Wind Energ. Sci.

, 5, 1521–1535, 2020
https://doi.org/10.5194/wes-5-1521-2020
© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

Reliability analysis of offshore wind turbine

foundations under lateral cyclic loading
Gianluca Zorzi1 , Amol Mankar2 , Joey Velarde3 , John D. Sørensen2 , Patrick Arnold1 , and Fabian Kirsch1
1 GuD Geotechnik und Dynamik Consult GmbH, 10589 Berlin, Germany
2 Department of the Built Environment, Aalborg University, Aalborg, 9100, Denmark
3 COWI A/S, Aarhus, 8000, Denmark

Correspondence: Gianluca Zorzi (zorzi@gudconsult.de)

Received: 20 August 2019 – Discussion started: 3 September 2019

Revised: 28 July 2020 – Accepted: 29 September 2020 – Published: 10 November 2020

Abstract. The design of foundations for offshore wind turbines (OWTs) requires the assessment of long-term
performance of the soil–structure interaction (SSI), which is subjected to many cyclic loadings. In terms of ser-
viceability limit state (SLS), it has to be ensured that the load on the foundation does not exceed the operational
tolerance prescribed by the wind turbine manufacturer throughout its lifetime. This work aims at developing
a probabilistic approach along with a reliability framework with emphasis on verifying the SLS criterion in
terms of maximum allowable rotation during an extreme cyclic loading event. This reliability framework allows
the quantification of uncertainties in soil properties and the constitutive soil model for cyclic loadings and ex-
treme environmental conditions and verifies that the foundation design meets a specific target reliability level. A
3D finite-element (FE) model is used to predict the long-term response of the SSI, accounting for the accumu-
lation of permanent cyclic strain experienced by the soil. The proposed framework was employed for the design
of a large-diameter monopile supporting a 10 MW offshore wind turbine.

1 Introduction that the soil’s bearing capacity withstands the lateral loading
of the pile within the allowable deformations (i.e. pile deflec-
tion and pile rotation at the mud-line).
Offshore wind turbines are slender and flexible structures Subsequently, the pre-design is checked for the cyclic
which have to withstand diverse sources of irregular cyclic load. The verification of the pre-design for the cyclic load de-
loads (e.g. winds, waves and typhoons). The foundation, sign step regards three limit states: ULS, SLS and FLS. The
which has the function of transferring the external loads to cyclic stresses transferred to the soil can reduce the lateral re-
the soil, must resist this repeated structural movement by sistance by means of liquefaction (ULS); can change the soil
minimizing the deformations. stiffness which can cause resonance problems (FLS); and can
The geotechnical design of the foundation for an offshore progressively accumulate deformation into the soil, leading
wind turbine (OWT) has to follow two main design steps to an inclination of the structure (SLS). If one of these limit
named static load design (or pre-design) and cyclic load de- states is not fulfilled, cyclic loads are driving the design and
sign. A design step is mainly governed by limit states: i.e. the the foundation dimensions should be updated.
ultimate limit state (ULS), the serviceability limit state (SLS) Performing the checks for the cyclic load design step is
and the fatigue limit state (FLS). The design of an offshore very challenging due to the following: (i) a high number of
structure mostly starts with the static load design step in cycles is usually involved; (ii) soil subjected to cyclic stresses
which a loop between the geotechnical and structural engi- may develop non-linearity of the soil response, pore water
neers is required to converge to a set of optimal design di- pressure, changing in stiffness, and damping and accumula-
mensions (pile diameter, pile length and can thickness). This tion of soil deformation (Pisanò, 2019); (iii) the load char-
phase is governed by the ULS in which it must be ensured

Published by Copernicus Publications on behalf of the European Academy of Wind Energy e.V.
1522 G. Zorzi et al.: Reliability analysis of offshore wind turbine foundations under lateral cyclic loading

acteristics such as frequency, amplitude and orientation are – Some reasons are due to aesthetics.
continually varying during the lifetime; (iv) characteristic
In SLS designs, extreme and relevant accidental loads, such
of the soil such as type of material, porosity and drainage
as typhoons and earthquakes, should be accounted for as they
condition can lead to different soil responses; (v) the rele-
can be design-driving loads. A very strict tilting requirement,
vant codes (BSH, 2015; DNV-GL, 2017) do not recommend
i.e. 0.25◦ , in conjunction with these accidental conditions can
specific cyclic load methods for predicting the cyclic load
increase the foundation dimensions and significantly raise
behaviour of structures, which leads to the development of
the cost of the foundation.
various empirical formulations (Cuéllar et al., 2012; Hettler,
An advanced numerical method called the soil cluster
1981; LeBlanc et al., 2009) or numerically based models
degradation (SCD) method was developed (Zorzi et al.,
(Zorzi et al., 2018; Niemunis et al., 2005; Jostad et al., 2014;
2018). This method explicitly predicts the cyclic response
Achmus et al., 2007). Despite the different techniques used in
of the soil–structure interaction (SSI) in terms of the foun-
these models, they all predict the soil behaviour “explicitly”,
dation rotation. The main objective of this study is to use the
based on the number of cycles instead of a time domain anal-
SCD method within a probabilistic approach. The probabilis-
ysis (Wichtmann, 2016). Time domain analysis for a large
tic approach along with the reliability framework was used to
number of cycles is not convenient due to the accumulation
quantify the main uncertainties (aleatoric and epistemic), ex-
of numerical errors (Niemunis et al., 2005).
plore which uncertainty the response is most sensitive to and
In common practice due to the non-trivial task faced by the
de-sign the long-term behaviour of the foundation for a spe-
engineers, simplifications and hence introduction of uncer-
cific target reliability level. In this paper, first the developed
tainties and model errors are often seen. The application of
reliability-based design (RBD) framework is outlined in de-
probabilistically based methods for designing offshore foun-
tail. Then, an application of the proposed RBD framework is
dations is not a new topic (Velarde et al., 2019, 2020; Car-
presented for a large-diameter monopile supporting a 10 MW
swell et al., 2014), and it is mainly related to the static de-
offshore wind turbine.
sign stage. Very limited research has been developed regard-
ing the probabilistic design related to the cyclic load design
stage. 2 Development of the RBD framework
This current work focuses on the cyclic loading design
stage and the verification of the serviceability limit state. 2.1 Limit state function for SLS
During the design phase, the wind turbine manufacturers pro-
vide a tilting restriction for operational reasons. The recom- The rotation experienced by the foundation structure sub-
mended practice DNV-GL-RP-C212 (DNV-GL, 2017) pro- jected to cyclic loading is considered partially irreversible
vides the order of magnitude for the maximum allowed tilt- (irreversible serviceability limit states) because the soil de-
ing of 0.25◦ throughout the planned lifetime. This strict ver- velops an accumulation of irreversible deformation due to
ticality requirement may have originated from different de- the cyclic loading action. For this reason, it is noted that the
sign criteria, which, however, are mainly rooted within the accidental and environmental load cases for the SLS design
onshore wind turbine sector and are given below (extracted are the extreme loads that give the highest rotation. As for
from Bhattacharya, 2019). a deterministic analysis, the first step in the reliability-based
analysis is to define the structural failure condition(s). The
– Blade–tower collision. Owing to an initial deflection of term failure signifies the infringement of the serviceability
the blades, a possible tilting of the tower may reduce the limit state criterion, which is here set to a tilting of more than
blade–tower clearances. 0.25◦ . The limit state function g(X) can then be written as
– Reduced energy production. Change in the attack angle g(X) = θmax − θcalc (X), (1)
(wind blades) may reduce the total energy production.
where θmax = 0.25◦ is the maximum allowed rotation and
– Yaw motors and yaw breaks. Reduce motor capacity for θcalc (X) is the predicted rotation (i.e. the model response)
yawing into the wind. based on a set of input stochastic variables X.
– Nacelle bearing. A tilted nacelle may experience differ-
ent loadings in the bearing, causing a reduction of their 2.2 Estimation of the probability of failure
fatigue life or restriction of their movements.
The design has to be evaluated in terms of the probability
– Fluid levels and cooling fluid movement can vary. of failure. The probability of failure is defined as the prob-
ability of the calculated value of rotation θcalc (X) exceeding
– P − δ effect. The mass of the rotor–nacelle assembly is the maximum allowed rotation θmax as it does when the limit
not aligned with the vertical axis, and this creates an state function g(X) becomes negative, i.e.
additional overturning moment in the tower, foundation,
grouted connection and soil surrounding the foundation. Pf = P [g(X) ≤ 0] = P [θmax ≤ θcalc (X)] . (2)

