson’s
equation [43] is used for finding the force acting on the monopile, which
is expressed by the following equation
( ) ( )
π Dp (z)2 πDp (z)2
dFi (z, t)|i=10...14 = − CA ρüi (z, t)dz + CM ρafi (z, t)dz
4 4
1 ( ) ⃒ ⃒
+ CDr vfi (z, t) − vpi (z, t) ρDp (z)⃒vfi (z, t) − vpi (z, t)⃒dz
2
(20)
In the above equation, the first term denotes the added mass asso­
ciated with the motion of the pile while the second and third term imply
fluid inertia force and viscous drag force, respectively. Here, ρ is the Fig. 4. Bending moment and axial force of monopile for stress calculation.

6
M.M. Sajeer et al. Soil Dynamics and Earthquake Engineering 144 (2021) 106674

where Vx and Vy are the shear forces in two orthogonal directions, βτ is where f eq and Tj are the frequency of DES and the total time duration of
the angle between the point at which the internal stress is estimated and the jth time-history. The parameter m denotes the slope of the S–N curve.
the total internal shear vector. Mz is the torsional moment, Dp is the pile The lifetime damage-equivalent stress is evaluated from different load
diameter, and J is the polar moment of inertia of the cross-section. Once cases using the following expression
the component stresses are evaluated, von-Mises stress can be estimated ⎡∑∑ ( )m ⎤m1
[18] for failure analysis. fjL nji σae
Fatigue life estimation of a structure involves accurate counting of
ji
⎢ ⎥
(27)
j i
σ LDE = ⎣ ∑ L eq ⎦
stress reversals and corresponding accumulated damage for a given fj f Tj
stress time-history. The finite number of stress reversals is estimated
j

using the rainflow algorithm proposed by Downing and Socie [53]. Once
Once the damage-equivalent stress is estimated, the long-term per­
the number of cycles is estimated from the response time-history, the
formance of the monopile can be analysed to ensure its serviceability.
allowable number of cycle for the monopile material is estimated from
the S–N curve using equivalent alternating stress obtained from
6. Long-term performance
Goodman relation. Usually, the Goodman corrected fatigue life de­
creases with the increase of mean of the applied alternating load.
Long-term offshore wind turbines behaviour is designed by satisfying
Therefore, Goodman relation [54] and the S–N curve must be used to
the serviceability limit state (SLS) as per DNV guideline [39]. In general,
find the maximum number of alternating stress cycles that a material can
the accumulation of rotation (or tilt) and the deflection (or strain) at the
undergo until its fatigue failure. Goodman line is a linear relationship
seabed must be within the allowable limits (i.e. serviceability). The
between the mean stress and alternating stress to estimate the equivalent
long-term strain accumulation is investigated from the lifetime
alternating stress (σ ae
ji ), which can be expressed as stress-based damage-equivalent load. In this context, different methods
σaji (|σ mf | − σ u ) are often prescribed in the literature for finding the accumulated strain
σ ae
ji =
⃒ m⃒
⃒σ ji ⃒ − σu (23) after N cycles from the strain in the first load cycle, which are described
below.
where σ m a th
ji and σ ji are the mean stress and alternating stress of the i cycle
6.1. Power-law
in the jth time-history. Parameter σ mf is the fixed mean stress, while σu is
the ultimate stress of the material. Power-law is based on the pressuremeter test on monopiles in sandy
The S–N curve for steel is selected based on DNV–OS–J101 [55] soil under lateral cyclic loading [59]. The lateral deformation after N
guideline, and the damage accumulation is evaluated using cycle is evaluated as
Palmgren-Miner summation, which was first proposed by Palmgren [56]
and popularized by Miner [57]. According to their model, the damage yN = y1 N α (28)
accumulated in each cycle is equal to 1/N, when the material fails after
In the above equation, y1 is the static displacement corresponding to
N number of constant stress cycles. Therefore, the damage accumulated
the first cycle, and α is the cyclic degradation parameter depending on
from the jth stress time-history (i.e. short-term damage) is evaluated as the soil and monopile properties, installation method and loading
∑ nji characteristics. Therefore, the strain accumulation in a typical monopile
DST
j = (24)
i
Nji foundation after N cycles can be expressed as

εPL
N = ε1 N
α
(29)
where nji is the rainflow cycle count and Nji is the allowable number of
load cycles corresponding to the ith stress reversal. where ε1 is the strain in the first cycle of loading. This law does not
The lifetime damage is estimated using the probability of occurrence account for the stiffness degradation. Hence, it is applicable only for
and the availability factor during its design life. In general, the hourly piles that undergo little or no flexural stiffness degradation during its
mean wind speed for a year is modelled by Weibull distribution [49], lifetime.
while the availability factor indicates its effective duration. Therefore,
the damage accumulation during the lifetime is estimated as
6.2. Logarithmic method
∑
DL = AfjL DST
j (25)
j The power-law expression is further modified to incorporate degra­
dation factor, which depends on the soil properties, types of pile
where A is the availability factor of the wind turbine and fjL is the scaling installation/loading, pile penetration length and the ratio of pile/soil
factor for the jth time series to account for the probability of occurrence. relative stiffnesses based on the experimental results [60]. In this
The material fails when the damage index DL reaches 1. The fatigue life method, a logarithmic relation for the accumulated strain is used
can be obtained using the damage accumulation over the flow-duration εLL (30)
N = ε1 [1 + βln(N)]
(i.e. the time duration multiplied with the inverse of the damage
accumulation). Here, β is the degradation factor, which is evaluated using the
Further, the short-term and lifetime damage-equivalent loads (DELs) following expression
need to be investigated for safe design. The DEL is defined as the fatigue
β = 0.032λFd Fi Fl (31)
load with constant mean and amplitude occurring at a fixed frequency,
which induces the same damage resulting from variable spectrum load. Parameter λ in the above equation is the ratio of the length of
In this study, the damage equivalent stress (DES) is evaluated instead of monopile (Lp ) to its characteristic length (Lc ), which depends on the
DEL. Thus, the short-term DES is computed as [58]. 1
properties of monopile and soil, i.e. Lc = (Ep Ip /nh )5 , and nh is the coef­
⎡∑ ( )m ⎤m1 ficient of subgrade reaction. Factors Fd , Fi and Fl account for soil density,
nji σae
ji
monopile installation method, and cyclic load ratio, respectively. For a
σ ST
DEj =
⎣ i ⎦ (26) driven monopile in medium dense soil with one-way loading, these
f eq Tj factors are Fd = 1.125, Fi = 1.0 and Fl = 1.0. However, this method is
developed using small-scale pile test in the laboratory, and hence, its

