You are on page 1of 6

Australasian Coasts & Ports 2019 Conference – Hobart, 10-13 September 2019

Numerical Prediction of Wave Excitation Forces on a Fixed Tension Leg Platform Concept for Offshore Wind Turbines
S. Khanam, N. Abdussamie, and R. Ojeda

Numerical Prediction of Wave Excitation Forces on a Fixed Tension Leg


Platform Concept for Offshore Wind Turbines
S. Khanam, N. Abdussamie, and R. Ojeda
National Centre for Maritime Engineering & Hydrodynamics, Australian Maritime College, University of
Tasmania, Newnham, Australia.
email: nagia@utas.edu.au

Abstract
To predict wave excitation forces on a fixed Tension Leg Platform Wind Turbine (TLPWT), two numerical
software packages were used, namely the CFD code STAR-CCM+ and ANSYS AQWA. The CFD code solved
the fully-nonlinear Navier-Stokes equations in a time domain whereas ANSYS AQWA solved the linear
diffraction problem in a frequency domain. In both domains, the horizontal wave excitation forces on a generic
TLP structure under several deterministic regular wave conditions of a mild sea state were predicted. The size
of the mesh/panel was fine-tuned to yield optimum results for both solvers. It was found that the horizontal
forces obtained by the linear diffraction solver were larger in magnitude than CFD predictions. The linear
diffraction solver provided an insight into the horizontal force experienced by the TLP structure, and hence it
could be used to verify CFD results. However, as the linear diffraction theory does not account for the effect
of turbulence, viscous effects or the effect of the air phase, CFD codes could be used to investigate such
phenomena and provide further details.

Keywords: Offshore wind turbines, tension leg platforms, excitation forces, numerical predictions.

1. Introduction Morison equation, Froude-Krylov and MacCamy &


The total capacity of offshore wind has increased Fuchs techniques, which all are based on potential
considerably in the last decade, with global capacity flow theory. Each method has its limitations and
reaching a recorded 19.27 gigawatts (GW) in 2017 conditions which need to be satisfied before it can
[1]. Many major countries are continuing to develop be employed. For instance, the Morison equation is
offshore wind technology. The current rate of only applicable when the wave length λ ≥ 5D where
development is only set to increase, with predictions D is the diameter or other projected cross-sectional
of up to 120 GW to be installed by 2030 [2]. The dimension of the member [6]. The Froude-Krylov
electricity generated offshore is transformed method is applicable to both slender and large
through cables buried under the sea floor to a structures. However, for this method to be
substation either onshore or near shore and is then applicable to large structures, wave reflection due to
delivered to the electrical grid. the presence of the structure should be minimal i.e.
The demand for offshore wind turbines at deeper no significant amount of flow separation from the
waters has recently increased [3]. There are few surface. Based on linear diffraction theory, the
types of offshore wind turbine structures that have MacCamy & Fuchs method can be used to obtain
been utilised based on the water depth. These wave forces acting on a finite vertical cylinder
include gravity base structures and monopiles length.
which are suitable for water depth up to 25 m, while The use of Computational Fluid Dynamic (CFD)
jacket type structures and tripods are suitable for based techniques has received an increasing
water depth of 20 m to 50 m [4]. amount of attention in recent years for predicting the
Full-scale prototypes of floating offshore wind hydrodynamic characteristics, wave-induced loads
turbines have been deployed including a SPAR and response of marine structures. However, CFD
buoy (Norway in 2009), followed by a semi- results must be validated with experimental data to
submersible (Portugal in 2011), and three more verify its accuracy. Nevertheless, the experimental
semi-submersibles (Japan in 2011-2015). Tension approach is costly, time-consuming and involves
Leg Platform (TLP) concepts have been used several other constraints such as scaling effects [7].
widely for supporting oil and gas topsides due to Alternatively, verification and validation of the CFD
their high stiffness of the tendon system. However, model results can be done through industry
TLP concepts are yet to be proven for supporting accepted software packages such as WAMIT and
offshore wind turbines [5]. The accurate prediction ANSYS AQWA.
of wave excitation forces on TLP wind turbines is The focus of this study is to utilise the commercial
among several design challenges. Wave excitation CFD code STAR-CCM+ and ANSYS AQWA to
forces are defined as the loads exerted on a establish a numerical procedure for predicting the
submerged structure held in equilibrium position in wave excitation forces acting on a fixed TLP
waves. offshore wind turbine support structure. Numerical
Commonly used approaches for estimating wave domains were created and the Volume of Fluid
excitation forces are simplified methods such as the (VOF) method was chosen to accurately capture the
Australasian Coasts & Ports 2019 Conference – Hobart, 10-13 September 2019
Numerical Prediction of Wave Excitation Forces on a Fixed Tension Leg Platform Concept for Offshore Wind Turbines
S. Khanam, N. Abdussamie, and R. Ojeda

