Download as pdf or txt
Download as pdf or txt
You are on page 1of 19

Ocean Engineering 262 (2022) 112221

Contents lists available at ScienceDirect

Ocean Engineering
journal homepage: www.elsevier.com/locate/oceaneng

A novel framework for modeling floating offshore wind turbines based on


the vector form intrinsic finite element (VFIFE) method
Yu Zhang a, Wei Shi b, c, *, Dongsheng Li d, **, Xin Li e, Yuanfeng Duan f, Amrit Shankar Verma g
a
Department of Civil Engineering, Dalian University of Technology, Dalian, 116024, China
b
State Key Laboratory of Coastal and Offshore Engineering, Dalian University of Technology, Dalian, 116024, China
c
Deepwater Engineering Research Center, Offshore Renewable Energy Research Center, Dalian University of Technology, Dalian, 116024, China
d
Guangdong Engineering Center for Structure Safety and Health Monitoring, MOE Key Laboratory of Intelligent Manufacturing Technology, Shantou University, Shantou,
Guangdong, 515063, China
e
Institute of Earthquake Engineering, Faculty of Infrastructure Engineering, Dalian University of Technology, Dalian, 116024, China
f
College of Civil Engineering and Architecture, Zhejiang University, Hangzhou, 31000, China
g
Department of Mechanical Engineering, University of Maine, USA

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

Keywords: Currently, there is a rapid demand for the development of floating offshore wind turbines (FOWTs) for
Spar-FOWT deployment in sites with deep water depths. FOWTs are highly complex structures that are subjected to com­
Novel modeling framework bined loading from wind, hydrodynamic, hydrostatic, and mooring loads, and these loads have a significant
VFIFE method
influence on their dynamic behavior. The complexity of such a multi-body system makes the dynamic analysis
Multi-body system dynamics
Dynamic of mooring system
considerably challenging and demands an efficient model to capture the physical characteristics of the system
Gyroscopic moments accurately. This paper proposes a novel framework for modeling floating offshore wind turbines (FOWTs) based
on the vector form intrinsic finite element (VFIFE) method. In this framework, the multi-body dynamics (MBD) is
used to handle the rigid body motion, and the analysis of structural deformation and the solution of governing
equation of motions (EOMs) is implemented based on the VFIFE method. The methodology considers the FOWT
as two rigid bodies: (a) the tower, which is a structural assembly of platform and nacelle; and (b) the rotor that
can mechanically rotate relative to the nacelle. Further, a dynamic and flexible model of the mooring system is
included, and their axial extension, inertia, and hydrodynamic loads are considered. The EOMs of the FOWT and
mooring system are derived by the Newton-Euler (NE) and Newton’s second law, respectively and the central
difference scheme is implemented to solve the EOMs. Finally, the developed model is subjected to different load
cases under the combined action of wind and waves, and the responses of platform motion and the axial tension
of the mooring system are calculated and verified against the FAST. After thorough verification of the VFIFE
model which showed excellent agreement with FAST results, the proposed model is used (a) to highlight the
variations caused by neglecting the dynamic of the mooring system where the results are compared with the
quasi-static mooring model developed using the MAP++ module in FAST, and (b) reveal the coupled mechanism
of the platform pitch and yaw motion when the rotor is spinning.

1. Introduction significantly slow down global warming and reduce CO2 emissions
(Barthelmie and Pryor, 2021; Zeng et al., 2022). A wind turbine is a
In order to achieve the goals of carbon neutrality, the power pro­ promising technology to exploit wind energy, and specifically, offshore
duced from renewable energy is being paid more and more attention by wind turbines are installed in the ocean to harness the kinetic energy
governments around the world. Wind energy is one of the most prom­ from the moving wind. Currently, the majority of offshore wind turbines
ising forms of clean energy and it is expected that the development and are based on bottom-fixed foundations deployed in shallow water depths
utilization of this energy both at onshore and offshore locations will (Zhang et al., 2020). However, not all locations around the world have

* Corresponding authorState Key Laboratory of Coastal and Offshore Engineering, Dalian University of Technology, Dalian, 116024, China
** Corresponding author
E-mail addresses: weishi@dlut.edu.cn (W. Shi), lids@stu.edu.cn (D. Li).