7
M.M. Sajeer et al. Soil Dynamics and Earthquake Engineering 144 (2021) 106674

prediction for an actual monopile is doubtful. speed is kept zero, while it is adjusted above the rated speed to keep the
rpm unaltered. The complete system is mounted over a steel monopile
6.3. Stiffness degradation method with a depth of 36 m from the mudline. The total length is 76 m, whose
height above the mean sea level is 10 m. The properties of soil and
This method for estimating accumulated strain (using stiffness monopile are selected from the NREL report [62]. The Young’s modulus
degradation is based on drained cyclic triaxial tests combined with finite and shear modulus of the steel monopile are 210 GPa and 80.8 GPa,
element analysis) was developed by Achmus et al. [61]. The plastic respectively, whose material density is 8500 kg/m3 . The response
strain in each cycle is inversely proportional to the secant stiffness of the time-histories are evaluated for fatigue design of the monopile using
soil. Therefore, the accumulated strain after the N cycle is obtained as these details of the benchmark turbine. However, before fatigue anal­
ysis, the model is validated with FAST [28,63] in the following
ES1
εPN = εP1 (32) subsection.
ESN

where εPN and ESN are the plastic strain and Young modulus of soil at the
7.1. Validation
Nth load cycle. Achmus et al. [61] proposed an empirical form of the
above equation based on cyclic stress ratio, i.e. The validation of the proposed model with FAST is carried out at the
rated speed with 10% turbulence considering 1.634 m wave height with
(33)
b
SD b 1 Xc 2
ε = εP1 N
N
5.838 s peak spectral period. The responses are simulated without
In the above expression, b1 and b2 are the model parameters, while Xc considering SSI to maintain the uniformity between these models. Under
is the characteristic cyclic stress ratio that varies from 0 to 1. This these conditions, the blade and tower responses are evaluated for
method is useful for large monopiles in the marine environment. comparison, as shown in Fig. 5. In this figure, u denotes the total
Various other methods are also proposed in the literature [6], which deformation, the superscripts b and t represent the blade and tower
are not described here for brevity. The procedures mentioned above are while the subscripts op, ip, fa and ss denote the out-of-plane, in-plane,
used to investigate the long-term behaviour using MBD-SSI model in the fore-aft and side-to-side directions, respectively. These results show that
following section. the responses simulated from the two codes (i.e. proposed model and
FAST) are consistent in amplitude and frequency content. Small varia­
7. Numerical results and discussion tions are observed due to the difference in the numerical solvers of these
codes. However, these small differences are negligible for all practical
In this study, a benchmark NREL 5 MW OC3 Phase II [62] wind propose. Other turbine parameters such as drivetrain torsional rotation
turbine on a monopile foundation is considered for numerical analysis. It (DrTr), nacelle yaw motion (NacYaw) and rotor speed (Ω) are also
has three blades of length 61.5 m, connected to a hub of diameter 3 m. compared with FAST, which are given in Fig. 6. The bending moments
The nacelle of this turbine is placed at the height of 87.6 m from the acting at the mudline of the monopile are critical for determining its
mean sea level. The blades are made of 8 different airfoils, whose lift and fatigue life. Fig. 7 shows the bending moments obtained from FAST and
drag coefficients, mass distribution, bending stiffness, aero-twist are the code developed in this study, where the superscript m denotes the
given in the NREL report [62]. It operates within the wind speeds of 3 mudline. All the comparisons made in this study are plotted after
m/s to 25 m/s (i.e. cut-in and cut-out), while the rated speed for optimal removing the initial transience (about first 300 s), and the results are
performance is 11.4 m/s (i.e. 12.1 rpm). The pitch angle up to the rated compared for a significantly longer duration (700 s). These results show

Fig. 5. Blade and tower displacement time-history for 10% turbulent wind speed of 11.4 m/s; (a) blade tip out-of-plane, (b) blade tip in-plane, (c) tower-top out-of-
plane & (d) tower-top in-plane.

8
M.M. Sajeer et al. Soil Dynamics and Earthquake Engineering 144 (2021) 106674

Fig. 6. Response time-history of rotor and nacelle at rated wind speed with 10% turbulence; (a) drivetrain torsion, (b) yaw motion of nacelle & (c) rotor speed.

Fig. 7. Bending moment time-history at mudline for 10% turbulent rated wind speed; (a) fore-aft & (b) side-to-side.

consistency and are in good agreement with each other. Therefore, it

proves the accuracy of the mathematical model developed in this study.
Hence, it can be used further for fatigue life estimation and long-term
performance analysis of the monopile foundation.

7.2. Response analysis

The monopile turbine system with soil-structure interaction is

modelled, as discussed in Section 3 using the soil characteristics given in
Fig. 8. The model is further used for response time-history analysis to
study its fatigue life and long-term behaviour. For this purpose, the
combined system is solved for different wind flow conditions covering
the operational range as prescribed in IEC 61400–3 [43]. Thus, 17 load
cases are selected in this study as per data provided for K13 site in Dutch
Fig. 8. Soil profile for OC3 Phase II [62].
North Sea [64], which are given in Table 2. V and T are the mean wind
speed and turbulence intensity at the hub height in this table. Parame­
ters Hw and Tp are the wave height and corresponding time period. Six
load cases in Table 2 (i.e. LC1 and LC13-LC17) are in the non-operating
range. For LC17, the mean wind speed is set to 38 m/s [19]. The wind
fields for all the 17 load cases are simulated using TurbSim. The corre­
sponding wave fields are generated as described in Subsection 4.2,

9
M.M. Sajeer et al. Soil Dynamics and Earthquake Engineering 144 (2021) 106674

Table 2 analysis is used for multi-body dynamic analysis. The process is repeated
Wind and wave parameters. at every time step to ensure the numerical convergence of the complete
LC No. V [m/s] T [%] Hw [m] Tp [s] system. Fig. 11b shows the monopile trajectory at the mudline, whose
maximum value in the fore-aft direction is 4.755 mm. This figure in­
1 2 29.2 1.07 6.03
2 4 20.4 1.10 5.88 dicates that the soil nonlinearity in fore-aft and side-to-side directions
3 6 17.5 1.18 5.76 are not fully mobilised. The response is also evaluated using the p-y
4 8 16.0 1.31 5.67 curve proposed by Lee et al. [40]. Fig. 12a and Fig. 12b show this p-y
5 10 15.2 1.48 5.74 curve and monopile trajectory at the mudline, respectively. This p-y
6 12 14.6 1.70 5.88
7 14 14.2 1.91 6.07
curve offers less soil stiffness compared to the API curve. It is reflected in
8 16 13.9 2.19 6.37 the little increase of mudline deformation (refer the difference in
9 18 13.6 2.47 6.71 Figs. 11b and 12b). However, as the DNV guideline [39] refers API
10 20 13.4 2.76 6.99 curve, it is used further in this analysis.
11 22 13.3 3.09 7.40
The platform motion in the fore-aft and side-to-side directions are
12 24 13.1 3.42 7.80
13 26 12.0 3.76 8.14 shown in Fig. 13 along with their Fourier amplitude spectrum, where
14 28 11.9 4.17 8.49 superscript p denotes platform. Fig. 13 also indicates that the first nat­
15 30 11.8 4.46 8.86 ural frequency lies in between the rotational frequency (i.e. 1P) and the
16 32 11.8 4.79 9.12 blade passing frequency (i.e. 3P). Thus, the design of the monopile
17 34–42 11.7 4.90 9.43
comes under the soft-stiff category, which is generally preferred. The
NB: LC stands for load case. response around 0.17 Hz in Fig. 13b corresponds to the spectral fre­
quency of the hydrodynamic load.
which are further used to solve the multi-body dynamic system in Further, deformations of the combined system at different locations
MATLAB using ode45 solver [65]. are obtained using the soil-structure interaction model (i.e. model-I) and
In this subsection, the results are presented for the monopile of compared with the fixed bottom case (i.e. model-II), shown in Table 3.
diameter 7 m and thickness 0.07 m, subjected to the 6th load case (i.e. This table indicates that the model-II always under-estimates the
LC6). The wind-wave misalignment is assumed to be 0◦ , as the fatigue response, which can be as high as 100%. Therefore, these results
failure is most prone to this angle between wind and wave. Fig. 9a shows advocate for the detailed multi-body dynamic analysis considering SSI.
the wind speed at the hub height and wave elevation for LC6, where Fig. 14 shows the variation of peak displacement at mudline, platform
Fig. 9b shows the Fourier amplitude spectrum for the same. The wave and nacelle in the fore-aft and side-to-side directions over the complete
elevation reaches its peak at 0.17 Hz, as shown in Fig. 9b. The aero­ operational window of the turbine (i.e. cut-in to cut-out).
dynamic and hydrodynamic loads acting on the blade and monopile are Besides displacement, rotation/tilt of the structure is equally
demonstrated in Fig. 10. important from the serviceability point of view. Fig. 15a and Fig. 15b
Fig. 11a illustrates the p-y curve of the sand layers used in this study show the fore-aft and side-to-side rotation of the platform, where the
for soil-structure interaction. Similarly, the axial resistance due to soil peak, mean and rms values are included within the figure. The allowable
skin friction and soil end bearing are modelled using t-z and Q-z curves, tilt at the mudline is limited to 0.5◦ , including the tolerance of 0.25◦ for
as shown in Fig. 11c and d, respectively. The soil-pile finite element tower installation [39]. These deformations at the mudline for the
model is developed using these soil spring properties. The equivalent benchmark turbine are found to be within the allowable limits of rota­
stiffness of the soil-monopile system obtained from the finite element tion, i.e. the peak fore-aft and side-to-side rotation at this location are