interface between the air and water phases [8-14]. wave height of 3 m were tested (Table 2). The
The TLP dimensions and wave conditions were selection of such wave conditions (H versus T) was
based on the TLPWT given in [3, 5, 6]. The results on the basis of realistic sea states in the North Sea
of the horizontal forces obtained by the CFD code [6] (Hs = 3.0 and Tz = 7.5 s) such that H = 3 m (H/Hs
were compared with AQWA predictions. = 1.0) and T = 8 – 12 s (T/Tz = 1.1 – 1.6). Based on
the maximum wave period tested, the tank had a
2. Methodology length of 10 λ, a height of 200 m and a width of 5 m
The main dimensions of the TLP structure are with a constant (still) water depth of 150 m (λ =
summarised in Table 1. Figure 1 shows the main wavelength). The wave elevations were initialised to
components of the TLP structure in which its four be fully developed at 0.25 λ away from the centre of
submerged pontoons are denoted by P1 – P4. the tank (at 4.75 λ).
Table 1. Main dimensions of the TLP structure.

Parameter Value Reference


Hull column Bachynski and
diameter 18.0 m Moan [5]
Hull column Bachynski and
height 52.6 m Moan [5]
Draft 47.9 m Matha [3]
Freeboard 5.0 m Matha [3]
Tower height 81.0 m Matha [3]
Bachynski and
Pontoon size 18.0x2.4x2.4 m Moan [5]

Bachynski and
Water depth 150.00 m Moan [5]

Figure 1 Fixed TLP structure (left) and top view of the four
pontoons (right). Figure 2 Methodology flowchart.
Figure 2 shows a flowchart of the methods applied
in this study to predict the horizontal wave excitation A free surface zone was introduced in the domain
forces. The first step was to validate CFD wave with a height of 6 m (3 m in the air zone and the
elevations using the Stokes 5th order wave theory. other 3 m in the water zone) to ensure that the
The fixed TLP structure was then modelled in both interface between the two immiscible fluids could be
CFD and AQWA codes. The results obtained from accurately captured (Figure 3). To obtain an
both packages were then compared to ascertain accurate prediction of the wave propagation in the
how much difference in the predicted force free surface part, the optimum number of cells
magnitudes for each wave condition. The procedure required per wave height and wave length was
was repeated such that the size of the mesh/panel found to be approximately 20-30 cells and 80 cells,
was fine-tuned to yield optimum results for both respectively [8-10]. Furthermore, a value of 0.01s
solvers. was used for the time step (dt), as it was found to
be an optimum value to capture the dynamics of a
2.1 CFD Method sharp wave free surface and maintain an optimal
To carry out a CFD study, two separate domains solution using the High-resolution Interface
were created namely a Numerical Wave Tank capturing (HRIC) scheme [10, 12, 14]. The selected
(NWT) and a wave-structure interaction domain time step of 0.01 s resulted in lower CFL numbers
[10]. Figure 3 shows the setup of the NWT created (less than 0.5). The non-dimensional CFL number
in STAR-CCM+. Five wave periods at a constant (also known as the Courant number) is a function of
the maximum wave velocity, time step, and mesh
Australasian Coasts & Ports 2019 Conference – Hobart, 10-13 September 2019
Numerical Prediction of Wave Excitation Forces on a Fixed Tension Leg Platform Concept for Offshore Wind Turbines
S. Khanam, N. Abdussamie, and R. Ojeda