Wind Energ. Sci., 5, 1521–1535, 2020 https://doi.org/10.5194/wes-5-1521-2020

G. Zorzi et al.: Reliability analysis of offshore wind turbine foundations under lateral cyclic loading 1523

Once the probability of failure is calculated, the reliability tests are performed with different combinations of cyclic am-
index β is estimated by taking the negative inverse standard plitude and average load for N number of cycles. These di-
normal distribution of the probability of failure: agrams provide a 3D relation between the stress level and
number of cycles for an investigated variable: accumulation
β = 8−1 (Pf ) , (3) of strain, pore pressure, soil stiffness or damping. The cyclic
contour diagrams have been applied successfully for many
where 8( ) is the standard normal distribution function. The years for the design of several offshore foundations (Jostad
probability of failure in this work is estimated using the et al., 2014; Andersen, 2015); however careful engineering
Monte Carlo (MC) simulation. For each realization, the MC judgement is required for the construction and interpretation.
simulation randomly picks a sequence of random input vari- The loading input for the model must be a design storm
ables, calculates the model response θcalc (X) and checks if event simplified in a series of regular parcels. This load-
g(X) is negative (Fenton and Griffiths, 2008). Thus, for a to- ing assumption is also recommended by DNV-GL-RP-C212
tal of n realizations the probability of failure can be com- (DNV-GL, 2017) and the BSH standard (BSH, 2015). The
puted as method is implemented in the commercial code PLAXIS 3D
nf (PLAXIS, 2017).
Pf = , (4) Three stochastic input variables (X = [X1 X2 X3 ]) are nec-
n
essary for the SCD model:
with nf being the number of realizations for which the limit
state function is negative (rotation higher than 0.25◦ ). – X1 is soil stiffness that is derived from the cone pene-
IEC 61400-1 (IEC, 2009) sets as a requirement with regard tration test (CPT),
to the safety of wind turbine structures an annual probability
– X2 is the cyclic contour diagram that is derived from the
of failure equal to 5×10−4 (ULS target reliability level). This
cyclic laboratory tests, and
reliability level is lower than the reliability level indicated in
the Eurocodes EN1990 for building structures where an an- – X3 denotes extreme environmental loads that are de-
nual reliability index equal to 4.7 is recommended. Usually, rived from metocean data and a fully coupled aero-
in the Eurocodes, for the geotechnical failure mode consid- hydro-servo-elastic model.
ered in this paper the irreversible SLS is used. In EN1990
Annex B, an annual target reliability index for irreversible These inputs have to be quantified in terms of their point
SLS equal to 2.9 is indicated, corresponding to an annual statistics (e.g. the mean, standard deviation and probability
probability of failure of 2 × 10−3 . distribution type) representing the uncertainties. When us-
IEC61400-1 does not specify the target reliability levels ing the MC simulation, 100/pf realizations are needed to
for the SLS condition. Therefore, it can be argued that the estimate an accurate probability of failure, which makes it
target for SLS in this paper should be in the range of 5 × challenging to apply it in combination with the FE simula-
10−4 –2 × 10−3 . In this work, the same reliability target for tions. Since the SCD model is based on 3D FE simulations,
ULS of 5 × 10−4 is also considered for the irreversible SLS it is computationally intensive and hence expensive to com-
as a conservative choice. plete a large number of realizations. One FE simulation takes
approximately 30–40 min. For this reason, a response sur-
2.3 Derivation of the model response θcalc
face (RS) is trained in such a way that it yields the same
model response θcalc as the SCD model for the studied range
The calculation of the model response θcalc is based on the of the input variables X. The response surface is a function
soil cluster degradation (SCD) model. The SCD method ex- (usually first- or second-order polynomial form) which ap-
plicitly predicts the long-term response of an offshore foun- proximates the physical or FE models but allows the relia-
dation accounting for the cyclic accumulation of permanent bility assessment of the investigated problem with resealable
strain in the soil. The SCD model is based on 3D finite- computational effort.
element (FE) simulations, in which the effect of the cyclic The design of experiment (DoE) procedure is used to ex-
accumulation of permanent strain in the soil is considered plore the most significant combinations of the input vari-
through the modification of a fictional elastic shear mod- ables X. Based on the developed FE simulation plan, the ob-
ulus in a cluster-wise division of the soil domain. A simi- tained outputs θcalc are used to fit the response function.
lar approach of reducing the stiffness in order to predict the Figure 1 summarizes the methodology for the reliability
soil deformation can be found in Achmus et al. (2007). The analysis design for lateral cyclic loading. The framework
degradation of the fictional stiffness is implemented using a starts with the uncertainty quantification from the available
linear-elastic Mohr–Coulomb model. Reduction of the soil data (CPT, cyclic laboratory tests of the soil, and metocean
stiffness is based on the cyclic contour diagram framework and aero-hydro-servo-elastic model) and the derivation of the
(Andersen, 2015). The cyclic contour diagrams are derived stochastic input variables (soil stiffness, cyclic contour dia-
from a laboratory campaign using cyclic test equipment. The gram and storm event). The chosen stochastic variables are

https://doi.org/10.5194/wes-5-1521-2020 Wind Energ. Sci., 5, 1521–1535, 2020

1524 G. Zorzi et al.: Reliability analysis of offshore wind turbine foundations under lateral cyclic loading

Figure 1. Methodology of reliability analysis.

the inputs of the SCD model. Based on the stochastic input 3 Case study: reliability design for a monopile
variables, a response surface is then trained to yield the same supporting a 10 MW wind turbine
output (in terms of structural tilting) of the 3D FE simula-
tions. The response surface is then used to calculate the prob- In this section, firstly, the monopile pre-design (static load
ability of failure passing through the formulation of the limit design step) is carried out in which the subsoil conditions of
state equation and the MC simulation. If the calculated prob- the case study and the ULS design of the monopile geometry
ability of failure does not meet the target probability, then supporting a 10 MW wind turbine are explained. The pre-
the foundation geometry has to be changed, and the method- design of the monopile is developed using the hardening soil
ology is repeated to check whether the new design is safe. model in finite-element model to predict the static response
of the monopile.