Table 3
Deformation of wind turbine, NB: Model-I is using SSI, Model-II is using fixed bottom method.
Location Model Fore-aft [mm] Side-to-Side [mm]

peak mean rms peak mean rms

Mudline I 4.755 2.294 2.332 0.711 − 0.305 0.321

II 0.000 0.000 0.000 0.000 0.000 0.000
Platform I 81.750 39.798 40.563 11.960 − 3.000 3.511
II 57.890 26.762 27.262 8.842 − 2.092 2.645
Tower-top I 650.000 314.362 319.575 124.600 − 53.578 56.504
II 337.700 161.676 164.407 74.590 − 29.416 31.559

Fig. 9. Load case 6; horizontal wind velocity at hub height and wave elevation (a) time history & (b) Fourier amplitude.

10
M.M. Sajeer et al. Soil Dynamics and Earthquake Engineering 144 (2021) 106674

Fig. 10. Aerodynamic and hydrodynamic load with 30◦ wind-wave misalignment for LC6; (a) aerodynamic load on blade out-of-plane direction (b) aerodynamic
load on blade in-plane direction, (c) hydrodynamic load in fore-aft direction & (d) hydrodynamic load in side-to-side direction.

Fig. 11. (a) API p-y curve for different soil layers, (b) monopile trajectory at mudline for LC6, (c) t-z curve & (d) Q-z curve.

0.0074◦ and 0.0046◦ , respectively. direction experiences 45∼70% change in magnitude compared to
The response of the wind turbine is also affected by the wind-wave 5∼10% in the fore-aft direction due to wind-wave misalignment. The
misalignment in the marine environment. Thus, the response of the shear force in the fore-aft direction reaches its peak at 60◦ wind-wave
turbine is simulated for various misalignment cases from 0◦ to 90◦ , misalignment, whose impact on the fatigue damage is presented in the
which are summarised in Table 4. These data show that the magnitudes following subsection.
of the internal forces (i.e. shear force and bending moment) in the fore- The bending stresses at the mudline are evaluated for 100 points over
aft direction are significantly higher than side-to-side direction. It is the cross-section, where the mean and peak stresses due to the fore-aft
because the aerodynamic force is more in this direction. The results in moment are 27.495 MPa and 56.347 MPa, respectively. In contrast,
Table 4 also highlights that the wind-wave misalignment has more the same parameters due to the side-to-side bending moment are 1.987
impact on the side-to-side response. The response in the side-to-side MPa and 8.454 MPa, respectively. In addition to bi-axial bending

11
M.M. Sajeer et al. Soil Dynamics and Earthquake Engineering 144 (2021) 106674

Fig. 12. (a) p-y curve for different soil layers as per Lee et al. [40] & (b) monopile trajectory at mudline for LC6.

Fig. 13. Platform deformation with monopile diameter 7 m and thickness 0.07 m for LC6; (a) fore-aft, (b) Fourier amplitude spectrum of fore-aft deformation, (c)
side-to-side & (d) Fourier amplitude spectrum of side-to-side deformation.

stresses, the stress due to axial force is also evaluated, whose mean and 7.3. Fatigue life estimation
peak values are 5.511 MPa and 9.416 MPa, respectively. The total lon­
gitudinal stress distribution is estimated using Eq. 26, which is domi­ The fatigue life is evaluated using the S–N curve recommended in the
nated by the fore-aft moment. The critical location for maximum stress DNV guideline [67] for steel structures in the marine environment (refer
over the cross-section is at xc = − 0.2198 m, yc = 3.4931 m, as shown in to Fig. 16b), whose ultimate strength is 440 MPa. The availability factor
the inset of Fig. 16a. The shear stress due to torsion is also evaluated, and is assumed to be 1, i.e. the turbine offers full power generation within its
the mean and peak values are obtained as 8.063 kPa and 598.942 kPa, operational window. Besides material properties and availability factor,
respectively. Thus, the mean and peak of the shear stress time history are lifetime fatigue design requires the probability distribution of environ­
estimated as 0.788 MPa and 2.411 MPa, respectively, which are less in mental loads at the site, modelled as Weibull distribution for each load
this case than other stresses. The von-Mises stress time history is ob­ case given in Table 2. The parameters of this probability density function
tained from the different stress components, whose mean and peak are c = 11.31 m/s (i.e. scaling parameter) and k = 1.97 (i.e. shape
values are 33.080 MPa and 62.108 MPa, respectively. parameter) [64].
The mean and peak values of strains at the critical location are The fatigue life is estimated separately from the fore-aft moment and
157.51E-6 and 295.75E-6, respectively, and are well within the allow­ stress as 29.87yrs and 26.92yrs, respectively. Fore-aft moment-based
able limit (i.e. 1262E-6 [66]). However, the turbine undergoes a large calculation overestimates the fatigue life by 11%, mainly due to two
number of cyclic loading during its lifetime in the marine environment. reasons - (i) change in critical location and (ii) the additional stress due
Therefore, fatigue life and long-term performance are investigated to the axial force. The fatigue lives without Goodman correction in these
further for the design of the monopile foundation. cases are 38.05yrs and 36.15yrs, respectively.
Further, the impact of wind-wave misalignment on monopile fatigue

12
M.M. Sajeer et al. Soil Dynamics and Earthquake Engineering 144 (2021) 106674

wind and wave loads are coincident. The minimum damage occurs when
θma is 60◦ due to the change in the fore-aft shear force.
Once the fatigue life is estimated, both short-term and lifetime
damage-equivalent stresses are evaluated to study the serviceability
criteria. As discussed earlier, the maximum short-term DES corre­
sponding to the 12th load case has a mean value of 23.00 MPa with an
amplitude of 9.99 MPa. Thus, the strain during this cycle is 157.10E-6,
which is well within the design limit. The lifetime DES is obtained as
6.94 MPa corresponding to 580848302.6 equivalent cycles of load with
a mean 23.00 MPa at 1 Hz, which is further used in the following sub­
section to study the long-term performance.