https://doi.org/10.1016/j.oceaneng.2022.112221
Received 11 May 2022; Received in revised form 24 July 2022; Accepted 5 August 2022
Available online 24 August 2022
0029-8018/© 2022 Elsevier Ltd. All rights reserved.
Y. Zhang et al. Ocean Engineering 262 (2022) 112221

shallow water depths and the seabed can have deep bathymetry. Hence, procedure. The EL method establishes the EOMs of the multi-rigid-body
the floating foundations will become an inevitable choice for deploy­ system based on the energy theory. The EL method can overcome the
ment of offshore wind technology in sites with deep water depths (Wang shortcomings of the NE method because it is not required to calculate the
et al., 2022). There are other commercial advantages of floating offshore internal forces. Nevertheless, the EL method involves a series of partial
wind turbines (FOWTs) such as the wind speed is more stable and derivatives of the energy with respect to related degrees of freedom
stronger, and the wind shear is smaller. Moreover, placement in deep (DOF). The derivation of these partial derivatives is also laborious. Kane’s
water avoids the problems associated with noise and visual impacts (Ren method combines the advantages of the NE method and EL method, which
et al., 2022). In addition, the FOWTs can be assembled on the quayside avoids the calculation of the partial derivatives and the internal forces. In
and can be towed to the target location, thus, it will significantly reduce FAST, Kane’s method is used to establish the dynamic equation of the
the installation and construction costs associated with expensive crane FOWT (Sarkar and Fitzgerald, 2021).
operations. Among the many multi-body dynamics models of FOWTs, except for
FOWTs are subjected to combined loads, including wind and hy­ a few well-known codes (such as FAST and HAWC2), currently, the
drodynamic, hydrostatic, and mooring loads, and these loads have a dynamic modeling of the mooring system is not considered in many in-
significant influence on the dynamic behavior. This demands a precise house programs. Although, the linear spring and catenary theory are
multi-body dynamic model to predict the system’s dynamic perfor­ usually used to calculate the mooring tension, they are static or quasi-
mance. The complexity of such multi-body system makes the dynamic static methods (Shen et al., 2016). Leble and Barakos (2016) used a
analysis greatly challenging. Few theoretical multi-rigid-body dynamic high-fidelity method, which is precise in aerodynamic loads and hy­
models of FOWTs have been presented in literature. Liu et al. (2019) drodynamic loads calculation. The mooring cables are modeled as a set
used a fully coupled computational fluid dynamics (CFD) and of springs and dampers. The mooring system is an integral part of the
multi-body dynamics (MBD) method to calculate the aerodynamic loads, FOWTs, and its primary function is to provide station keeping and
in their work the platform surge motion on results was considered. Using maintain the safety and stability of the FOWTs. The analysis methods of
the same approach, Li et al. (2017) presented a high-fidelity simulation mooring systems can be classified as quasi-static and dynamic. The
framework to investigate the drivetrain dynamics. Interactions between quasi-static method has the advantage of lower computational costs.
drivetrain dynamics and servo-control dynamics were studied. Tan et al. However, the inertia effects and internal damping of the line together
(2019) proposed a novel multi-body dynamic drivetrain model, which with external forces such as the added mass and drag force of water are
considers the dynamic response caused by platform motion. Ma et al. neglected, so it cannot truly reflect the platform-line coupling motion
(2019) proposed a time-domain simulation program (named DUTMST) (Hall and Goupee, 2015). Therefore, to overcome the defects of the
based on MATLAB to research the dynamic response of a tension leg quasi-static method, it is necessary to establish the dynamic model of the
platform (TLP) wind turbine. Then this model was verified by experi­ mooring system in the multi-body model of FOWTs.
mental data. Park et al. (2016) proposed a relatively complex dynamic The vector form intrinsic finite element (VFIFE) method is an inno­
model, where the flexible multi-body dynamics were employed to vative approach for structural motion behavior analysis (Ting et al.,
construct the equations. This system consisted of the platform, tower, 2004a; 2004b; Shin et al., 2004). The VFIFE method combines vector
nacelle, hub and three blades. The tower and three blades were con­ mechanics and numerical computation theory. In this approach, based