cell size as given in Equation (1). It should be noted 2 1.000 0.125 8.000 279,924
that the maximum velocity, Umax, was obtained 3 0.5 0.125 8.000 546,308
using the Stokes second order formula [6].

𝑈𝑚𝑎𝑥 𝑑𝑡
𝐶𝐹𝐿 =
min⁡(𝑑𝑥, 𝑑𝑧) (1)

To enable the simulation of nonlinear wave


conditions, the NWT used the fifth order of the
Stokes wave theory. The fifth order wave was
deemed applicable in this condition as the Ursell
number, which is a function of wave height, wave
length, and water depth, was found to be less than
30 [15]. In wave generation, it is necessary to
implement a damping zone to minimise the effect of
wave reflection which can affect the quality of the
waves generated throughout the NWT. The NWT
had a damping zone of 2 λ at the far end of the tank
to dampen the motions of the free surface [11].

Table 2 Test conditions at H = 3 m.


Figure 4 Mesh study results of wave elevation for
Condition T (s) λ (m) Ursell No. Condition 1 (H = 3 m and T = 12 s).
1 12.0 224.7 0.05 Figure 5 shows the results of wave height for all
2 11.0 188.9 0.03 conditions as a function of wave steepness (S).
3 10.0 156.1 0.02 Conditions 1 and 3 (S = 0.013 and 0.019) were
4 9.0 126.5 0.01 found to be in close agreement with the theoretical
5 8.0 99.9 0.01 wave height. The small variation between CFD and
theory can be attributed to the viscous effects which
are neglected in potential flow theory.

Figure 3 Profile view of wave generation domain (NWT).

The mesh sizes were obtained through a mesh


convergence study that was carried out for each test
condition where the mesh size in the x- and z-
direction was varied (Table 3). A base cell of 6.0 m
was applied throughout the domain except in the
free surface zone where refined mesh was selected.
By referring to Figure 4, it was found that Mesh 1
resulted in a relative error of 4.4 % between CFD
results and the input wave height whereas Mesh 2
and Mesh 3 had an error of 5.5% and 6.5%,
Figure 5 Wave height comparison between CFD and
respectively.
Stokes wave theory as a function of wave steepness.
Table 3 Mesh study parameters for condition 1 (H = 3 m
and T = 12 s). The TLP structure was placed in the middle of the
tank to minimise the effect of reflected waves from
Mesh dx (m) dz Base Total no. of the far boundaries. To capture the fine flow details
(m) size cells around the TLP structure, a finer mesh compared to
(m) the NWT mesh size was applied such that a surface
1 0.750 0.094 6.000 472,280
cell size of 0.3125 m was used (
Australasian Coasts & Ports 2019 Conference – Hobart, 10-13 September 2019
Numerical Prediction of Wave Excitation Forces on a Fixed Tension Leg Platform Concept for Offshore Wind Turbines
S. Khanam, N. Abdussamie, and R. Ojeda

Figure 6). To account for the turbulence flow, the k-


ω model was selected, and similarly, a 0.01 s time
step was adopted.