Wind Energ. Sci., 5, 1521–1535, 2020 https://doi.org/10.5194/wes-5-1521-2020

G. Zorzi et al.: Reliability analysis of offshore wind turbine foundations under lateral cyclic loading 1525

the ULS geotechnical verification of the preliminary de-

sign of the monopile is carried out, using the finite-element
method in PLAXIS 3D.
The monopile is modelled in PLAXIS as a hollow steel
cylinder using plate elements. For the steel, a linear elastic
material is assumed with a Young’s modulus of 200 GPa and
a Poisson coefficient of 0.3. The interface elements are used
to account for the reduced shear strength at the pile’s surface.
The soil model used is the hardening soil model with
small-strain stiffness (HSsmall) (PLAXIS, 2017). The hard-
ening soil model with small strain stiffness can predict the
non-linear stress–strain behaviour of the soil. It considers a
stress and strain stiffness dependency, can predict the higher
stiffness of the soil at small strain which is relevant for cyclic
loading condition, and distinguishes between loading and un-
loading stiffness.
On the other hand, the Mohr–Coulomb model approxi-
mates the complex non-linear behaviour of the soil by a
linear-elastic perfectly plastic constitutive law.
The soil model parameters for the two layers are derived
from the tip resistance (Fig. 2) and listed in Table 1. The rel-
Figure 2. CPT profile.
ative density (which is related to the soil porosity) of the two
layers is calculated using the formula from Baldi et al. (1986)
with the over-consolidated parameters (typical for offshore
Then the reliability framework for the cyclic load design conditions), leading to a mean value of 70 % and 90 % for
shown in Fig. 1 is applied to the monopile to check if the the first and second layers, respectively.
pre-design satisfies the SLS criteria. The following subsec- The monopile design requires a loop between the struc-
tions discuss the derivation of input uncertainties for the SCD tural and geotechnical engineers to update the soil stiffness
method, derivation of the response surface and probability of and loads at the mud-line level. A fully coupled aero-hydro-
failure, and reliability index calculation. servo-elastic model using HAWC2 (Larsen and Hansen,
2015) is developed to perform the time-domain wind tur-
3.1 Monopile pre-design: subsoil condition and pile bine load simulations (Velarde et al., 2020). The soil–
geometry structure interaction model is based on the Winkler-type ap-
proach, which features a series of uncoupled non-linear soil
For the present case study, a tip resistance from the cone pen- springs (so called p − y curves) distributed every 1 m. The
etration test (CPT) and the boring profile are used to deter- force (p) − deformation (y) relations are extracted from the
mine the geotechnical properties and soil stratigraphy at the PLAXIS 3D model. At each meteor section, the calculation
site, where the monopile is assumingly installed. A CPT is of the force (p) is carried out by integrating the stresses
basically a steel cone which is pushed into the ground and along the loading direction over the surface. The displace-
the tip resistance is recorded. Based on the recorded tip resis- ment (y) is taken as the plate’s displacement. The PISA
tance, soil stratigraphy and soil properties can be empirically project (Byrne et al., 2019) highlights that additional soil
derived. reaction curve components (distributed moment, horizontal
The CPT, shown in Fig. 2, features an increase in the tip base force and base moment) are needed in conjunction with
resistance with increasing depth, which is typical for sand. In the p − y curves in order to have a more accurate soil struc-
combination with the borehole profile, the tip resistance from ture interaction behaviour. For the sake of simplicity, only the
the CPT suggests that the soil can be divided into two differ- p − y curves extracted from the FE model are considered.
ent layers. At approximately −10 m there is a jump in the tip The final pile design consists of an outer pile diameter at
resistance marking a transition to another layer with a higher the mud-line level of 8 m a pile thickness of 0.11 m, and a
magnitude visible, leading to the conclusion that denser sand pile embedment length of 29 m. The natural frequency of the
is present. The characterization of the soil extracted from the monopile is 0.20 Hz and is designed to be within the soft-stiff
boreholes shows the first layer (from 0 to −10 m) consisting region. Fatigue analysis of the designed monopile is also car-
of fine to medium sand and the second layer (from −10 m) ried out (Velarde et al., 2020). Figure 3a shows the horizontal
consisting of well-graded sand with fine gravel. displacement contour plot at 3.5 MN horizontal force, while
To accurately predict the soil–structure interaction and in- Fig. 3b shows the horizontal load-rotation curve at the mud-
corporate the rigid behaviour of the large-diameter monopile, line.

https://doi.org/10.5194/wes-5-1521-2020 Wind Energ. Sci., 5, 1521–1535, 2020

1526 G. Zorzi et al.: Reliability analysis of offshore wind turbine foundations under lateral cyclic loading

Table 1. Soil model parameters.

Soil Parameter Value Soil Parameter Value

Fine– E50 (MPa) 33.3 Medium- E50 (MPa) 98.3
medium Eoed (MPa) 33.3 coarse sand Eoed (MPa) 98.3
sand Eur (MPa) 99.9 Eur (MPa) 295
m (–) 0.5 m (–) 0.5
Depth: c (kN m−2 ) 0.1 Depth: from c (kN m−2 ) 0.1
from 0 to ϕ (◦ ) 39 −10 m ϕ (◦ ) 42
−10 m ψ (◦ ) 9 ψ (◦ ) 12
G0 (MPa) 116 G0 (MPa) 196.6
Relative γ0.7 (–) 0.0001 Relative γ0.7 (–) 0.0001
density: density:
70 % 90 %

Figure 3. (a) Horizontal displacement contour plot at 3.5 MN horizontal load; (b) monopile rotation.