structed as flexible bodies using the three-dimensional beam element, on the physical model of the structure, the structure is described as a
and the rotor rotation is also considered. Ding et al. (2017) established a collection of space particles connected by the massless elements but can
numerical model of submerged TLP wind turbine based on multi-body bear the internal forces. A series of path units are used to describe the
dynamics, and the stability and towline force in the transport phase process of deformation, whereas the pure deformation and internal force
was studied. Then, the environmental load effects on the towing stability are calculated by fictitious reverse motion. The governing equations of
of the FOWTs were investigated. space particles are derived from Newton’s second law. The solution of
FOWTs are a typical multi-body system and can be regarded as a the governing equations is performed by the central difference method,
combination of multi-rigid-body. However, they can also be considered as an explicit integral solution method that does not require an iterative
a rigid-flexible coupled structure to obtain more accurate results (Kar­ solution to the governing equation. This method is suitable for large
imirad and Moan, 2012). The most common modeling strategy is to deformations and geometrical changes of the structure, especially
consider the FOWT separated into four bodies: floating platform, tower, elastic-plastic problems and collision and collapse problems, and is
nacelle, and rotor (Al-Solihat et al., 2019). If the elastic deformation of the capable of nonlinear and discontinuous mechanical behavior analysis of
structure is smaller compared to the displacement, then the elastic the structure. The VFIFE method is a complete structural dynamic and
deformation can be neglected, and the FOWTs can be modeled as a static analysis method and has been applied to many engineering fields.
multi-rigid-body system (Wang and Sweetman, 2013; Sweetman and Civil engineering is the most widely used field, such as the vibration
Wang, 2014). But, if the elastic deformation cannot be neglected in certain analysis of train–track–bridge coupled system (Wang et al., 2021) and
conditions, then the FOWTs should be treated as a rigid-flexible coupled the earthquake-induced collapse of bridges (Duan et al., 2014). Espe­
multi-body system. Such as in Al-Solihat and Nahon’s research (2018), the cially, Tan et al. (2021) released an object-oriented structure analysis
tower is regarded as a deformable flexible body, and the remaining parts platform named openVFIFE. This is a unified general analysis frame­
of the floating wind turbines are considered as rigid bodies. In Chen’s work, which will significantly promote the applications and de­
study (2017), the tower and blade are regarded as deformable flexible velopments of VFIFE. In the solid mechanics fields, the simulation of
bodies, and the platform and nacelle are considered to be rigid bodies. crack propagation (Duan et al., 2017) and the elastoplastic analysis of
Based on the momentum theory or energy theory, the methods to fine beams (Yuan et al., 2018) can also be analyzed using the VFIFE
establish the equation of motion (EOM) of multi-body systems include method. In the ocean engineering fields, the VFIFE method has been
Newton-Euler (NE) method, Euler-Lagrange (EL) method and Kane’s widely used in recent years. Li et al. (2018a, 2018b, 2020, 2021) used
method (Sweetman and Wang, 2014). The EOMs obtained by these clas­ the VFIFE method to analyses the vortex-induced vibration of
sical analytical methods are essentially equivalent, but the application three-dimensional marine risers. Gu et al. (2022) studied the collision
scope may differ from each other in terms of specific problems. In the NE probability problem of vertical risers considering the wake interference
method, the EOMs are established for each rigid body, and the internal effect. Xu et al. (2021, 2022) used the VFIFE method to simulate the
forcing at each joint must be considered. Thus, it may be complex to derive deep-water laying pipeline. In addition, the VFIFE method was used to
EOMs using the NE method when the number of rigid bodies is large. An establish an integrated model of the wind turbine system, and the dy­
advantage of the NE method is that the change of a newly added body on namic response under the wind loads was analyzed (Chen et al., 2017). It
the EOM can be represented conveniently by a recursive formulation can be found that the VFIFE method has a wide range of applications,

