Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

Wave loads on offshore wind turbine foundations in

shallow water
Engineering models vs. refined flow modelling
Erik Damgaard Christensen Erik Asp Hansen
DHI DNV Wind Energy
edc@dhigroup.com erik.asp.hansen@dnv.com

Lars Yde Niels Jacob Tarp-Johansen


DHI DONG Energy
lay@dhigroup.com niels.jacob.tarp-johansen@risoe.dk

Helge Gravesen Mathilde Lindhardt Damsgaard


Vattenfall DONG Energy
helge.gravesen@grontmij-carlbro.dk matda@dongenergy.dk

Abstract: In the simplest case, normally used in the


industry, the wave loads are determined by use of
This study presents comparisons between two
the Morison formula using the undisturbed
methods to estimate wave loads on wind turbine
kinematics. Morison’s formula is used with
foundations. The analyses have been restricted
constant drag and inertia coefficients, and wave
to analyses of loads on mono-piles. The
forces at different levels are assumed to be
traditional method is based on a combination of
independent, which are both debatable
undisturbed wave kinematics and a Morison force
assumptions.
model. A refined model is based on CFD
techniques where the wave interaction with the
The two limitations of a) describing the wave field
structure is modelled directly. The comparisons
accurately and b) obtaining the loads faithfully,
are made for regular waves, irregular waves, and
can be met by use of a fully non-linear flow
breaking waves. The effect of scour holes/scour
model, incorporating the structure as part of the
protection has also been analysed.
modelling domain. In this study, such an
approach is taken, using the ‘NS3’ model, a finite
volume solver for the 3D Navier-Stokes
Keywords: Wave load, CFD, Morisons equations, allowing for a free surface through the
equation. volume of fluid (VOF) technique. Hereby wave
loads obtained by the approximate methods can
be evaluated, using the NS3 results as a
1 Introduction benchmark.

Frandsen et al. (2006) presents a list of Application of the NS3 code to wave loads and
possibilities of improving existing methods for the run-up on wind turbine foundations has been
design of Offshore Wind Turbines (OWT). The reported already in e.g. Christensen et al. (2005),
present paper deals only with the hydrodynamic and for influence of a scour / scour protection in
loads. OWT are often placed in fairly shallow Hansen et al (2007). NS3 has also been used for
waters which gives rise to significantly non-linear non-cylindrical foundations; see Bredmose et al.
(often breaking) waves. These waves are three- (2006).
dimensional irregular waves and can co-exist with
a current. A correct modelling of the associated The purpose of the study is to identify the cases
wave loads requires a faithful description of the where conventional engineering models are
kinematics. The complexity of this topic is sufficiently accurate, and conversely to identify
contrasted by the theories and methods, which the cases where more precise models are
are normally applied in the design process. The needed. Results of the different modelling
most widely applied methods make use of linear approaches will be examined with respect to
wave theory combined with “Wheeler stretching” kinematics and hydrodynamic loads. The study
from the mean water level to the instantaneous will focus on cylindrical foundations at relatively
sea surface position. Other methods use more shallow water. Both irregular waves as well as
complicated approaches involving higher-order regular waves in combination with a current are
Stokes or stream function wave models for the treated.
kinematics. These non-linear approaches,
however, are restricted to regular waves.
2 Design practice and rules Step 2 Determination of the undisturbed wave
and current kinematics at the tower centre
This section gives an overview of how the
hydrodynamic loads on vertical circular mono- From the undisturbed conditions determined in
piles exposed to wave-current action are step 1, (the conditions that would have been there
determined. At the end of the section the applied if the turbine wasn’t there), the following
assumptions / simplifications are discussed. hydrodynamic properties at the turbine centre are
determined:
The hydrodynamic loads are in principle
determined by completing the 3 steps procedure a) Wave elevation.
described below. b) Vertical distribution of the wave induced
velocities (in 2 horizontal directions )
c) Vertical distribution of the wave induced
Procedure acceleration (in 2 horizontal directions)
Step 1 Determination of the design basis for d) Vertical distribution of the wave induced
wave and current. pressure
e) Vertical distribution of the current (in 2
An offshore wind turbine has to be designed for horizontal directions)
the extreme conditions (ULS), a 50 year return
period is often used to design a turbine for 20 The current is normally characterised by the
years. Furthermore, the turbine has to be depth averaged current, and the velocity profile is
designed for accumulated fatigue damage (FLS) typically assumed to be constant, logarithmic, or
which it will experience during design lifetime. following a power law.