3.2 Input uncertainties for the SCD model from the CPT test (Fig. 2). The layering of the soil domain is
assumed to be deterministic as explained in Sect. 3.1.
The application of the SCD model requires three inputs – The design tip resistance is established by means of the
soil stiffness (for the Mohr–Coulomb soil model), cyclic con- best-fit line in the data. A linear model is fitted to the data
tour diagrams and a design storm event. The laboratory test- for each layer (Fig. 4, green line). The maximum likelihood
ing and field measurements are used to estimate the inputs estimation (MLE) is used for estimating the parameters of
for the model. In this estimation process, different sources the linear model along with the fitting error (assumed to be
of uncertainty of unknown magnitude are introduced (Wu normally distributed and un-biased). From the MLE method,
et al., 1989). These parameters then have to be modelled as the standard deviations and correlations of the estimated pa-
stochastic variables with a certain statistical distribution. rameters (Sørensen, 2011) are obtained. The linear model is
expressed by means of Eq. (5) as below:
3.2.1 Soil stiffness
qc = Xa z + Xb + ε, (5)
The uncertainties of the soil stiffness used in the SCD model
are analysed. The soil model employed in the SCD method is where Xa and Xb are stochastic variables modelling param-
the Mohr–Coulomb model, with a stress-dependent stiffness eter uncertainty related to the parameters a and b, respec-
(i.e. the stiffness increases with depth). For cyclic loading tively; ε is the fitting error; and z is the depth (m). Table 2
problems, the unloading–reloading Young’s modulus Eur is shows a summary of the fitting parameters.
used. This soil modulus is obtained from the tip resistance

Wind Energ. Sci., 5, 1521–1535, 2020 https://doi.org/10.5194/wes-5-1521-2020

G. Zorzi et al.: Reliability analysis of offshore wind turbine foundations under lateral cyclic loading 1527

highly dependent on the soil, stress history, relative density,

effective stress level and other factors (Lunne et al., 1997;
Bellotti et al., 1989; Jamiolkowski et al., 1988).
To understand the uncertainty in the stiffness modulus, α is
treated as a stochastic normal variable varying from αmin = 3
to αmax = 8 with a mean µ = 5.5 and standard deviation σ =
1.25. The standard deviation is calculated by (αmin −αmax )/4,
assuming that 95.4 % of the values are enclosed between the
α values of 3 and 8.
Thus, the calculation of the drained constraint modulus in
unloading–reloading, covering all possible uncertainties, is
summarized as follows:

Es = Xα [Xa z + Xb + ε] . (7)

Depending on the size of the foundation, the local fluctuation

(physical uncertainty) of the tip resistance can have a signif-
icant impact on the structural behaviour. If the size of the
foundation is large enough, the soil behaviour is governed
by the average of the global variability of the tip resistance
(mean trend value). For a smaller foundation, the local ef-
fect, i.e. the local physical variability of the tip resistance,
Figure 4. Average tip resistance. governs the soil behaviour. If the local variability of the tip
resistance does not affect the foundation behaviour compared
Table 2. Stochastic input variable for tip resistance. to the fitted linear model, it can be neglected. Moreover, the
uncertainty related to the empirical formulation for calculat-
Parameter Distribution Mean Standard ing the soil stiffness (Xα ) has a higher influence compared
deviation to the one used to approximate the tip resistance with a lin-
Xa (first layer) Normal −0.42 0.049 ear model (Xa Xb ε). The preliminary results show that the
Xa (second layer) Normal −0.53 0.024 uncertainty associated with approximating the tip resistance
Xb (first layer) Normal 6.35 0.28 with the mean trend line is negligible due to the size of the
Xb (second layer) Normal 34.05 0.72 monopile. For this reason, Xa Xb ε values are considered de-
ε (first layer) Normal 0 3.14 terministic at their mean value.
ε (second layer) Normal 0 16.06 (MPa) Figure 6 shows the variability of the soil modulus Es over
ρXa ,Xb (first layer) – 0.86 – depth. The red lines are the realizations, using the MC simu-
ρXa ,Xb (second layer) – 0.98 – lation by performing random sampling on the stochastic vari-
able Xα . The black points are the deterministic multiplication
of the tip resistance with a mean value of α = 5.5.
The residuals are then plotted to check the assumption of The drained constraint modulus in unloading–reloading
the normality of the model error. For the first layer (Fig. 5a), Es is then converted to the drained triaxial Young’s modu-
the distribution of the residual is slightly skewed to the right. lus in unloading–reloading Eur used in the Mohr–Coulomb
This means that the trend line under-represents the tip resis- soil model in PLAXIS. Assuming an elastic behaviour of the
tance due to the presence of high peaks at the boundary layer. soil during unloading–reloading, Es and Eur can be related
For the second layer (Fig. 5b), a normal distribution about the as
zero mean is visible, implying that a better fit is achieved.
(1 − υur )
An empirical linear relationship is used to calculate Eur = Es , (8)
the drained constraint modulus in unloading–reloading Es (1 + υur ) · (1 − 2 · υur )
(Lunne et al., 1997; Lunne and Christoffersen, 1983): where υur is the Poisson ratio (= 0.2).
The soil stiffness depends on the depth. In the Mohr–
Es = Xα qc , (6) Coulomb model, a linear increase in the stiffness with depth
is accounted for using the following formula:
where Xα is a unitless stochastic variable. For over-
consolidated sand, which is typical of offshore conditions, E(z)ur = E(z)ref
ur + (zref − z) Einc , (9)
a value of α = 5 is recommended (Lunne and Christoffersen,
1983). However, there is no unique relation between the stiff- where E(z)ur is the Young’s modulus for unloading–
ness modulus and the tip resistance because the α value is reloading at a depth z, E(z)ref
ur is the Young’s modulus for

https://doi.org/10.5194/wes-5-1521-2020 Wind Energ. Sci., 5, 1521–1535, 2020

1528 G. Zorzi et al.: Reliability analysis of offshore wind turbine foundations under lateral cyclic loading

Figure 5. Histogram of residual for layer 1 (a) and layer 2 (b).

Figure 6. (a) Variability of the soil modulus Es over depth; (b) histogram of the soil stiffness at zref = 0 m; (c) histogram of the soil stiffness
at zref = −10 m.

unloading–reloading at a reference depth zref and Einc is the – for the first layer at zref = 0 (Fig. 6b): µEurref =
increment of the Young’s modulus. Using this equation for 32.25 MPa and σEurref = 7.06 MPa;
a given input value of Eur ref and the increment E , E can
inc ur
be derived at a specific depth below the surface and com- – for the second layer at zref = −10 m (Fig. 6c): µEurref =
pared to Es , as specified in the design soil profile. For all re- 196.90 MPa and σEurref = 43.14 MPa.
alizations of different soil stiffness values (Fig. 6 red lines),
ref and the increment E
Eur inc are calculated:
Other soil properties, such as specific weight, friction angle
and relative density are considered to be deterministic. A full

Wind Energ. Sci., 5, 1521–1535, 2020 https://doi.org/10.5194/wes-5-1521-2020

G. Zorzi et al.: Reliability analysis of offshore wind turbine foundations under lateral cyclic loading 1529

positive correlation between the two soil layer stiffness val-

ues is assumed.