7.4. Long-term deformation estimation

The long-term performance of monopile is usually estimated to check

the serviceability limit. In this study, the long-term performance is
evaluated in terms of strain using lifetime damage-equivalent stress
obtained from the fatigue analysis. The maximum absolute stress in the
first cycle of damage-equivalent stress for diameter 7 m and thickness
0.07 m in 30 m water depth is 29.94 MPa. Therefore, the corresponding
Fig. 14. Peak deformations of an offshore wind turbine in fore-aft and side-to- strain in the first cycle is 142.57E-6. The long-term performance of the
side directions. [N.B.: figures are not in scale]. monopile is evaluated by three different methods, as discussed in Sec­
tion 7.4. Fig. 18 shows the variation of strain vs the number of cycles,
damage is investigated (refer to Fig. 17). The critical location is changed where the soil condition is considered to be medium dense sand. This
to xc = − 0.4387 m, yc = 3.4724 m for higher wind-wave misalignment, figure also reveals that the overall dimensions of the monopile satisfy its
i.e. θma ≥ 30∘ . It is observed that the maximum short-term damage oc­ serviceability limit.
curs during the 12th load case, which corresponds to the maximum wind
speed within the operational window. However, the 12th load case has 7.5. Sensitivity analysis
less probability of occurrence than load case 6. Due to this reason, the 6th
load case contributes most in lifetime damage accumulation. Fig. 17c The aquatic profile in the marine environment varies with its spatial
shows the lifetime damage for various wind-wave misalignments. This extend. The hydrodynamic load increases with the water depth, which
figure also indicates that the maximum lifetime damage occurs when the ultimately affects the fatigue life of the foundation. Thus, a sensitivity

Fig. 15. Rotation of wind turbine at platform level with monopile diameter 7 m and thickness 0.07 m for LC6; (a) fore-aft & (b) side-to-side.

Table 4
Shear force and bending moment of monopile at mudline for various wind-wave misalignment cases.
Reactions Wind-wave misalignment Fore-aft Side-to-Side

peak mean rms peak mean rms

Shear Force [kN] 0◦ 1525.000 592.600 650.421 287.200 − 6.433 60.005

15◦ 1497.000 592.623 647.584 343.500 − 6.647 90.910
30◦ 1416.000 592.687 639.898 517.800 − 6.843 140.015
45◦ 1286.000 592.787 629.303 698.900 − 7.008 185.848
60◦ 1299.000 592.917 618.597 838.200 − 7.129 221.452
75◦ 1322.000 593.068 610.753 924.600 − 7.200 243.348
90◦ 1367.000 593.229 608.062 952.000 − 7.215 249.706
Bending Moment [kNm] 0◦ 147.300 71.876 73.349 22.100 5.194 6.199
15◦ 147.500 71.877 73.329 22.170 5.205 7.382
30◦ 148.200 71.879 73.276 29.090 5.214 9.078
45◦ 149.300 71.882 73.205 34.560 5.222 10.687
60◦ 150.800 71.885 73.137 38.130 5.227 11.873
75◦ 152.700 71.889 73.092 40.400 5.230 12.465
90◦ 154.900 71.894 73.084 40.350 5.230 12.391

13
M.M. Sajeer et al. Soil Dynamics and Earthquake Engineering 144 (2021) 106674

Fig. 16. (a) Longitudinal stress of monopile at the critical point & (b) S–N curve for offshore steel material [67].

Fig. 17. Damage of monopile for different load cases with various wind-wave misalignment; (a) short-term damage, (b) weighted damage with the probability of
occurrence & (c) lifetime damage.

analysis is conducted for various water depths, which is presented in

Table 5. A comparison of fatigue life using fore-aft moment and longi­
tudinal stress is also presented in this table, which clearly shows the
fatigue life is overestimated using the fore-aft moment-based approach.
The results in this table highlights the impact of diameter and thickness,
i.e. as diameter increases, the required thickness decreases and vice
versa. However, for specific water depth, a minimum thickness corre­
sponding to a particular diameter is recommended by API [36] guide­
lines to avoid local buckling. This issue is discussed further in the
following subsection.
Besides buckling, different monopile parameters (i.e. diameter,
thickness and water depth) also affect the short-term DES. Fig. 19 shows
this analysis for different load cases that offer fatigue life above its
design value, i.e. 20 yrs. The general trend of the short-term DES (i.e.
Fig. 18. Variation of strain with respect to the number of cycles. refer to Fig. 19) follows the damage accumulation pattern (i.e. refer to
Fig. 17) for different wind-wave misalignments. It indicates that the
short-term DES increases with load case number and it is maximum at
the cut-out wind speed. However, due to more probability of occurrence,
rated speed (i.e. load case 6) contributes more damage accumulation
during the lifetime, as shown in Fig. 19b.

14
M.M. Sajeer et al. Soil Dynamics and Earthquake Engineering 144 (2021) 106674

Fig. 19. Short-term damage-equivalent stress (DES) of various monopile model for different load cases; (a) short-term DES & (b) relative short-term DES weighted
with the probability of occurrence.

Table 5
Fatigue life of OC3 monopile wind turbine.
Dp [m] tp [m] Fatigue life [years]

using fore-aft moment using longitudinal stress

Dw [m] Dw [m]

10.00 20.00 30.00 40.00 50.00 10.00 20.00 30.00 40.00 50.00

6.00 0.05 4.50 – – – – 3.36 – – – –

6.00 0.06 11.54 6.02 – – – 8.78 4.95 – – –
6.00 0.07 25.53 13.41 – – – 19.63 11.10 – – –
6.00 0.08 49.47 26.22 11.32 – – 38.37 21.85 10.53 – –
6.00 0.09 87.84 46.30 20.96 – – 68.81 39.00 19.63 – –
6.00 0.10 – – 36.15 12.81 5.90 – – 31.20 11.29 5.23
6.00 0.11 – – – 20.55 10.21 – – – 18.07 9.04
7.00 0.06 57.71 30.35 14.14 – – 41.86 23.85 12.68 – –
7.00 0.07 124.94 66.91 29.87 12.46 – 90.69 52.64 26.92 11.29 –
7.00 0.08 240.99 130.33 57.39 26.51 10.62 175.67 103.06 52.00 24.00 9.93
7.00 0.09 428.08 234.65 100.20 48.52 17.95 313.61 186.14 91.32 44.08 16.87
7.00 0.10 – – – – 29.71 – – – – 26.07
8.00 0.08 – – 261.92 103.37 27.87 – – 226.09 91.01 25.84
8.00 0.09 – – 475.65 184.87 48.83 – – 409.06 162.99 45.35
8.00 0.10 – – 799.09 304.73 87.84 – – 684.93 269.85 82.13

7.6. Site-specific design curve diameter of the monopile. The dotted line in this figure corresponding to
7 m diameter indicates the region vulnerable to local buckling. Table 6
The sensitivity analysis presented in the above subsection clearly shows the variation of first natural frequency in two orthogonal di­
shows the impact of water depth on fatigue life. It motivates to develop rections, fatigue life and corresponding strain levels for design curve in
the design curve for a given power rating at a particular site. As stated Fig. 20b. The first natural frequency in two orthogonal directions (i.e.
earlier, the minimum thickness for a given diameter is recommended by fore-aft and side-to-side) lie between 1P and 3P, as shown in Fig. 21.
API guideline [i.e. tpmin = 0.01 Dp + 0.00635] to avoid local buckling. Hence, the monopile design using the dimensions obtained from these
Fig. 20a shows the variation of diameter and thickness for different curves (i.e. Fig. 20) does not suffer resonance due to blade rotation.
water depth and power rating, i.e. 5 MW. Here, it is worth mentioning Table 6 also highlights the strains in different cycles using techniques
that fatigue life for each case is more than its design value, i.e. 20 yrs. discussed in Section 7.4. These results indicate that soil stiffness
The outer diameter of a monopile depends on the size of the tower and degradation significantly affects the performance of the monopile.
transition piece. Fig. 20b shows the effect of wall thickness for a given However, the data in Table 6 shows that the long-term performance (i.e.
strain in the Nth cycle) is satisfactory (i.e. within the allowable limit).