2
Y. Zhang et al. Ocean Engineering 262 (2022) 112221

the fracture process of the mooring line and the nonlinear constitutive
relation of the mooring line material. In sections 2 and 3, the mooring
system is regarded as flexible and deformable, while the FOWT is
regarded as an assembly of multi-rigid-bodies. Based on the dynamic
model of the mooring line developed in our previous work (Zhang et al.,
2022), the multi-body dynamics model of FOWT was further developed
in this paper based on the Newton-Euler (NE) method. Meanwhile, the
two models were coupled to form a new framework for modeling
FOWTs. The mooring system module and the main structure module are
bilaterally weak coupled. Those are described in section 4. Thus, the
dynamic effects of the mooring system in the mooring tension calcula­
tion and the coupling motion between the platform and mooring system
can be considered. In the VFIFE model, the aerodynamic loads are
simplified as an approximate concentrated thrust force. The hydrody­
namic loads of the spar platform and mooring system are calculated with
the Morison’s equation. Three load cases are adopted to verify the sys­
tem dynamic response of FOWTs obtained by VFIFE method and the
results are verified with analysis performed in FAST. Finally, the
coupling of the gyroscopic moments arising from rotor spin and platform
rotation is investigated and described in section 5.

2. Multi-body modeling of FOWT

2.1. Coordinate frames and cardan angles

The whole structure of FOWT is regarded as a simple two rigid body


dynamic system: the support structure and the rotor. The support
structure is a structural assembly, including the nacelle, the tower, and
Fig. 1. Coordinate frames used in the VFIFE model. the platform; and the rotor is an assemble of hub and three blades, which
can mechanically rotate relative to the nacelle. The implementation of
and the reason for this widespread popularity comes from its theoretical the MBD method requires use of several coordinate frames to derive the
innovation. EOMs for the FOWT system. The coordinate frames defined for the
This paper proposes a novel framework for modeling FOWTs using multi-body system of the FOWT are shown in Fig. 1. The systems
the VFIFE method. In this framework, the MBD method is used to handle (X, Y, Z) and (xg , yg , zg ) are the earth-fixed global coordinate frames. The
the rigid body motion, the analysis of structural deformation and the origin O of coordinate frame (X, Y, Z) is located at the seabed; the origin
solution of governing EOMs are conducted based on the VFIFE method. Og of coordinate frame (xg , yg , zg ) is located at the intersection of the
Hence, this framework is capable for handling the rigid-flexible coupling central axis of the tower and the still water level surface when the
problem. Considering that the VFIFE method does not require the as­ platform is in its undisplaced position, and this frame named as e(g) . The
sembly of the global stiffness matrix, and the equations of motion are systems (xt , yt , zt ) and (xr , yr , zr ) are the body-fixed coordinate frames.
established based on every node. Thus, the singularity problem can be The name of the frame (xt , yt , zt ) is e(t) and the origin Ot initially co­
well avoided in this framework. Another distinguishing feature of the incides with the global coordinate frame (xg , yg , zg ) and then moves with
VFIFE method is code modularity and sequencing. The program of this
the FOWT. The frame e(t) defined like this is convenient for comparing
model changes little if the boundary conditions change or some struc­
the simulation results with FAST. The name of the frame (xr , yr , zr ) is e(r)
ture is added to this model. Therefore, it is very easy to extend the model
and the origin Or located at the center of the hub, the coordinate frame
with different configurations. In addition, this framework can simulate
e(r) fixed on the rotor and is parallel to the frame e(t) initially, the rotor

Fig. 2. Cardan angles α, β, and γ corresponding to three rotational motions.