3.2.2 Cyclic contour diagrams

The aim of the contour diagrams is to provide a 3D variation

in the accumulated permanent strain in the average stress ra-
tio (ASR), which is the ratio of the average shear stress to
the initial vertical pressure or confining pressure; the cyclic
stress ratio (CSR), which is the ratio of the cyclic shear stress
to the initial vertical pressure or confining pressure; and the
number of cycles (N ). An extensive laboratory test campaign Figure 7. Slice of the cyclic contour diagram.
is needed to have an accurate 3D contour diagram. The lab-
oratory campaign generally consists of carrying out different
regular cyclic load tests with different average and cyclic am- interpolated to create the final smooth 3D contour diagram.
plitude stresses for a certain number of cycles. This procedure and its validation are explained in Zorzi et
For this work, a series of undrained single-stage two-way al. (2019a).
cyclic simple shear tests were performed at the Soil Mechan- The power-law function can be written in the form of
ics Laboratories of the Technical University of Berlin. The Eq. (1).
tests were carried out on reconstituted soil samples. The sam-
ples were prepared by means of the air pluviation method. CSR = Xc N Xd + Xe + ε, (10)
The initial vertical pressure was 200 kPa and no pre-shearing
was considered. where Xd represents the shape of the curve, Xc is a scaling
The cyclic behaviour of the upper layer of sand was evalu- factor, Xe is the intersection with the CSR axis and ε is the
ated with samples prepared at a relative density of 70 %. For fitting error. Using the maximum likelihood method (MLM)
the lower layer sand, a 90 % relative density was used. Two- it is possible to fit the mathematical model and estimate the
way cyclic loading tests were carried out, testing different standard deviation of the fitting error and the standard devia-
combinations of ASR and CSR. All the tests were stopped tion of the parameters c and e. During the fitting procedure,
at 1000 cycles or at the start of the cyclic mobility phase. the shape parameter d is assumed fixed at −0.35 for the lower
For the results on the cyclic behaviour of various tests and layer and −0.50 for the upper layer.
relative densities, refer to Zorzi et al. (2019b). Based on the results of the fitting procedure, a standard
All the data extracted from the laboratory tests were as- deviation of the fitting error of 0.008 is chosen for the two
sembled in a 3D matrix (ASR, CSR, N), and a 3D interpola- diagrams for the two soils. The parameters c and e are con-
tion of the permanent shear strain (γp ) was created to map the sidered deterministic, as the associated standard deviation is
entire 3D space. The repeatability of the cyclic simple shear very low. Preliminary simulations show that the uncertainty
tests is an important aspect to consider in evaluating the un- of a and c derived from the MLM has less influence than the
certainties in the cyclic contour diagram. Cyclic simple shear uncertainty in the fitting error.
tests feature a low repeatability for dense sand, which can be It has to be noted that the fitting error, to some extent, re-
attributed to the relatively small specimen size used for test- flects the uncertainties of repeatability of the tests. Moreover,
ing (Vanden Bergen, 2001). This makes the cyclic tests sensi- the relative density of the soil samples is based on the empir-
tive to sample preparation, resulting in, for example, different ical relation applied to the tip resistance (Sect. 3.1). To ac-
initially measured relative densities, soil fabric and void ratio count for the uncertainty in the relative density, different sets
non-uniformities. of contour diagrams should have been derived from several
Owing to this variability of the test, a mathematical for- tests performed with soil samples at different relative densi-
mulation was fitted to the raw interpolation. For this rea- ties.
son, different two-dimensional slices (CSR vs. N) at differ- The contour diagrams for two different ASR slices are pre-
ent ASR values were extracted. Figure 7 represents a slice sented in Figs. 8 and 9 for the upper and lower layers, respec-
of ASR equal to 0.06. The different coloured points repre- tively.
sent the strain surfaces γp for different levels of deformation.
The raw interpolation of data and the uncertainty related to 3.2.3 Load uncertainty
the low sample repeatability of the tests cause an unrealis-
tic non-smooth shape of the strain surfaces. Therefore, each The load input parameter for the SCD model is characterized
slice is assumed to follow a power-law function (variation in by a regular loading package with a mean and cyclic am-
CSR as power of N) for different strain levels and then cali- plitude load and an equivalent number of cycles (hereafter
brated to fit the data. Finally, the calibrated strain surfaces are called “load inputs” for simplicity). In common practice, the

https://doi.org/10.5194/wes-5-1521-2020 Wind Energ. Sci., 5, 1521–1535, 2020

1530 G. Zorzi et al.: Reliability analysis of offshore wind turbine foundations under lateral cyclic loading

Figure 8. Cyclic contour diagram for the first layer.

Figure 9. Cyclic contour diagram for the second layer.

structural engineer provides the irregular history at mud-line

level by means of the aero-hydro-servo-elastic model. There-
fore, a procedure is needed to transform the irregular design
storm event to one single regular loading parcel. The envi-
ronmental load used for the cyclic loading design relies on
the chosen return period for the load. The statistical distri-
bution of the environmental loads is then based on different
return periods.
The design storm event is here defined as a 6 h duration of
the extreme load (also called the peak of the storm) (DNV-
GL, 2017). The underlying assumption in considering only
the peak is that most of the deformations, which the soil ex-
periences, happen at the peak of the storm. The considered
design load case is DLC 6.1 (IEC, 2009; BSH, 2015), i.e.
when the wind turbine is parked and yaw is out of the wind.
The ULS loads are considered for the cyclic load design.
To derive the irregular load history at the mud-line level, Figure 10. Environmental contour plot for extreme sea states (Ve-
the fully coupled aero-hydro-servo-elastic model is devel- larde et al., 2019).
oped in the wind turbine simulation tool, HAWC2. Based on
5-year in situ metocean data from the North Sea, the envi-
ronmental contours for different return periods are derived as The five design sea states for maximum wind speed are
shown in Fig. 10 (Velarde et al., 2019). The marginal extreme summarized in Table 3. To account for short-term variability
wind distribution is derived using the peak-over-threshold in the responses, 16 independent realizations are considered
method for wind speed above 25 m s−1 . Furthermore, it is as- for each design sea state.
sumed that maximum responses are given by the maximum Time-domain simulations provide an irregular force his-
mean wind speed and conditional wave height for each return tory of 10 min at the mud-line. To transform the 10 min irreg-
period (red point in Fig. 10). ular loading to a 6 h storm, each 10 min interval is repeated
36 times.

Wind Energ. Sci., 5, 1521–1535, 2020 https://doi.org/10.5194/wes-5-1521-2020

G. Zorzi et al.: Reliability analysis of offshore wind turbine foundations under lateral cyclic loading 1531

Table 3. Design sea state for maximum wind speed. Table 4. Gumbel parameters of the distribution for the load inputs.