Fig. 20. Design curve for NREL 5 MW offshore wind turbine monopile dimensions with the depth of water; (a) for varying diameter & (b) for fixed diameter.

15
M.M. Sajeer et al. Soil Dynamics and Earthquake Engineering 144 (2021) 106674

Table 6
Fatigue life and long-term strain of monopile with 6 m and 7 m diameter for various water depth.
Dp Dw tp ffa fss Fatigue Life Strian in 1st cycle Strain in Nth cycle

[m] [m] [m] [Hz] [Hz] [Years] με1 μεPL

N μεLL
N μεSD
N

6.00 10.00 0.075 0.2967 0.2883 26.22 160.48 910.16 600.43 541.54
20.00 0.085 0.2567 0.2783 27.56 160.00 907.46 588.24 539.19
30.00 0.100 0.2567 0.2617 31.20 140.48 796.73 504.98 446.44
40.00 0.120 0.2567 0.2567 28.30 137.62 780.53 482.62 433.45
50.00 0.140 0.2567 0.2483 29.20 136.67 775.13 469.55 429.15
7.00 10.00 0.055 0.2967 0.2950 25.80 162.38 920.97 593.00 550.97
20.00 0.062 0.2967 0.2817 27.99 161.43 915.57 579.64 546.24
30.00 0.070 0.2567 0.2650 30.23 142.57 808.62 503.32 456.07
40.00 0.082 0.2567 0.2617 26.72 140.95 799.43 486.85 448.62
50.00 0.100 0.2567 0.2567 26.07 137.14 777.83 461.10 431.30

Fig. 21. Fourier amplitude spectrum of platform deformation with monopile dimensions as per design curve for LC6; (a) Dp = 6m, fore-aft, (b) Dp = 6m, side-to-
side, (c) Dp = 7m, fore-aft & (d) Dp = 7m, side-to-side.

Fig. 22. (a) Weibull distribution of wind speed for different scaling and shape parameters & (b) Fatigue life variation for various Weibull parameters.

16
M.M. Sajeer et al. Soil Dynamics and Earthquake Engineering 144 (2021) 106674

Finally, the variation of fatigue life due to changes in wind speed misalignment is sixty degrees. The maximum short-term damage
distribution is investigated. Fig. 22a shows the variation of wind dis­ occurs at the cut-out speed, while the rated wind speed mostly
tribution for different scale and shape parameter (i.e. c & κ) and Fig. 22b contributes to the lifetime damage. Therefore, long-term damage due
illustrates the impact of these parameters on the fatigue life. The mini­ to fatigue load is governed by the rated speed as the turbine wit­
mum fatigue life from this analysis is obtained as 23.97 years corre­ nesses maximum exposure to this operational condition.
sponding to c = 13 m/s and κ = 3, i.e. at the maximum value of scale and • The work also investigates the long-term behaviour and the param­
shape parameter. Thus, the curves in Fig. 20 can be used to design eters that influence damage-equivalent stress to meet the stringent
monopiles for similar turbines in dense sandy soil. serviceability criteria. The results demonstrate the impact of stiffness
degradation on monopiles’ long-term performance in the marine
8. Conclusions environment.
• Finally, the study offers the design curves for a given power rating of
The work presented in this paper develops a multi-body dynamic the turbine. These design curves mostly highlight the interrelation
soil-structure interaction model of an offshore wind turbine for fatigue between three key parameters, i.e. diameter, thickness and depth of
design and long-term performance evaluation. The paper demonstrates water. The dimensions obtained using these design curves are
stress-based fatigue analysis considering various monopile parameters, satisfactory in terms of fatigue life and serviceability criteria. Thus,
water depth and operational conditions. The significant contributions of these curves can be helpful for the design of similar structures and for
this study are as follows. project planning that resembles the same marine environment.

• The combined system is modelled in Kane’s approach using equiv­ Credit Author Statement
alent monopile properties obtained from a detailed substructure
analysis. It uses Winkler’s beam theory to model the soil layers M. Mohamed Sajeer: Conceptualization, Methodology, Software
defined by their respective p-y curves. The model is validated using Development and Writing - Original draft preparation
the properties of a benchmark turbine for turbulent wind. The pro­ Arka Mitra: Multi-body dynamic analysis including SSI, Software
posed model offers good agreement with FAST. It is further used to Development and Writing - Original draft preparation
model soil-structure interaction in the marine environment. Arunasis Chakraborty: Conceptualization, Investigation, Supervi­
• The numerical results demonstrate fatigue design of monopile sion, Writing- Reviewing and Editing
foundation considering different load cases prescribed in IEC
guidelines [43]. These results highlight the significance of Declaration of competing interest
stress-based modelling using soil-structure interaction for long-term
performance evaluation. The authors declare that they have no known competing financial
• As evident, the maximum damage occurs when the aerodynamic and interests or personal relationships that could have appeared to influence
hydrodynamic forces coincide, and their effect is minimum when the the work reported in this paper.

A. Monopile-tower system

The transition piece rigidly connects the monopile and tower to be modelled together as a cantilever with distributed mass, stiffness and other
properties. The global or inertial reference frame ̂ x is defined at the monopile base, which coincides with the local coordinate system ̂ a at a height h.
The position vector of a point on the monopile-tower can be written as
[ ( ) ( ) ] [ ( ( ) ( ) ( ) )]
rmb ,T (h) = φTFA
1 h .u10 + φ TFA
2 h .u12 a
.̂ 1 + h − 0.5 S TFA
1,1 h .u2
10 + S TFA
2,2 h .u2
12 + 2S TFA
1,2 (h).u 10 .u12 + S TSS
1,1 (h).u 2
11 + S TSS
2,2 (h).u 2
13 + 2S TSS
1,2 h .u 11 .u13 a2
.̂
[ ( ) ( ) ]
+ φTSS1 h .u11 + φTSS
2 h .u13 .̂ a3 (A.1)

where φTFA TFA TSS TSS

1 (h), φ2 (h), φ1 (h) and φ2 (h) are the first two mode shapes of the monopile-tower in the fore-aft and side-to-side directions, respectively.
TFA TSS
The coefficients Si,j (h) and Si,j (h) are given by
∫ h [ TFA ′ ′
]
dφi (h ) dφTFA (h )
(A.2a)
j ′
Si,j (h) = . .dh for i, j = 1, 2
0 dh′ dh′

∫ [ ′
]
h
dφTSS
′
i (h )
dφTSS
j (h )
(A.2b)
′
Si,j (h) = ′ . .dh for i, j = 1, 2
0 dh dh′

Now, linear velocities at this point can be obtained from the position vector as follows
[ ( ) ( ) ] [ ( ( ) ( ) ( )
vT (h) = φTFA
1 h u̇10 + φTFA
2 h u̇12 ̂ TFA
a 1 + h − 0.5 S1,1 TFA
h u10 .u̇10 + S2,2 TFA
h u12 .u̇12 + S1,2 TSS
(h). u̇10 .u12 + u10 .u̇12 + S1,1 TSS
(h).u11 .u̇11 + S2,2 (h).u13 .u̇13
( )( )) ] [ ( ) ( ) ]
TSS
+S1,2 h u̇11 .u13 + u13 .u̇11 a 2 + φTSS
̂ 1 h u̇11 + φTSS
2 h u̇13 ̂ a3