The ULS hydrodynamic loading depends on: Normally the kinematics is found based on the
assumption that the seabed is horizontal:
a) Water level (that may vary due to tide, surge,
wind set-up, and future sea level rise) For the FLS calculations the wave kinematics is
b) Bed level normally determined by a sum of linear waves
c) Current velocities and directions and the current. For the ULS calculations the
d) Waves, height, and associated period and wave kinematics is determined by a higher order
direction Stokes wave or by the stream function theory.

Normally a number of representative wave time Step 3: Determination of the Tower Loads
series are used to determine the fatigue damage, from the wave-current kinematics
each time series is described by
The horizontal hydrodynamic loads on mono-piles
1) The significant wave height, Hs are today determined from the relatively simple
2) The power spectral density (i.e. the form of Morison’s equation where the force intensity
wave spectrum) characterised by the peak (n/m) is the sum of the inertia and drag terms.
period, Tp, and one or more parameters. For r
example for the JONSWAP spectra by an r du 1 rr
f ( y , t ) = ρCM ⋅ A ⋅ + ρCD ⋅ D ⋅ u u
additional parameter, the peak enhancement dt 2
factor
3) Mean wave propagation direction r
where u( y , t ) is the velocity, as a function of
4) Directional wave spreading
the vertical coordinate and time (from step 2),
5) Water depth r
6) Current and current direction (if important) du
( y , t ) the acceleration, as a function of the
dt
In order to design the turbine the above vertical coordinate and time (from step 2), CM (y)
parameters including their internal correlation and inertia coefficient, determined theoretically or
the correlation with the wind speed and direction empirically, CD (y) drag coefficient, (determined
have to be predicted for the entire lifetime of the empirically), D(y) tower width, perpendicular to
turbine. the current, A(y) cross-section area in a vertical
section, ρ water density.
The parameters are normally determined from
measurements or from numerical hindcast The total force on the tower is found by
simulation, see for instance Jacobsen and integrating the force intensity from the seabed y=-
Rugbjerg (2005). De to the instantaneous water surface y=η.
r η r most of them have been conducted with a circular
F( t ) = ∫
−De
f ( y , t )dy cylinder in U-tubes or in towing tanks. Most of the
tests have been sinusoidal oscillations, where the
velocity and accelerations are constant along the
cylinder and vary only in time:
FLS, flow above mean water level. Wheeler
stretching
 2πt 
u( t ) = U w cos  
The FLS calculations are normally based on  T 
linear wave theory, i.e. the wave heights are du( t ) 2π  2πt 
assumed to be small; in such cases the total = U w sin  
dt T  T 
force is found by integrating the force from the
seabed to the mwl (y=0).
where U w is the orbital velocity, and T is the
r 0 r period of the oscillation.
F( t ) = ∫
−De
f ( y , t )dy
Dimensional analysis shows that for a cylinder
with smooth surface the force coefficients depend
In cases where the wave heights cannot be
on two parameters, the Keulegan-Carpenter
considered small, it is normal to use a stretching
number, and the Reynolds number.
method, for example the Wheeler stretching, see
ISO 19901-1, Wheeler stretching of the linear
wave kinematics to get the wave kinematics in TU w
KC =
the entire water column to the instantaneous D
water surface.
Uw D
y +D Re =
uWeeler ( y , t ) = u lin ( η+De De − De , t ) ν
e

y = [− De : η]
For a rough cylinder with roughness k s the
for
relative roughness becomes a parameter.

Slamming load ks
k s* =
D
The instantaneous force: the horizontal slamming Inserting the velocity and acceleration in the
load, per vertical length unit on this section and Morison equation one gets:
on underlying sections, which has not yet fully
penetrated the sloping water surface, can be r  2πA D  2πt   2πt   2πt  
calculated, according to DNV-OS-J101’, as: f ( t ) = ρUw2 D CM ⋅ 2 ⋅ sin  + ½CD ⋅ cos  cos 
 D TUw  T   T   T  

r rr Or written in a non-dimensional form


fslam ( y , t ) = ½ ρCs ⋅ D ⋅ u u
r
f (t )  2πt   2πt  2πA 1  2πt 
= ½C D ⋅ cos  cos  + CM ⋅ 2 ⋅ sin  
where C s is the slamming coefficient whose ρU w2 D  T   T  D KC  T 

value can be taken as: (for 0<s<D)