Annual Return Wind Wave Wave Load α β µ σ σε

exceedance period speed height period Fa 1.092 0.113 1.158 (MN) 0.382 (MN) 0.0040
probability (yr) Uw Hs Tp Fcly 3.66 0.093 3.71 (MN) 0.347 (MN) 0.011
(q) (m s−1 ) (m) (s) Neq 329.75 70.08 370.2 (cycles) 9.49 (cycles) 0.024
0.63 1 37.4 3.17 7.95
0.10 10 44.5 4.10 8.84
0.02 50 50.6 4.90 9.54 To obtain a statistical distribution, the mean force, cyclic
0.01 100 53.3 5.24 9.83 amplitude force and equivalent number of cycles are plotted
0.002 500 59.4 6.04 10.44 versus the probability of non-exceedance for each return pe-
riod.
The black points in the three following figures are, respec-
tively, the mean load, cyclic amplitude and number of cycles
of the regular packages obtained from the previous procedure
and plotted vs. the probability of not exceedance for each re-
turn period. Assuming that for each return period the black
points have a normal distribution, the 0.50 fractile (red cir-
cles) and the 0.95 fractile (blue circles) are obtained.
The statistical distributions for the loads are derived by fit-
ting a Gumbel distribution to the 0.95 fractile values (NOR-
SOK, 2007). The MLM is employed to fit the cumulative
Gumbel distribution to the extreme response (blue circles).
The cumulative density function distribution is defined as
Figure 11. Rain flow matrix for 100-year-return-period wind speed. CDF(x) = exp(− exp(−(x − α)/β)), (11)
µ = α + β0.5772, (12)
The irregular load histories have to be simplified to one π
σ= β. (13)
equivalent regular package with a specific mean and cyclic 2.44
load amplitude and an equivalent number of cycles that lead
The Table 4 summarizes the parameters of distribution for the
to the same damage accumulation (accumulation of soil de-
three load inputs. The standard deviation of the fitting error
formation) as that of the irregular load series.
is small, marking a good fitting of the distribution function.
The following procedure is used (Andersen, 2015).
Looking at the distribution of Fig. 12a–c, larger 0.50 frac-
– The rain flow counting method is utilized to break down tiles (red circles) are present when increasing the return pe-
the irregular history into a set of regular packages with riod. This is more pronounced for the mean force and is ex-
different combinations of mean force Fa and cyclic am- pected because the higher the return period, the higher the
plitude force Fcly and number of cycles N. Figure 11 mean pressure on the wind turbine tower.
shows an example of the output from the rain flow The scatter for each return period is more significant when
counting. the return period increases. This can have different reasons;
for example, the “rare” storms with a lower probability of
– All the bins are ordered with increasing maximum occurrence could cause more non-linearity problems, vary-
force Fmax obtained from the sum of the mean and the ing the wave and wind speeds in the aero-hydro-servo-elastic
cyclic amplitude (Fmax = Fa + Fcly ). model. It could also depend on the model uncertainty in the
time domain simulations.
– 3D contour diagrams in conjunction with the strain ac-
The correlation coefficients ρ for the 0.95 fractile val-
cumulation method are then used to calculate the ac-
ues between the mean and cyclic loads and the equivalent
cumulation of deformation. After scaling the loads to
number of cycles are ρFa −Fcly = 0.77, ρNeq −Fcly = 0.81 and
shear stresses, the result of this procedure gives the
ρNeq −Fa = 0.85. The three coefficients mark a strong positive
equivalent number of cycles for the highest maximum
max , which in turn gives the same accumulation correlation between the three load inputs.
force Fmax
of deformation of the irregular load history.
3.2.4 Model error
This procedure is applied for all simulations with different
return periods. This type of error is difficult to estimate because it requires
the validation of the numerical error against different model

https://doi.org/10.5194/wes-5-1521-2020 Wind Energ. Sci., 5, 1521–1535, 2020

1532 G. Zorzi et al.: Reliability analysis of offshore wind turbine foundations under lateral cyclic loading

tests. In the case of the SCD model, this error arises due to
the simplification of the model for a much more complex
behaviour of the soil–structure interaction under cyclic load-
ing. The model error εmodel is estimated as a random variable
and multiplied to predict the structural tilting (Eq. 14). The
model error is assumed to be normally distributed with a uni-
tary mean and a coefficient of variation of 10 %. Ideally, this
model uncertainty should be quantified comparing the results
from the SCD model with several different test results. How-
ever, such a large number of tests is not feasible.

g(X) = θmax − εmodel θcalc (X) (14)

3.3 Derivation of the response surface

The stochastic variables are summarized in Table 5. For sim-

plicity, a full correlation between the soil stiffness of the two
layers and the loads is assumed.
Once the stochastic variables are defined, the 3D FE model
has to be substituted by a response surface.
The DoE is used to obtain the training point from the
FE simulation. As most of the variables are correlated, three
stochastic variables are considered: (i) the stiffness of the up-
per soil layer Eur , (ii) the fitting error of the cyclic contour
diagram CCDerr and (iii) the mean load Fa . The indepen-
dent input stochastic variables have the statistical distribution
shown in Table 5. For each factor, three different levels are
assumed: minimum value µ−2·σ , average value µ and max-
imum value µ + 2 · σ . A full factorial design in three levels is
implemented. Therefore, a total of 33 simulations are needed
to explore all possible combinations.
Based on visual inspection of the output from the 3D FE
model, a second-order polynomial function is fitted to the
sample data. The linear regression method is used to estimate
regression coefficients of the polynomial function. The fol-
lowing function is the outcome of the linear regression anal-
ysis:

θcalc = 0.248Fa − 0.007Eur Fa − 0.144Fa CCDerr

2
+ 0.0000746Eur Fa + εfit . (15)

An un-biased fitting error (ε) with normal distribution is as-

sumed and the estimate of residual standard deviation (σεfit )
is 0.0013. R-squared is a statistical measure of how close the
data are to the fitted regression line. For the fitted function,
the R-squared value is 0.9984, underlining a good fit of the
function to the data and hence the choice of the initial choice
of the second-order polynomial function.
Figure 13a shows the function at the CCDerr = 0 (the mean
value). The surface shows that at a lower soil stiffness and a
high force, a higher rotation of the monopile is reached. Val-
Figure 12. Distribution of the load inputs. ues higher than 0.25 are considered failures. The red points
are from the numerical simulations. The 3D plot (Fig. 13b)
shows the response surface for the mean value of the force
Fa = 1.158 MN. It is apparent that the fitting error for the

Wind Energ. Sci., 5, 1521–1535, 2020 https://doi.org/10.5194/wes-5-1521-2020

G. Zorzi et al.: Reliability analysis of offshore wind turbine foundations under lateral cyclic loading 1533

Table 5. Summary of the stochastic variables.