(A.3)
Similarly, the angular velocities at the same point can be expressed using small-angle approximations as
[ ] [ ]
dφTSS
1 (h) dφTSS (h) dφTFA
1 (h) dφTFA (h)
T
ω (h) = .u̇11 + 2 u̇13 .̂
a1 − .u̇10 + 2 u̇12 .̂
a3 (A.4)
dh dh dh dh

17
M.M. Sajeer et al. Soil Dynamics and Earthquake Engineering 144 (2021) 106674

The rigid body components of total linear and angular velocities are zero, i.e. vT(h)
t = 0 and ωT(h)
t = 0. Hence, the linear components of these
velocities are obtained by rearranging Eq. A.3 and Eq. A.4 in the format of Eq. (5a) and Eq. (6b), respectively as follows
⎧ [ ]
⎪
⎪
⎪ φTFA
⎪ 1 (h).̂
TFA
a 1 − S1,1 TFA
(h).u10 + S1,2 a 2 for i = 10
(h).u12 .̂
⎪
⎪ [ ]
⎪
⎪
⎪ TSS
⎨ φ1 (h).̂ TSS TSS
a 3 − S1,1 (h).u11 + S1,2 (h).u13 .̂a 2 for i = 11
vT(h)
i = [ ] (A.5a)
⎪
⎪ TFA TFA TFA
⎪ φ2 (h).̂
⎪ a 1 − S2,2 (h).u10 + S1,2 a 2 for i = 12
(h).u12 .̂
⎪
⎪ [ ]
⎪
⎪ TSS
⎪
⎩ φ2 (h).̂ TSS
a 3 − S2,2 TSS
(h).u11 + S1,2 a 2 for i = 13
(h).u13 .̂

⎧ dφTFA (h)
⎪ 1
a3 fori = 10
⎪−
⎪
⎪ dh
.̂
⎪
⎪
⎪
⎪
⎪
⎪ dφTSS (h)
⎪
⎪
⎪ 1
⎨ a1
.̂ fori = 11
dh
ωT(h)
i = (A.5b)
⎪ dφTFA (h)
⎪
⎪
⎪ 2
⎪−
⎪ a3
.̂ fori = 12
⎪
⎪ dh
⎪
⎪
⎪
⎪
⎪ TSS
⎩ dφ2 (h)
a1
.̂ fori = 13
dh
Using above two expressions for partial linear and angular velocities (i.e. vT(h)
i and ωT(h)
i ), the expression for partial linear and angular acceleration
can be obtained by differentiating Eq. A.5a and Eq. A.5b with respect to time, i.e.
⎧[ ]
⎪
⎪
⎪−
⎪
TFA
S1,1 TFA
(h).u̇10 + S1,2 (h).u̇12 .̂a2 for i = 10
⎪
⎪
⎪
⎪[ ]
⎪
⎨− TSS
S1,1 (h).u̇11 + STSS a2 for i = 11
1,2 (h).u̇13 .̂
d ( T(h) )
vi = [ ] (A.6a)
dt ⎪
⎪ TFA
− S2,2 TFA
(h).u̇10 + S1,2 (h).u̇12 .̂a2 for i = 12
⎪
⎪
⎪
⎪ [ ]
⎪
⎪
⎪ TSS
⎩ − S2,2 (h).u̇11 + STSS
1,2 (h).u̇13 .̂a2 for i = 13

d ( T(h) )
ω =0 (A.6b)
dt i
In this context, it is necessary to mention that the derivative of the rigid body components of the partial velocities will be equal to zero, i.e. dtd (vT(h)
t )
= 0 and dtd (ωT(h)
i ) = 0. Using all appropriate expressions of linear and angular velocities and accelerations in Eq. (4) and the distributed properties of
monopile-tower, the generalised inertial force due to tower mass can be written as
∫ Htw
*
Fi,tw =− μT (h).vTi (h).aTi (h).dh − MYB .vTi (Htw ).aTi (Htw )
0

∫ [( ) ( )] [( ) ( )] (A.7)
Htw ∑
13 ∑13
d{ T } ∑
13 ∑13
d{ T }
=− μT (h).vTi (h). vTi (h).üi + v (h) .u̇i .dh − MYB .vTi (Htw ). vTi (Htw ).üi + v (Htw ) .u̇i
0 i=10 i=10
dt i i=10 i=10
dt i

where Htw is the height of the tower defined from the monopile base and MYB is the mass of the yaw bearing. The distributed mass per unit length of the
tower is expressed as μT (h) in the above equation. Hence, the components of the system matrices can be obtained by rearranging Eq. A.7 in the
following form
∫ Htw
[M(u, t)]tw (i, j) = μT (h).vTi (h).vTj (h).dh + MYB .vTi (Htw ).vTj (Htw ) for i, j = 10...13 (A.8)
0

∫ ( ) ( )
Htw ∑ 13
d{ T } ∑ 13
d{ T }
[ − f (u̇, u, t)]tw (i) = − μT (h).vTi (h). vi (h) .u̇i .dh − MYB .vTi (Htw ). vi (Htw ) .u̇i for i, j = 10...13 (A.9)
0 i=10
dt i=10
dt

The elastic and damping forces, aerodynamic and hydrodynamic forces and gravitational self-weight of the tower contribute to the generalised
active forces. The generalised elastic forces can be obtained from the potential energy stored in each member and are given by
⎧
⎪
⎪ TFA
− k1,1 TFA
.u10 − k1,2 .u12 for i = 10
⎪
⎪
⎪
⎪
⎨ − kTSS .u11 − kTSS .u13 for i = 11
(A.10)
1,1 1,2
[ − f (u̇, u, t)]tw,el (i) = TFA TFA
⎪
⎪ − k2,1 .u10 − k2,2 .u12 for i = 12
⎪
⎪
⎪
⎪ TSS TSS
⎩ − k2,1 .u11 − k2,2 .u13 for i = 13

where kTFA TSS

i,j and ki,j are the generalised stiffness of the monopile-tower, which are expressed as follows

18
M.M. Sajeer et al. Soil Dynamics and Earthquake Engineering 144 (2021) 106674

∫ Htw
d2 φTFA 2 TFA
1 (h) d φ2 (h)
TFA
ki,j = EI TFA (h). 2
. dh (A.11a)
0 dh dh2
∫ Htw
d2 φTSS 2 TSS
1 (h) d φ2 (h)
TSS
ki,j = EI TSS (h). . dh (A.11b)
0 dh2 dh2

In the above equation, EITFA (h) and EITSS (h) are the distributed stiffness properties of the tower in the fore-aft and side-to-side directions,
respectively. Considering Rayleigh’s damping for the monopile-tower system, the dissipating forces can be expressed as
⎧
⎪
⎪
⎪
⎪ ξTFA TFA
1 .k1,1 ξTFA TFA
2 .k1,2
⎪
⎪ − .u̇10 − .u̇12 for i = 10
⎪
⎪ π .f1TFA π .f2TFA
⎪
⎪
⎪
⎪
⎪
⎪ ξTSS TSS
1 .k1,1 ξTSS TSS
2 .k1,2
⎪
⎪ .u̇11 − .u̇13 for i = 11
⎨− π .f1TSS π.f2TSS
[ − f (u̇, u, t)]tw,d (i) = (A.12)
⎪
⎪ ξTFA TFA
ξTFA TFA
⎪ 1 .k2,1 2 .k2,2
⎪
⎪ − .u̇10 − .u̇12 for i = 12
⎪
⎪ π .f1TFA π .f2TFA
⎪
⎪
⎪
⎪
⎪
⎪ ξTSS TSS
1 .k2,1 ξTSS TSS
2 .k2,2
⎪
⎪ − .u̇11 − .u̇13 for i = 13
⎪
⎩ π .f1TSS π.f2TSS

where ξTFA
i and ξTSS i denote the structural damping ratio in the ith mode of vibration in the fore-aft and side-to-side directions, respectively. The natural
frequencies of the monopile-tower in the ith mode of vibration in the fore-aft and side-to-side directions (i.e. fiTFA and fiTSS ) can be expressed as
√̅̅̅̅̅̅̅̅̅̅
1 kTFA
(A.13a)
TFA i,i
fi =
2π mTFA i,i