3
Y. Zhang et al. Ocean Engineering 262 (2022) 112221

rotates along the xr axis. g


The key point of rigid body space motion is the description of relative t
ṙog →or = T g→b ⋅ ṙog →ot − t ̃
ρot →or ⋅t ωt (9)
orientation between two coordinate frames. Here, the Cardan angle
approach (Tupling and Pierrynowski, 1987) is applied to describe the t ¨
rog →Or
g
= T g→b ⋅ r¨og →Ot − t ̃
t (
ρot →or ⋅ ω̇t + t ωt × t ωt × t ρot →or
)
(10)
rotational angles of a rigid body. In order to implement the large-amplitude
rotational motion in space, the rotational order must be specified clearly.
The three rotational motions used here are defined in Fig. 2, wherein α, β, 2.2.2. Rotational motion
and γ are the Cardan angles, which are written as follows: Angular velocity is the most important and easily misunderstood
concept in rigid body kinematics. If the projection of the angular ve­
θ=[α β γ ]T (1) locity ωi written as π̇i , then the πi has no actual physical meaning (it is
not Cardan angles here) and is usually referred to as pseudo-coordinates.
Thus, the rotational transformation matrix (RTM) from the global
In general, the angular velocity vector of the rigid body rotates along a
coordinate frame to the body-fixed coordinate frame is
⎡ ⎤ fixed axis can be written as ω = θ̇e, the modulus of angular velocity is
cβcγ sαsβcγ + cαsγ sαsγ − cαsβcγ equal to the rate of rotation with the same direction as the rotation axis.
T g→b = ⎣ − cβsγ cαcγ − sαsβsγ cαsβsγ + sαcγ ⎦ (2) Thus, if using the Cardan angle approach to describe the rigid body
sβ − sαcβ cαcβ rotation, it can be seen in Fig. 2 that the three-rotation axis are

where c denotes a cosine function, and s denotes a sine function. p1 = g1


p2 = e12 = cosαg2 + sinαg3 (11)
2.2. Rigid body kinematics p3 = e23 = sinβg1 − sinαcosβg2 + cosαcosβg3

2.2.1. Translational motion The angular velocity vector of support structure assembly expressed
The position vector of the center of mass (COM) of the support by Cardan angles in global coordinate frame is
structure assembly with respect to the global coordinate frame e(g) can g Δθ
ωt = lim = α̇p1 + β̇p2 + γ̇p3 = [α̇ + γ̇sinβ]g1 + [β̇cosα − γ̇sinαcosβ]g2
be written as (Shabana, 2013) Δt→0 Δt

+ [β̇sinα + γ̇cosαcosβ]g3
g
rog →ct = g rog →ot + g rot →ct = g rot →ct + T b→g ⋅t ρot →ct (3)
(12)
where the symbol g ( ⋅) in above formula denotes the vector belonging to Then equation (12) can also be written as a matrix form
the frame e(g) ; while t ( ⋅) denotes the vector belonging to the frame e(t) .
The vector g rog →ct is the radius vector point from the Og to the Ot . The
g
ωt = Gθ̇ (13)
vectors t ρot →ct and g rot →ct are the radius vector point from the Ot to the
in which
COM of support structure assembly. The matrix T b→g represents the RTM ⎡ ⎤
from the body-fixed coordinate frame to the global coordinate frame. 1 0 sβ
Then, the velocity and acceleration of the COM of the support structure G = ⎣ 0 cα − sαcβ ⎦ (14)
0 sα cαcβ
assembly are
g It has been pointed that the EOMs of FOWT are established in the
g
ṙog →ct = ṙog →ot + g ωt × g rot →ct (4) body-fixed coordinate frame. Thus, the angular velocity vector
expressed in the body-fixed coordinate frame is
g (g )
g ¨
rog →ct = g r¨og →ct + ω̇t × g rot →ct + g ωt × ωt × g rot →ct (5) t
ωt = T g→b ⋅g ωt = T g→b ⋅Gθ̇ = Lθ̇ (15)
in which ωt and ω̇t are respectively the angular velocity and the
g g
in which
angular acceleration of support structure assembly expressed in frame ⎡ ⎤
e(g) . cβcγ sγ 0
In order to simplify the calculation of angular momentum of the two L = ⎣ − cβsγ cγ 0⎦ (16)
rigid bodies, the EOMs of the FOWT in this paper are established in the sβ 0 1
frame e(t) , thus, the velocity and acceleration should be expressed in the The angular acceleration vector expressed by Cardan angles in global
frame e(t) , they are coordinate frame is
g g
dω g
dp g
dp g
dp ( )
t
ṙog →ct = T g→b ⋅ ṙog →ot t ̃
ρot →ct ⋅t ωt (6) g
ωt = = α̈p1 + β̈p2 + γ̈p3 + α̇ 1 + β̇ 2 + γ̇ 3 = α̈p1 + β̈p2 + γ̈p3
dt dt dt dt
t ( ) + (α̇β̇p1 × p2 + γ̇ β̇p2 × p3 + α̇γ̇p1 × p3 )
t ¨
rog →ct = T g→b ⋅ g r¨og →ot − t ̃
ρot →ct ⋅ ω̇t + t ωt × t ωt × t ρot →ct (7)
(17)
The equation (17) describes the components of the angular acceler­
in which (⋅)
̃ denotes the skew symmetric matrix of a vector ( ⋅), such as
ation vector of arbitrary rotation. Then, the angular acceleration vector
an arbitrary vector r = ( r1 r2 r3 ), then ̃r is written as
⎡ ⎤ of support structure assembly expressed in body-fixed coordinate frame
0 − r3 r2 is (Shabana, 2013)