X Unit PDF µ σ CoV ρ

(%)
ref
Soil stiffness layer 1 Eur MPa Normal 32.25 7.06 21.9
ref 1
Soil stiffness layer 2 Eur MPa Normal 196.90 43.14 21.9
Cyclic contour diagram fitting error CCDerr – Normal 0 0.008 – –
Input load Fa MN Gumbel 1.158 0.382 32.9
Input load Fcly MN Gumbel 3.71 0.347 9.3 1
Input load Neq Cycles Gumbel 370.2 9.49 2.5

Figure 13. Response surfaces.

contour diagram is small and thus does not have a significant

influence on the results.

3.4 Reliability analysis

The limit state function is written as
g(X)= 0.25◦ − (0.248Fa − 0.007Eur Fa − 0.144Fa CCDerr

2
+0.0000746Eur Fa + εfit εmodel . (16)

A total of 107 MC simulations were performed by random

sampling of the input stochastic variables. This number was
the minimum required to keep the relative error of the re-
Figure 14. Sensitivity plot.
liability index lower than 1 %. The stochastic variables and
their probability distribution functions are given in Table 5.
The derivation of the design mean and standard deviation are
3.5 Sensitivity analysis
explained in Sect. 3.2.
With the analysed monopile design, the annual probabil- The sensitivity analysis of the stochastic input variables on
ity of failure is 2.7000 × 10−5 and the corresponding annual the reliability index is conducted by varying the coefficient
reliability index is 4.03. This means that the monopile meets of variation one at a time for each input (0.5 and 2 coeffi-
the target reliability index of 2.9–3.3 and is considered safe cient of variation, CoV). The inclination of dashed lines in
for long-term behaviour in terms of rotation accumulation for Fig. 14 marks the sensitivity of the stochastic variable. Mean
the design storm event. force Fa and the soil stiffness Eur both influence the reliabil-
ity index significantly more than the fitting error and numer-
ical model error do.

https://doi.org/10.5194/wes-5-1521-2020 Wind Energ. Sci., 5, 1521–1535, 2020

1534 G. Zorzi et al.: Reliability analysis of offshore wind turbine foundations under lateral cyclic loading

4 Conclusion Competing interests. The authors declare that they have no con-
flict of interest.
During the lifetime of wind turbines, storms, typhoons or
seismic action are likely to cause permanent deformation of
the structure owing to the accumulation of plastic strain in the Special issue statement. This article is part of the special issue
soil surrounding the foundation. The serviceability limit state “Wind Energy Science Conference 2019”. It is a result of the Wind
criteria require that the long-term structural tilting does not Energy Science Conference 2019, Cork, Ireland, 17–20 June 2019.
exceed the operational tolerance prescribed by the wind tur-
bine manufacturer (usually less than 1◦ ) with a specific target
reliability level. In this study, the SLS design for long-term Acknowledgements. This research is part of the Innovation
structural tilting is addressed within a reliability framework. and Networking for Fatigue and Reliability Analysis of Struc-
tures – Training for Assessment of Risk (INFRASTAR) project.
This framework is developed based on the 3D FE models for
This project has received funding from the European Union’s
the prediction of the SSI under cyclic loading. For the case Horizon 2020 Research and Innovation programme under the
study of a large monopile installed on a typical North Sea Marie Skłodowska-Curie grant agreement no. 676139. The labora-
environment, a reliability index of 4.03 was obtained. Sen- tory tests are provided by Chair of Soil Mechanics and Geotechnical
sitivity analysis also shows that uncertainties related to the Engineering of the Technical University of Berlin. The authors are
soil stiffness and the environmental loads significantly affect grateful for the kind permission to use those test results.
the reliability of the structure. For regions where assessment
against accidental loads due to typhoons is necessary, uncer-
tainty of the extreme environmental loads can increase by up Financial support. This research has been supported by the In-
to 80 %. Such load scenarios can significantly reduce the reli- novation and Networking for Fatigue and Reliability Analysis of
ability index and therefore become the governing limit state. Structures – Training for Assessment of Risk (INFRASTAR) (grant
A discussion has to be started in the offshore community no. 676139).
regarding the very strict tilting requirement (i.e. 0.25◦ ). This
very small operational restriction can lead to foundations of
Review statement. This paper was edited by Athanasios Kolios
excessively large dimensions, which are unfeasible from an
and reviewed by Federico Pisano and one anonymous referee.
economic point of view. On the other hand, a less strict verti-
cality requirement (which could be a function of the dimen-
sion and type of the installed wind turbine), for example an
angle of rotation of 1–3◦ , could lead to a smaller foundation References
size and still meet the safety requirements. For this reason,
by means of aero-elastic analyses, the investigation of the Achmus, M., Abdel-Rahman, K., and Kuo, Y.: Behaviour of large
position of the natural frequency of the whole system and fa- diameter monopiles under cyclic horizontal loading, in: Twelfth
tigue analysis should be carried out when a wind turbine is International Colloquium on Structural and Geotechnical Engi-
tilted at 1–3◦ . Allowing a less stringent tilting of the founda- neering, 10–12 December 2007, Cairo, Egypt, 2007.
tion can also be beneficial during the monopile installation. A Andersen, K. H.: Cyclic soil parameters for offshore foundation de-
small foundation dimension saves vessel and equipment cost, sign, in: Frontiers in Offshore Geotechnics III, ISFOG’2015, 10–
which contributes significantly to the overall cost reduction 12 June 2015, Oslo, Norway, 5–82, ISBN 978-1-138-02848-7,
of the foundation. 2015.
Baldi, G., Bellotti, R., Ghionna, V., Jamiolkowski, M., and
In this paper, a simplified model to calculate the perma-
Pasqualini, E.: Interpretation of CPTs and CPTUs; 2nd part:
nent rotation (the SCD method) is implemented. It is noted drained penetration of sands, in: Proc. 4th Int. Geotech. Semi-
that other models of varying complexity can also be used in nar, Singapore, 143–156, 1986
the proposed probabilistic framework. If new inputs are in- Bellotti, R., Ghionna, V. N., Jamiolkowski, M., and Robertson, P.
troduced, the respective uncertainties should be considered K.: Design parameters of cohesionless soils from in situ tests,
in the reliability calculation and the function for the response Transport. Res. Rec., 1235, 45–54, 1989.
surface should be adjusted accordingly. Bhattacharya, S.: Challenges in Design of Foundations for
Offshore Wind Turbines, Eng. Technol. Ref., 9, 1–9,
https://doi.org/10.1049/etr.2014.0041, 2019.
Code and data availability. The codes can be made available by BSH: Standard Design – minimum requirements concerning the
contacting the corresponding author. constructive design of offshore structures within the Exclu-
sive Economic Zone (EEZ), Federal Maritime and Hydro-
graphic Agency (BSH), Federal Maritime and Hydrographic
Author contributions. GZ, AM, JV and JDS designed the pro-
Agency (BSH), Hamburg, Rostock, 2015.
posed methodology. GZ prepared the manuscript with the contribu- Byrne, B. W., Burd, H. J., Zdravkovic, L., Abadie, C. N., Houlsby,
tions from all co-authors. G. T., Jardine, R. J. and Taborda, D. M. G.: PISA De-
sign Methods for Offshore Wind Turbine Monopiles, in: Off-