When the velocity and acceleration are known
together with the measured force, the
 D 0.107s 
Cs = 5.15 +  coefficients, CD and CM ,are determined by a least
 D + 19s D  square fit. The book by Sumer and Fredsøe
(1996) gives a good overview of the force
The penetration distance, s, is the horizontal coefficients variation, the book also addresses
distance from the periphery on the wet side of the the force coefficients in a combined wave and
cylinder to the sloping water surface, measured in current flow.
the direction of the wave propagation.
For practical engineering purposes, see e.g.
‘DNV-OS-J101’, the force coefficients are often
only taken to be a function of the KC number,
Determination of Force coefficients with a differentiation whether the tower surface is
rough or smooth. The DNV-OS-J101 coefficients
The force coefficients to be used in step 3 have are shown in figure 1 below. Even though the
been determined through physical experiments; influence of the roughness seems to be quite
important, it turns out that for sinusoidal flow for
KC number less than 10, the influence of the
1.8
roughness on the peak force is less than a few
per cent. This is illustrated in figure 2 where the 1.6

forces for KC number 10 are shown both for a

Fmax-rough/Fmax-smmoth
1.4
rough and a smooth cylinder. Figure 3 shows the 1.2
ratio between the max forces on a rough and a
1
smooth cylinder as a function of the KC number.
0.8

force coefficients j101 0.6


2
smooth 0.4
1.5 rough
0.2
CD

1
0
2 4 6 8 10 12 14 16 18 20
0.5
KC

0
0 10 20 30 40 50 60 70 80 90 100 Figure 3: Peak forces for KC number 10 for a
rough and smooth cylinder based on coefficients
2 presented in ‘DNV-OS-J101’.
smooth
1.5 rough
Wave forces by diffraction theory
CM

0.5 When the dimension of the structure in question


0 is large compared with the wave length, typically
0 10 20 30 40 50 60 70 80 90 100
KC
when D > 0.2L, Morison’s equation is no longer
valid. The inertia force will then be dominating
Figure 1; ‘DNV-OS-J101’ Force coefficients as a and can be predicted by means of diffraction
function of the KC number theory. For small waves the inertia coefficients
can be calculated by the MacCamy Fucks
KC=10 force variation j101
1 analytical solution.
0.8

0.6
smooth
rough 2
0.4

0.2
MacCamy Fuchs
Force

1.5 Morison
0

-0.2
CM

-0.4 1

-0.6

-0.8
0.5
-1
0 1 2 3 4 5 6 7 8 9 10
t

0
Figure 2: Force variation for KC number 10 for a 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
D/L
rough and a smooth cylinder based on
coefficients presented in ‘DNV-OS-J101’. Figure 4: Inertia coefficient, CM, calculated by
MacCamy Fucks as a function of ratio between
the diameter and the wave length.

Discussion

The 3 step procedure described above contains a


number of assumptions, engineering
approximations, and raises a number of practical
questions. The complicated task of establishing
the design waves and current (task 1) is not
treated here.
Step 2 The free surface geometry is governed by the
How are forces below the seabed level kinematic boundary condition where a particle on
determined if a scour hole has been developed the free surface follows the local fluid velocity.
around the turbine?
How are the kinematics affected by a possible The kinematic boundary condition is included by
scour protection placed on the seabed affecting extrapolation of the velocities within the fluid
the forces? domain to the surface and through use of the
How accurate is the ULS and FLS wave Volume of Fluid (VOF) that tracks the free
kinematics if the turbines are placed near or close surface. The present method is described in Hirt
to a sloping bed? And how will the forces be and Nichols (1981), however, with an improved
affected by local wave breaking etc.? scheme for the advection of the conservative
How accurate is the Wheeler stretching quantity, F, cf. Ubbink (1997).
compared with a full non-linear wave kinematics?
The dynamic boundary condition in the case of an
Step 3 invicid fluid is given as:
How shall the KC number be determined in case
of irregular waves? psurf = patm
How is the KC number determined above the
‘mwl’ (here the turbine is only exposed to forward The atmospheric pressure is set to zero.
directed flow)?
How is the KC number determined for a stream The spatial discretisation is based on the finite-
function wave, where the flow isn’t sinusoidal? volume approach on a multi-block grid. The time
How are the forces influenced by vertical integration of the Navier-Stokes equations is
velocities, horizontal and vertical gradients of the performed by application of the fractional step
horizontal velocities? (For small waves this can method. The turbulence is not fully resolved in the
be treated by the MacCamy Fucks analytical set-up in this study.
solution).
How well are the ULS determined, as they are
based on force coefficients based on regular
oscillations?