̃r = r3 0 − r1 ⎦ (8) g
− r2 r1 0 t
ω̇t = T g→b ⋅ ω̇ = Lθ̈ + α̇Zθ̇ + α̇β̇Z0 = Lθ̈ + Y (18)
Meanwhile, the relationship r × v = ̃r⋅v = − v ̃ ⋅r is also used in the
in which
derivation process of Eqs. (6) and (7), r and v are two arbitrary vectors.
⎡ ⎤
Using the same approach, the velocity and acceleration of the COM of 0 − sβcγ − cβsγ
the rotor Or can also be formulated with the same form Z = ⎣ 0 sβsγ − cβcγ ⎦ (19)
0 cβ 0

4
Y. Zhang et al. Ocean Engineering 262 (2022) 112221

Fig. 6. Path unit.


Fig. 3. Thrust force coefficient as function of relative wind velocity (Nielsen
et al., 2006). model only allows the relative rotation between the rotor and support
structure assembly, thus, the rotor and support structure assembly are
connected by a revolute joint. The angular velocity vector of support
structure assembly expressed in e(t) is t ωt , the angular velocity vector of
rotor with respect to support structure assembly is t ωrt . Thus, the ab­
solute angular velocity of rotor can be written as
t
ωr = t ωt + t ωrt (21)

Differentiating equation (21) with respect to time yields


t t
t
ω̇r = ω̇t + ω̇rt (22)

According to the relationship of the time derivative of vector with


respect to different coordinate systems, we can obtain the expression of t ω̇t

(23)
t t ′
ω̇rt = ωrt + t ωt × t ωrt

where the symbol ( ) denotes differentiation with respect to co­


ordinates. The term t ωt × t ωrt is the primary source of the gyroscopic


moments caused by rotor spin. With consideration of only exists the
relative rotational motion between the rotor and support structure as­
Fig. 4. Modeling of a hull with variable diameter.
sembly, thus, the expressions of t ωrt and ωrt are
r ′

t
ωrt = φ̇er1 (24)

(25)
t ′
ωrt = φ̈er1

where er1 is the direction vector of axis xr ; φ̇ and φ̈ are the angular ve­
locity and acceleration of rotor respect to the support structure assem­
bly. Hence, the equations (21) and (22) can be rewritten as
t
ωr = t ωt + t ωrt = t ωt + φ̇er1 (26)
Fig. 5. Discretization of the mooring line. t t t
(27)
t ′
ω̇r = ω̇t + ωrt + t ωt × t ωrt = ω̇t + φ̈er1 + t ωt × φ̇er1
⎡ ⎤

Z0 = ⎣ − sγ ⎦ (20) The term ωt × ωrt represents the Coriolis effects between the
t t

0 angular velocity of the platform and the relative angular velocity of the
rotor, which is also the main cause for the gyroscopic effects in a FOWT
Considering the rotor rotates along the xr axis and has no other system (Chen et al., 2017).
relative motion between the rotor and support structure assembly; thus,
the angular velocity and acceleration of the rotor are equal to the sup­
2.3. Equations of motion of FOWT
port structure assembly when the rotor is parked.

2.3.1. Derivation of EOMs


2.2.3. Relative motion of adjacent rigid body
Considering that only two rigid bodies exist in the model of FOWT,
Multi-body systems can be modeled as a set of bodies connected in a
the Newton-Euler method is convenient for deriving the EOMs. The
kinematic chain by sets of mechanical joints such as revolute, prismatic,
Newton-Euler equations with respect to the point O are used to derive
and/or cylindrical joints. In this paper, with consideration of the FOWT
the EOMs of the FOWT system. We apply the Newton-Euler equations to

5
Y. Zhang et al. Ocean Engineering 262 (2022) 112221

Fig. 7. Fictitious reverse motion; (a) element forward motion during a path unit (ta ≤ t ≤ tb ); (b) the fictitious reverse translation motion; (c) the fictitious reverse
rotation motion.