TLP column

Figure 6 Profile view of mesh refinement around the TLP structure and in the free surface zone.
2.2 Panel Method at the different mesh cases 1-5. Mesh case 1 which
ANSYS AQWA, as an industry accepted software, had a coarse mesh (lower number of elements)
was chosen as a tool to verify the CFD results. A resulted in the largest force magnitudes for the
wave frequency range from 0.04 Hz to 0.2 Hz (wave whole wave period range tested. Mesh cases 2-5
period 5-25 s) was selected to solve the linear has quite similar results, and hence mesh case 5
diffraction problem in the frequency domain. To was selected because it had the lowest number of
save computational time and attain an optimal mesh elements in comparison with the other three cases.
size to capture wave excitation forces acting on the
TLP structure, a mesh study was also carried out
using five different cases as presented in Table 4.
Mesh condition 1 represents the default mesh size
whereas mesh conditions 2-5 were tested to
ascertain the differences in the force magnitudes.
Defeaturing tolerance is a function in AQWA that
automatically defeatures small features within the
geometry. To further investigate the effect of the
defeaturing tolerance and the maximum element
size, the defeaturing tolerance was kept constant at
0.4 for mesh cases 2-5, and only the maximum
element size varied. The total number of elements
was found to be affected only by the change of the
maximum element size.
Table 4 Mesh study for AQWA simulations.
Defeaturing Maximum
No. of
Case tolerance element
elements
(m) size (m)
Mesh 1 2.0 5.0 713
Mesh 2 0.4 0.7 30,414
Mesh 3 0.4 0.8 23,821 Figure 7 Horizontal forces at different mesh sizes given at
Mesh 4 0.4 0.9 18,926 H = 3.0 m.
Mesh 5 0.4 1.0 15,534
3. Results and Discussion
The results obtained from the CFD and AQWA
Figure 7 shows a comparison of the horizontal wave
codes are discussed and compared below for the
excitation force (denoted by Fx) obtained by AQWA
Australasian Coasts & Ports 2019 Conference – Hobart, 10-13 September 2019
Numerical Prediction of Wave Excitation Forces on a Fixed Tension Leg Platform Concept for Offshore Wind Turbines
S. Khanam, N. Abdussamie, and R. Ojeda

fixed TLP offshore wind turbine structure. Figure 8


shows a comparison between the CFD wave
elevations and the Stokes wave theory, and the
resulting horizontal forces for conditions 2 and 3.
The wave results presented here were obtained
while the TLP in the domain (wave-structure
domain). As can be seen, there was generally a
good wave elevation agreement between CFD and
theory, particularly for condition 3.

Figure 9 Wave forces acting on the pontoons for condition


1 (H = 3 m and T = 12 s).
By comparing CFD and AQWA results, the
magnitudes of the horizontal force are in good
agreement as given in Table 5. It was found that the
predicted horizontal force acting on the fixed TLP
structure by the CFD code had a mean relative
difference ranging between 5% and 10%. Overall,
the force magnitudes predicted by AQWA were
higher than those predicted by the CFD code. Such
discrepancies could be attributed to the viscous
effects being captured in CFD.

Table 5 Horizontal wave forces predicted by AQWA and


CFD codes.
Figure 8 Time history of wave elevation (top) and
horizontal force (bottom) for condition 2 (left) and Condition T(s) AQWA (MN) CFD (MN)
condition 3 (right).
1 12 5.54 5.25
Figure 9 shows only the CFD forces experienced by 2 11 6.19 5.66
the platform pontoons for condition 1. As expected,
all the horizontal forces acting on the submerged 3 10 6.73 6.09
pontoons were out of phase from each other. 4 9 7.13 6.59
Overall, it was found that the force magnitudes 5 8 7.35 6.76
acting on the pontoons contributed about 3% to 4%
to the total force of the wave excitation force. It could
also be seen that the port pontoon (refer to Figure 4. Conclusions
1) yielded the highest force followed by the forward The horizontal wave excitation force acting on a
(FWD) and aft (AFT) pontoons. fixed TLP offshore wind turbine due to several
deterministic regular wave conditions was predicted
using a CFD based code. A code-to-code
comparison with the commercial software, ANSYS
AQWA, was then carried out. It was found that the
CFD code under predicted the horizontal wave
forces acting on the structure compared to AQWA
results. The mesh sizes had a significant impact on
the horizontal forces obtained in the CFD code.
Furthermore, it was found that the submerged
pontoons contributed approximately 4% of the total
forces predicted by CFD. The linear diffraction
solver provided an insight into the horizontal force
experienced by the TLP structure, and hence it
could be used to verify CFD results. However, as
the linear diffraction theory does not account for the
effect of turbulence, viscous effects or the effect of
the air phase, CFD codes could be used to
investigate such phenomena and provide further
Australasian Coasts & Ports 2019 Conference – Hobart, 10-13 September 2019
Numerical Prediction of Wave Excitation Forces on a Fixed Tension Leg Platform Concept for Offshore Wind Turbines
S. Khanam, N. Abdussamie, and R. Ojeda