Of course it is much easier to raise the questions


than to answer them, however, the process of
listing questions and trying to answer them, may
help identifying uncertainties, inconsistencies,
and possible errors. The refined flow simulations
modelling presented by Hansen and Christensen
(2007) has addressed the scour hole and scour
protection with CFD analyses.

3 The use of CFD


The Navier-Stokes equations in three dimensions
are solved using a finite-volume approach on a
multi-block grid. The viscous forces are in the
present study neglected, which reduces the
Navier-Stokes to the Euler equation, which is Figure 5 The mono-pile (wind turbine foundation)
given below: with indication of mesh bounaries. The position of
the free surface is shown in blue.
∂u i
=0 Figure 5 shows an example of results from a
∂x i
numerical simulation of the multi-block grid used
∂u i ∂u i u j ∂p ∂  ∂u i ∂u j 
 for the study of run-up on wind turbine
ρ +ρ = ρg i − + µ +
∂t ∂x j ∂x i ∂x j  ∂x j ∂x i 

foundations. The grid consists of 12 blocks.
The ui are the three velocity components, g the The CFD code solving the Navier-Stokes
gravitational acceleration, p is the pressure, ρ the equations as sketched above has been used and
fluid density, and µ the dynamic viscosity. validated in Mayer et al. (1998), Nielsen and
Mayer (2001), Christensen et al. (2005),
Christensen (2006).

4 Comparison to standard
methods
The comparisons between the traditional
methods and the CFD method have been split up
in three different analyses. In all the cases the
mono-pile diameter is 4.2 m.

4.1 Regular waves


The first analyses are based on regular waves
and the objective is to study the interpretation of a
wave kinematics into loads on the structure. The
wave kinematics are described by a stream
function theory, where the current is defined in
terms of cS, i.e. net mass flux velocity.
Figure 6 Case 3: Comparison of horizontal force
It is not clear from the traditional guidelines how (upper panel), overturning moment (middle
scour holes, scour protection, and secondary panel), and the surface elevation at three
structures should be incorporated. In this study positions (green, at the mono-pile). Legends for
the depth has been changed to the total depth in the two upper panels: - - streamfunction +
the deepest part of the scour hole, and to the top
Morison equation,  CFD.
level of the scour protection when the traditional
methods are used. The effect of a secondary
structure has been incorporated by modifying the
force coefficients. The effect of a scour
hole/protection and secondary structures can be
incorporated directly in the CFD calculations,
wherefore no assumptions are necessary.

Table 1 Nine test cases based on regular waves.