Wind Energ. Sci., 5, 1521–1535, 2020 https://doi.org/10.5194/wes-5-1521-2020

G. Zorzi et al.: Reliability analysis of offshore wind turbine foundations under lateral cyclic loading 1535

shore Technology Conference, 6–9 May 2019, Houston, Texas, Pisanò, F.: Input of advanced geotechnical modelling to the de-
https://doi.org/10.4043/29373-MS, 2019. sign of offshore wind turbine foundations, in: Proceedings of
Carswell, W. Arwade, S., DeGroot, D., and Lackner, the XVII ECSMGE-2019: Geotechnical Engineering founda-
M.: Soil–structure reliability of offshore wind tur- tion of the future, 1–6 September 2019, Reykjavik, Iceland,
bine monopile foundations, Wind Energy, 18, 483–498, https://doi.org/10.32075/17ECSMGE-2019-1099, 2019.
https://doi.org/10.1002/we.1710, 2014. PLAXIS: Plaxis 3D Reference Manual, edited by: Brinkgreve, R. B.
Cuéllar, P., Georgi, S., Baeßler, M., and Rücker, W.: On the quasi- J., Kumarswamy, S., Swolfs, W. M., and Foria, F., Plaxis B. V.,
static granular convective flow and sand densification around pile Delft, the Netherlands, 2017.
foundations under cyclic lateral loading, Granular Matter, 14, Sørensen, J. D.: Notes in Structural Reliability Theory and Risk
11–25, https://doi.org/10.1007/s10035-011-0305-0, 2012. Analysis, Aalborg University, Aalborg, 2011.
DNV-GL: DNVGL-RP-C212 – Offshore soil mechanics and Vanden Bergen, J. F.: Sand Strength Degradation within the
geotechnical engineering, DNV GL AS, 2017. Framework of Vibratory Pile Driving, PhD thesis, University
Fenton, G. and Griffiths, D.: Risk assessment in geotechnical Catholique de Louvain, Faculty of Applied Science Civil and En-
engineering, John Wiley & Sons, Inc., Hoboken, NJ, USA, vironmental Engineering Division, Louvain, 2001.
https://doi.org/10.1002/9780470284704, 2008. Velarde, J., Kramhøft, C., and Sørensen, J. D.: Reliability-based
Hettler, A.: Verschiebungen starrer und elastischer Gründungskör- Design Optimization of Offshore Wind Turbine Concrete Struc-
per in Sand bei monotoner und zyklischer Belastung, Institut für tures, in: 13th International Conference on Applications of
Bodenmechanik und Felsmechanik der Universität Fridericiana, Statistics and Probability in Civil Engineering, 26–30 May 2019,
Karlsruhe, 1981. Seoul, Korea, https://doi.org/10.22725/ICASP13.185, 2019.
IEC: IEC 61400-3 - Wind turbines Part 3: Design requirements for Velarde, J., Sørensen, J. D., Kramhøft, C., and Zorzi,
offshore wind turbines, International Electrotechnical Commis- G.: Fatigue reliability of large monopiles for off-
sion, Geneva, 2009. shore wind turbines, Int. J. Fatig., 134, 105487,
Jamiolkowski, M., Ghionna, V., Lancellotta, R., and Pasqualini, https://doi.org/10.1016/j.ijfatigue.2020.105487, 2020.
E.: New correlations of penetration tests for design prac- Wichtmann, T.; Soil behaviour under cyclic loading – experimental
tice, Int. J. Rock Mech. Mining Sci. Geomech., 27, A91, observations, constitutive description and applications, in: Ha-
https://doi.org/10.1016/0148-9062(90)95078-f, 1988. bilitation thesis, KIT – Karlsruher Institut für Technologie, Karl-
Jostad, H. P., Grimstad, G., Andersen, K. H., Saue, M., Shin, Y., sruhe, 2016.
and You, D.: A FE Procedure for Foundation Design of Offshore Wu, T., Tang, W., Sangrey, D., and Baecher, G.: Reliability of Off-
Structures – Applied to Study a Potential OWT Monopile Foun- shore Foundations-State of the Art, J. Geotech. Eng., 115, 157–
dation in the Korean Western Sea, Geotech. Eng. J., 45, 63–72, 178, https://doi.org/10.1061/(asce)0733-9410(1989)115:2(157),
2014. 1989.
Larsen, T. J. and Hansen, A. M.: How 2 HAWC2, the user’s man- Zorzi, G., Richter, T., Kirsch, F., Augustesen, A. H., Østergaard,
ual, Risø National Laboratory, Technical University of Denmark, M. U., and Sørensen, S. P.: Explicit Method to Account for
Roskilde, Denmark, 2015. Cyclic Degradation of Offshore Wind Turbine Foundations Us-
LeBlanc, C., Houlsby, G. and Byrne, B.: Response of stiff piles ing Cyclic Inter-action Diagrams, in: International Society of
in sand to long-term cyclic lateral loading, in: Geotechnique, Offshore and Polar Engineers, 10–15 June 2018, Sapporo, Japan,
Band 602, ICE Publishing, 2009. ISOPE-I-18-271, 2018.
Lunne, T. and Christoffersen, H. P.: Interpretation of Cone Zorzi, G., Kirsch, F., Richter, T., Østergaard, M., and Sørensen, S.:
Penetrometer Data for Offshore Sands, in: Offshore Tech- Validation of explicit method to predict accumulation of strain
nology Conference, 2–5 May 1983, Houston, Texas, during single and multistage cyclic loading, in: Proceedings of
https://doi.org/10.4043/4464-MS, 1983. the XVII ECSMGE-2019 – Geotechnical Engineering founda-
Lunne, T., Robertson, P. K., and Powell, J. J. M.: Cone Pene- tion of the future, 1–6 September 2019, Reykjavik, Iceland,
tration Testing in Geotechnical Practice, CRC Press, London, 2019a.
https://doi.org/10.1201/9781482295047, 1997. Zorzi, G., Kirsch, F., Richter, T., Østergaard, M., and Sørensen, S.:
Niemunis, A., Wichtmann, T., and Triantafyllidis, T.: A high-cycle Comparison of cyclic simple shear tests for different types of
accumulation model for sand, Comput. Geotech., 32, 245–263, sands, in: 2nd International Conference on Natural Hazards & In-
https://doi.org/10.1016/j.compgeo.2005.03.002, 2005. frastructure, 23–26 June 2019, Chania, Greece, 2019b.
NORSOK: NORSOK Standard N-003, Actions and action effects,
Standards Norway, Lysaker, Norway, 2007.

https://doi.org/10.5194/wes-5-1521-2020 Wind Energ. Sci., 5, 1521–1535, 2020