√̅̅̅̅̅̅̅̅̅
1 kTSS
(A.13b)
i,i
fiTSS =
2π mTSSi,i

Similarly, the modal mass of the monopile-tower system mTFA

i,i and mTSS
i,i can be obtained from
∫ Htw
mTFA
i,j = μT (h).φTFA
i (h).φTFA
j (h).dh (A.14a)
0

∫ Htw
mTSS
i,j = μT (h).φTSS TSS
i (h).φj (h).dh (A.4b)
0

The hydrodynamic load due to incident waves contribute to the generalised active force. The effect of added mass, fluid inertia and the viscous drag
forces are included in this analysis. Hence, the wave effect on the generalised active force is given by
∫ Hmsl [ ]
[ − f (u̇, u, t)]tw,hy (i) = − vTi (h).Ftw,hy
T
(h) + ωTi (h).Mtw,hy
T
(h) .dh (A.15)
Hmud

tb
In the above expression, Ftw,hy and MTtw,hy (h) represent the hydrodynamic force and moment expressed per unit length, respectively. Besides the
wave effect, these forces also depend on the motion of the monopile.
Similar to the hydrodynamic forces, the aerodynamic forces also act on the tower above the sea level. Thus, the elements of the system matrices due
to the aerodynamic force is obtained as
∫ Htw [ ]
[ − f (u̇, u, t)]tw,ae (i) = − vTi (h).Ftw,ae
T
(h) + ωTi (h).Mtw,ae
T
(h) .dh (A.16)
Hmsl

tb
In the above expression, Ftw,ae and MTtw,ae (h) represent the aerodynamic force and moment expressed per unit length, respectively. They are inte­
grated from the sea level, up to the height of the tower. Finally, the generalised active forces due to gravity on the monopile-tower are given by
∫ Htw { }
[ − f (u̇, u, t)]tw,gv (i) = − vTi (h). μT (h).g.̂
x 2 .dh − MYB .g.vTi (Htw ).̂
x2 (A.17)
0

where g is the acceleration due to gravity. Here, it must be noted that subscript i in equations Eq. A.15 to Eq. A.17 will range from DOF number
corresponding to monopile-tower, i.e. 10 to 13.
Following the same procedure, the equation of motion for each wind turbine component can be derived and assembled to form the governing
equation for the whole turbine system in its global reference frame.