Case h T H U Bed sec
[m] [s] [m] [m/s] form structure
1 9 4 2 0 plan no
2 15 4 2 0 plan no
3 15 8 7 0 plan no
4 15 8.6 8.1 0 plan no
5 15 8.6 8.1 1 plan no
scour
6 15 8.6 8.1 1 prot. no
scour
7 15 8.6 8.1 1 hole no
scour
8 15 8 7 1 prot. no
scour
9 15 8 7 1 hole no
Figure 7 Case 5: Comparison of horizontal force
Special cases (upper panel), overturning moment (middle
Scour protection: 20x20m area of 1 m height. panel), and the surface elevation at three
Slope of 1:2 from the top to the seabed. Scour positions (green, at the mono-pile). Legends for
hole: cone with depth 5.2 m at the mono-pile side the two upper panels: - - streamfunction +
and slope 30° to horizontal. Morison equation,  CFD.
Table 2 Nine test cases based on regular waves. The irregular analyses have been based on the
Max Horizontal Max Overturning following set-up:
Force [MN] Moment [MN] The depth at the wind turbine was 11.3 m. No
Case CFD Morison CFD Morison current which means that cS = 0, i.e. no net mass
1 0.26 0.27 1.55 1.62 flux velocity is set to zero in the present analyses.
2 0.26 0.28 2.96 3.13 Offshore wave condition: JONSWAP spectrum,
3 0.90 0.83 10.13 8.87 Hs=4.40m, Tp=9.44s. Wave condition at wind
4 1.19 0.97 15.35 11.30
turbine: Shoaling is insignificant.
5 1.17 1.10 14.63 13.10
A zero-crossing analysis of the horizontal force
6 1.20 1.11 15.85 12.94
and the overturning moment shows the
7 1.26 1.08 14.40 15.19 differences between the two models. For the
8 0.91 0.92 10.47 9.68 small load peaks the CFD loads are 6% and 18%
9 0.98 0.93 9.89 12.48 larger for the horizontal force and overturning
moment, respectively. However, the maximum
For small waves the difference between the horizontal force is predicted to be 38% larger in
traditional methods and CFD is rather small; in the CFD calculations compared to the traditional
the order of a few per cent. The largest relative methods, and for the overturning moment from
difference is found for case 4, which actually is a wheeler + Morison is 96% larger using CFD.
case with plane bed and no ambient current. The
st
main reason for the difference is the high non- It is well known that 1 order theory
linearity of the wave. Especially the rather steep underestimates the maximum forces. Therefore,
wave front has a major influence on the force. a part of the time series is often exchanged for a
The wave front or let’s say the effect of the free piece of stream function theory instead. However,
surface is not accounted for in the traditional it may be difficult to determine the part that will
method. give the largest load contribution. Figures 11 and
12 show 100 s sequences of the surface
The CFD results of scour hole/protection elevation predicted by the 1st order Wheeler
analyses show that the horizontal force method and horizontal force and overturning
decreases when going from scour hole to scour moment found from the Wheeler stretching +
protection, while the overturning moment Morison force and CFD, respectively. It is clear
increases. The traditional methods give the that the maximum forces are not found at the
completely opposite effect. same time, which of course makes it difficult to
st
choose correct 1 order wave to be exchanged
for wave kinematics obtained from stream
4.2 Irregular waves function theory.
Typical analyses of the dynamics of wind turbines
are based on 10 min. sequences of load from
wind, waves, and current. The CFD method uses
in this case a time series generated by a
Boussinesq model, MIKE21 BW. This gives a
good representation of the non-linear wave
kinematics for irregular waves.

The traditional method is based on Wheeler


stretching of 1st order Stokes wave theory. Using
st
1 order Stokes theory it is straightforward to
establish a time series of irregular waves. The
establishment of 1st order irregular time series is
based on the surface elevations used as
st
boundary condition for the CFD model. 512 1
order wave components have been found and
used to represent the incoming wave. If the wave
st
was close to 1 order the translation from the
boundary to the centre of the mono-pile (210 m) Figure 8 Comparison of max overturning
would be the same in CFD. However, due to non- moments found from zero-crossing analyses.
linearities the wave celerity is larger in the CFD
model, and therefore the waves arrive at the
cylinder in the order of 1 s earlier.
Force Moment
Wheeler CFD Wheeler CFD
[MN] [MN] [MNm] [MNm]
0.561 0.778 4.13 8.09
0.538 0.573 3.79 6.75
0.519 0.560 3.60 5.06
0.459 0.504 3.09 4.70
0.453 0.501 2.98 4.11
0.446 0.479 2.93 4.06

Figure 9 Comparison of max overturning


moments found from zero-crossing analyses.

The free surface and contours of the excess


pressure are shown in figure 10 at t = 592 s, i.e.
when the horizontal and overturning moment is
close at its maximum. Even though the modelled
wave loads are significantly larger using CFD the
wave surface is not in a state of breaking. This
shows that even for moderate waves the
simplifications in the use of Wheeler stretching +
Morisons equation may lead to significant Figure 10: The free surface and excess pressure
underestimation of the wave loads. at t = 592 s.