details. To gain a better understanding of how a TLP


support structure would react to wave loading
further study could be done on the structure in a
floating condition.

5. References
[1]. IRENA (2018), Renewable Capacity Statistics
2018 by The International Renewable Energy
Agency (IRENA).
[2]. GWEC (2017), GWEC Global Wind 2017 Report
by Global Wind Energy Council (GWEC).
[3]. Matha, D. (2009), Model development and loads
analysis of an offshore wind turbine on a tension
leg platform with a comparison to other floating
turbine concepts: National Renewable Energy
Lab.(NREL), Golden, CO (United States).
[4]. DNV, DNV-OS-J101 (2014), Offshore Standard:
Design of Offshore Wind Turbine Structures.
DNV AS, Høvik, Norway.
[5]. Bachynski, E.E. and T. Moan (2012), Design
considerations for tension leg platform wind [11]. Choi, J. and S.B. Yoon (2009), Numerical
turbines. Marine Structures, 29(1): p. 89-114. simulations using momentum source wave-
[6]. DNV (2010), Recommended Practice DNV-RP- maker applied to RANS equation model. Coastal
C205: Environmental conditions and Engineering, 56(10): p. 1043-1060.
environmental loads. Norway. [12]. Jasak, H., V. Vukčević, and I. Gatin (2015),
[7]. Chakrabarti, S.K. (1987), Hydrodynamics of Numerical simulation of wave loading on static
offshore structures. WIT press. offshore structures, in CFD for wind and tidal
[8]. Abdussamie, N., W. Amin, R. Ojeda, G. Thomas, offshore turbines. Springer. p. 95-105.
L. Tasmania, and Y. Drobyshevski (2014), [13]. Olsson, A. and M. Tunlid (2015), CFD simulation
Vertical wave-in-deck loading and pressure of wave-in-deck loads on offshore structures.
distribution on fixed horizontal decks of offshore Master Thesis in Naval Architecture and Ocean
platforms. in The Twenty-fourth International Engineering, Chalmers University, Sweeden.
Ocean and Polar Engineering Conference. [14]. Rhee, S.H., B.P. Makarov, H. Krishinan, and V.
International Society of Offshore and Polar Ivanov (2005), Assessment of the volume of fluid
Engineers. method for free-surface wave flow. Journal of
[9]. Abdussamie, N., R. Ojeda, W. Amin, G. Thomas, marine science and technology, 10(4): p. 173-
and Y. Drobyshevski (2014), Prediction of wave- 180.
in-deck loads on offshore structures using CFD. [15]. CD-Adapco (2012), User guide - Star-CCM+
in the 19th Australasian Fluid Mechanics Version 7.04., CD-Adapco.
Conference. Melbourne, Australia: AFMS.
Citeseer.
[10]. Abdussamie, N., R. Ojeda, G. Thomas, and W.
Amin (2016), Offshore Platforms in Waves-CFD
Simulations for Devastating Conditions. in The
Twelfth ISOPE Pacific/Asia Offshore Mechanics
Symposium. International Society of Offshore
and Polar Engineers.

You might also like