19
M.M. Sajeer et al. Soil Dynamics and Earthquake Engineering 144 (2021) 106674

References [32] Damiani R, Jonkman J, Hayman G. Subdyn user’s guide and theory manual, Tech.
rep. Golden, CO (United States): National Renewable Energy Lab.(NREL); 2015.
[33] Kane TR, Levinson DA. Dynamics, theory and applications. McGraw Hill; 1985.
[1] Adhikari S, Bhattacharya S. Vibrations of wind-turbines considering soil-structure
[34] Sajeer M, Mitra A, Chakraborty A. Spinning finite element analysis of
interaction. Wind Struct 2011;14(2):85.
longitudinally stiffened horizontal axis wind turbine blade for fatigue life
[2] Bisoi S, Haldar S. Dynamic analysis of offshore wind turbine in clay considering
enhancement. Mech Syst Signal Process 2020;145:106924. https://doi.org/
soil–monopile–tower interaction. Soil Dynam Earthq Eng 2014;63:19–35.
10.1016/j.ymssp.2020.106924.
[3] Zuo H, Bi K, Hao H. Dynamic analyses of operating offshore wind turbines
[35] Reddy JN. An introduction to the finite element method. New York: McGraw-Hill;
including soil-structure interaction. Eng Struct 2018;157:42–62.
2004.
[4] Bisoi S, Haldar S. Design of monopile supported offshore wind turbine in clay
[36] Rp2A-WSD, API, Recommended practice for planning, designing and constructing
considering dynamic soil–structure-interaction. Soil Dynam Earthq Eng 2015;73:
fixed offshore platforms–working stress design, Houston: American Petroleum
103–17.
Institute.
[5] Rong XN, Xu RQ, Wang HY, Feng SY. Analytical solution for natural frequency of
[37] I. Møller, T. Christiansen, Laterally loaded monopile in dry and saturated sand-
monopile. Wind Struct 2017;25(5):459–74.
static and cyclic loading: experimental and numerical studies, Masters Project,
[6] Bhattacharya S. Design of foundations for offshore wind turbines. Wiley Online
Aalborg University Esbjerg.
Library; 2019.
[38] Desai C, Zaman M. Advanced geotechnical engineering: soil-structure interaction
[7] Banerjee A, Chakraborty T, Matsagar V, Achmus M. Dynamic analysis of an
using computer and material models. CRC Press; 2013.
offshore wind turbine under random wind and wave excitation with soil-structure
[39] DNVGL-ST-0126, Support structures for wind turbines, Det Norske Veritas.
interaction and blade tower coupling. Soil Dynam Earthq Eng 2019;125:105699.
[40] Lee M, Bae K-T, Lee IW, Yoo M. Cyclic py curves of monopiles in dense dry sand
[8] Kementzetzidis E, Corciulo S, Versteijlen WG, Pisano F. Geotechnical aspects of
using centrifuge model tests. Appl Sci 2019;9(8):1641.
offshore wind turbine dynamics from 3D non-linear soil-structure simulations. Soil
[41] Boulanger RW, Curras CJ, Kutter BL, Wilson DW, Abghari A. Seismic soil-pile-
Dynam Earthq Eng 2019;120:181–99.
structure interaction experiments and analyses. J Geotech Geoenviron Eng 1999;
[9] Prendergast LJ, Gavin K. A comparison of initial stiffness formulations for small-
125(9):750–9.
strain soil–pile dynamic Winkler modelling. Soil Dynam Earthq Eng 2016;81:
[42] Choo YW, Kim D. Experimental development of the p-y relationship for large-
27–41.
diameter offshore monopiles in sands: centrifuge tests. J Geotech Geoenviron Eng
[10] Prendergast L, Wu W, Gavin K. Experimental application of FRF-based model
2016;142(1):04015058.
updating approach to estimate soil mass and stiffness mobilised under pile impact
[43] IEC-61400-3, Wind turbines - part 3: design requirements for offshore wind
tests. Soil Dynam Earthq Eng 2019;123:1–15.
turbines, International Electrotechnical Commission.
[11] Li J, Wang X, Guo Y, Yu XB. The loading behavior of innovative monopile
[44] Hansen MO. Aerodynamics of wind turbines. Routledge; 2015.
foundations for offshore wind turbine based on centrifuge experiments. Renew
[45] Burton T, Sharpe D, Jenkins N. Handbook of wind energy. John Wiley & Sons;
Energy 2020;152:1109–20.
2001.
[12] Katsikogiannis G, Bachynski EE, Page AM. Fatigue sensitivity to foundation
[46] Jonkman BJ. Turbsim user’s guide: version 1.50, Tech. rep. Golden, CO (United
modelling in different operational states for the DTU 10MW monopile-based
States): National Renewable Energy Laboratory (NREL); 2009.
offshore wind turbine. J Phys Conf 2019;1356(1):012019.
[47] Sarkar S, Chakraborty A. Optimal design of semiactive mr-tlcd for along-wind
[13] Schafhirt S, Page AM, Eiksund GR, Muskulus M. Influence of soil parameters on the
vibration control of horizontal axis wind turbine tower. Struct Contr Health Monit
fatigue lifetime of offshore wind turbines with monopile support structure. Energy
2018;25(2):e2083.
Procedia 2016;94:347–56.
[48] Ning SA. A simple solution method for the blade element momentum equations
[14] Rezaei R, Fromme P, Duffour P. Fatigue life sensitivity of monopile-supported
with guaranteed convergence. Wind Energy 2014;17(9):1327–45.
offshore wind turbines to damping. Renew Energy 2018;123:450–9.
[49] Burton T, Jenkins N, Sharpe D, Bossanyi E. Wind energy handbook. John Wiley &
[15] Carswell W, Johansson J, Løvholt F, Arwade S, Madshus C, DeGroot D, Myers A.
Sons; 2011.
Foundation damping and the dynamics of offshore wind turbine monopiles. Renew
[50] Pierson Jr WJ, Moskowitz L. A proposed spectral form for fully developed wind
Energy 2015;80:724–36.
seas based on the similarity theory of SA Kitaigorodskii. J Geophys Res 1964;69
[16] Aasen S, Page AM, Skau KS, Nygaard TA. Effect of foundation modelling on the
(24):5181–90.
fatigue lifetime of a monopile-based offshore wind turbine. Wind Energy Science
[51] K. Hasselmann, T. Barnett, E. Bouws, H. Carlson, D. Cartwright, K. Enke, J. Ewing,
2017;2:361–76.
H. Gienapp, D. Hasselmann, P. Kruseman, et al, Measurements of wind-wave
[17] Marino E, Giusti A, Manuel L. Offshore wind turbine fatigue loads: the influence of
growth and swell decay during the Joint North Sea wave project (JONSWAP),
alternative wave modeling for different turbulent and mean winds. Renew Energy
Ergänzungsheft 8-12.
2017;102:157–69.
[52] Das S, Sajeer MM, Chakraborty A. Vibration control of horizontal axis offshore
[18] De Risi R, Bhattacharya S, Goda K. Seismic performance assessment of monopile-
wind turbine blade using SMA stiffener. Smart Mater Struct 2019;28(9):095025.
supported offshore wind turbines using unscaled natural earthquake records. Soil
https://doi.org/10.1088/1361-665x/ab1174.
Dynam Earthq Eng 2018;109:154–72.
[53] Downing SD, Socie D. Simple rainflow counting algorithms. Int J Fatig 1982;4(1):
[19] Løken IB, Kaynia AM. Effect of foundation type and modelling on dynamic
31–40.
response and fatigue of offshore wind turbines. Wind Energy 2019;22(12):
[54] Goodman J. Mechanics applied to engineering. Longmans, Green; 1918.
1667–83.
[55] DNV-OS-J101, Guidelines for design of wind turbines, det norske veritas.
[20] Jacob A, Mehmanparast A, D’Urzo R, Kelleher J. Experimental and numerical
[56] A. Palmgren, Die lebensdauer von kugellagern (life length of roller bearings. in
investigation of residual stress effects on fatigue crack growth behaviour of S355
German), Zeitschrift des Vereines Deutscher Ingenieure (VDI Zeitschrift). ISSN
steel weldments. Int J Fatig 2019;128:105196.
(1924) 0341–7258.
[21] Igwemezie V, Mehmanparast A. Waveform and frequency effects on corrosion-
[57] Miner M, et al. Cumulative fatigue damage. J Appl Mech 1945;12(3):A159–64.
fatigue crack growth behaviour in modern marine steels. Int J Fatig 2020;134:
[58] Hayman G. MLife theory manual for version 1.00, vol. 74. Golden, CO: National
105484.
Renewable Energy Laboratory; 2012. p. 106. 75.
[22] Kim SB, Yoon GL, Jin HY, Lee JH. Reliability analysis of laterally loaded piles for
[59] Little RL, Briaud JL. Full scale cyclic lateral load tests on six single piles in sand,
an offshore wind turbine support structure using response surface methodology,
Tech. rep. Texas A and M Univ College Station Dept of Civil Engineering; 1988.
Wind and Structures. Int J 2015;21(6):597–607.
[60] Lin SS, Liao JC. Permanent strains of piles in sand due to cyclic lateral loads.
[23] Yi JH, Kim SB, Yoon GL, Andersen LV. Natural frequency of bottom-fixed offshore
J Geotech Geoenviron Eng 1999;125(9):798–802.
wind turbines considering pile-soil-interaction with material uncertainties and
[61] Achmus M, Kuo YS, Abdel Rahman K. Behavior of monopile foundations under
scouring depth. Wind Struct 2015;21(6):625–39.
cyclic lateral load. Comput Geotech 2009;36(5):725–35.
[24] Haldar S, Sharma J, Basu D. Probabilistic analysis of monopile-supported offshore
[62] Jonkman J, Butterfield S, Passon P, Larsen T, Camp T, Nichols J, Azcona J,
wind turbine in clay. Soil Dynam Earthq Eng 2018;105:171–83.
Martinez A. Offshore code comparison collaboration within IEA wind annex XXIII:
[25] Teixeira R, Nogal M, O’Connor A, Nichols J, Dumas A. Stress-cycle fatigue design
phase II results regarding monopile foundation modeling, Tech. rep. Golden, CO
with kriging applied to offshore wind turbines. Int J Fatig 2019;125:454–67.
(United States): National Renewable Energy Lab.(NREL); 2008.
[26] Velarde J, Kramhøft C, Sørensen JD. Global sensitivity analysis of offshore wind
[63] NWTC information portal (FAST v8) [Last modified 04-January-2018, https://
turbine foundation fatigue loads. Renew Energy 2019;140:177–89.
nwtc.nrel.gov/FAST8. [Accessed 4 December 2019].
[27] Jimenez-Martinez M. Fatigue of offshore structures: a review of statistical fatigue
[64] T. Fischer, W. De Vries, B. Schmidt, UpWind design basis (WP4: offshore
damage assessment for stochastic loadings. Int J Fatig 2020;132:105327.
foundations and support structures), Upwind.
[28] Jonkman JM, Buhl Jr ML, et al. Fast user’s guide, Tech. rep. Golden, CO (United
[65] The MathWorks inc, version 9.6, ordinary differential equation toolbox.
States): National Renewable Energy Lab.(NREL); 2005.
Massachusetts: Natick; 2019.
[29] E. Bossanyi, GH bladed user manual, garrad hassan and partners ltd 701.
[66] DNV-OS-B101, Metallic materials, Det Norske Veritas.
[30] J. De Vaal, T. Nygaard, 3DFloat user manual, Institute for Energy Technology,
[67] DNVGL-Rp-0005:2014-06, Rp-C203: Fatigue design of offshore steel structures, Det
Norway.
Norske Veritas.
[31] Larsen TJ, Hansen AM. How 2 HAWC2, the user’s manual. December 2007.

20