Table 3 Comparison of the 6 highest events of


horizontal force and overturning moment.
Figure 11: Sequence of horizontal force from the irregular timeseries: - - Wheeler stretching + Morison
equation,  CFD

Figure 12: Sequence of overturning moment from the irregular time series: - - Wheeler stretching +
Morison equation,  CFD

4.3 Shoaling waves over a slope an example of the horizontal force on the mono-
pile for the three cases. It is clear that just small
Test cases have also been performed for changes in the wave height may lead to
breaking waves over a slope. In this paper only significantly larger maximum wave loads.
an example of the effect of breaking waves is
shown. The water depth was 11.3 m and 12 s in
all three simulations, while the wave height was
7.8 m, 8.1 m, and finally 8.4 m. Figure 13 shows

Figure 13: An example of extreme wave loads on a mono-pile when the wave height is increased from 7.8
m to 8.4 m.
6 Summary Christensen, ED (2006): “Large eddy simulation
of spilling and plunging breakers”, Coastal
Engineering 2006, Vol. 53. pp 463-485.
Traditional methods, such as stream function
theory + Morison equation, have been used to
Christensen, E.D., Bredmose, H. and Hansen
calculate the forces on a mono-pile. The results
E.A. (2005). “Extreme wave forces and wave run-
have been compared to CFD results.
up on offshore windturbine foundations”.
Proceedings of Copenhagen Offshore Wind
The analyses of traditional force coefficients
2005, Copenhagen October 2005.
indicate that the effect of marine growth, i.e.
roughness, mainly has an effect on the phase
DNV-OS-J101. Design of Offshore Wind Turbine
rather than on the maximum force as long as the
Structures. June 2004. Det Norske Veritas.
KC number is below 10. In irregular waves it is
not straightforward to define the KC number.
Frandsen S., Tarp-Johansen, NJ, Hansen, EA.,
Høgedal, M., Ibsen, LB., Jensen, L., (2006):
The analyses of the regular waves show that
“Offshore Wind Turbine Design: Addressing
using a stream function + Morison equation gives
Uncertainty Drivers”, In proc. of 2006 EWEC
comparable results to the CFD calculations for
2006, pp 74-78.
small waves. The CFD method generally predicts
higher wave loads for high and steep waves.
Hansen, EA and Christensen, E.D. (2007): “Scour
holes or scour protection around offshore wind
The CFD results of scour hole/protection
turbine foundations: Effect on wave loads”. In
analyses show that the horizontal force
Proc. Of EWEC 2007.
decreases when going from scour hole to scour
protection, while the overturning moment
IEC 61400-3, ed.1 Design Requirements for
increases. The traditional methods give the
Offshore Wind Turbines. IEC TC88 WG3:
complete opposite effect.
Committee Draft, December 2005.
Irregular waves have been modelled by a
ISO 19901-1:2003, DIS: Specific requirements for
combination of 1st order wave theory of 512 wave
offshore structures – Part 1: Metocean design
components and Wheeler stretching. The
and operating conditions
Morison equation was used to predict the wave
loads from this wave kinematics. In general the
Jacobsen. V. and Rugbjerg, M (2005): “Offshore
horizontal force and the overturning moment was
wind farms – the need for Metocean Data”, In
underestimated in the order of 6 and 18%,
proceedings of Copenhagen Wind conference, 13
respectively. However, for the largest load the
pages.
difference was 38% and 96% compared to CFD
calculations. This clearly indicates that even for
MacCamy, R.C., and Fuchs R.A. (1954): “Wave
moderate waves the use of Wheeler stretching +
forces in piles: A diffraction theory”,. Technical
Morison equation significantly underestimates the
memorandum no. 69, Beach Erosion Board, US
wave loads on the mono-pile.
Navy Corps of Engineers.

Morison, J.R., O’Brien, M.P., Johnson, J.W., and


Acknowledgement Schaf, S.A.,(1950): “The forces exerted by
The financial support of the Danish Public Service surface waves on piles”, Petroleum Transactions,
Program (PSO) under contracts PSO: FU5101 AIME, Vol. 189, pp 149-157.
(ENS: 79030-0019, PSO-05) and PSO: 5876
(ENS: 79029-0030, PSO-04) is gratefully Sumer B.M. and Fredsøe, J. “Hydrodynamics
acknowledged. around Cylindrical Structures”, Advanced Series
on Ocean Engineering – Volume 12, World
References Scientific, London, 1997

Ubbink, O. (1997): ”Numerical prediction of two


Bredmose, H., Skourup, J., Hansen, E.A.,
Christensen, E.D., Pedersen, L.M. and Mitzlaff, A. fluid systems with sharp interfaces”, Ph.D. thesis,
(2006). Numerical reproduction of extreme wave University of London.
loads on a gravity wind turbine foundation. 25th
Int. Conf. Offshore Mech. Arctic Engng. Hamburg
2006. ASME

You might also like