Fig. 8. Modules of the proposed framework.

the support structure assembly and rotor, respectively: constraining forces and moments from the support structure assembly
For the tower acting on the rotor, respectively. There are no specific requirements
about the specific expressions of these internal forces while deriving the
(28)
t
mt ⋅t r̈og →ct = Ftex + t Fr→t EOMs, because of the following relationships

d( ) ( ) ( )
t
t
Fr→t = − t Ft→r (32)
Jc ⋅ t ωt + t ωt × Jct ⋅ t ωt + mt t ρot →ct × T g→b ⋅ g ro¨g →ot = Mtex + t Mr→t
dt t
(29)
t
Mr→t = − t Mt→r (33)
For the rotor Substituting equations (7) and (10) into equations (28) and (30),
respectively, then combining the equations (28) and (30), and with the
(30)
t t
mr ⋅ ro¨g →Or = Frex t
+ Ft→r consideration of equation (32), we can obtain
( ) ( t ( )) t
d( ) ( )
mall T g→b ⋅ g r¨og →ot − t̃rcall ⋅ ω̇t + t ωt × t ωt × t rcall = Fall (34)
Jcr ⋅ t ωr + t ωr × Jcr ⋅ t ωr + mr t ρot →Or × T g→b ⋅ g ro¨g →ot = t Mrex + t Mt→r ex
dt
(31)
where mall is the mass of the whole FOWT; t rcall is the radius vector point
where Jct and Jcr are the rotational inertia of the support structure as­ from Ot to the COM of whole structure.
Similar, substituting equations (21) and (27) into equations (29) and
sembly and the rotor respect to their COM point, respectively. Ftex and
t
(31), respectively, then combining the equations (29) and (31), and with
t
Frex are the external forces acting on the support structure assembly and
the consideration of equation (33), we can yield
rotor, respectively. Mtex and t Mrex are the external moments acting on the
t
t ( all t )t
support structure assembly and rotor, respectively. t Fr→t and t Mr→t are
t ′
Jall t t
c ω̇t + ωt × Jc ⋅ ωt ωt + mall ̃rcall ⋅T g→b ⋅ g ro¨g →ot +Jrc ⋅ ωrt
the constraining forces and moments acting on the support structure ( ) t
+ t ωt × Jrc ⋅ t ωrt = Mall (35)
assembly from the rotor, respectively. In contrast, t Ft→r and t Mt→r are the ex

6
Y. Zhang et al. Ocean Engineering 262 (2022) 112221

Fig. 9. Schematic diagram of mooring system and spar platform of VFIFE model in MATLAB.

where Jall along the xr -axis in e(r) frame; then the component of wind speed along
c is the rotational inertia of the whole structure.
the xr axis is used to calculated the thrust, the thrust vector is written as
Substituting equations (15) and (18) into equations (34) and (35), [ ]T
respectively, we can obtain the EOMs of the FOWT system in a matrix F = Fxwr 0 0 . It is a simplified method of computing the thrust,
r w

form thus the components in other two directions are neglected. The thrust is
assumed to act at the point where the rotor is attached to the hub, and it
will produce a moment about the origin Ot in the body-fixed coordinate

[ ][ ⎡ ⎤ [ ] [ ( ) ] ⎡t ⎤
] g ¨
mall I3 − mall t̃rcall T g→b 0 ⎣ rog →ot ⎦ − mall t̃rcall Y mall t ωt × t ωt × t rcall Fall
+ all + t ( ) t ′ ( ) = ⎣ ex ⎦
(36)
mall t̃rcall Jall
c
0 L θ̈ Jc Y ωt × Jall t t t
c ⋅ ωt + J rc ⋅ ωrt + ωt × Jrc ⋅ ωrt
t
Mall
ex

From Chen’s study (Chen et al., 2017) we can know that the terms
Jrc ⋅ ωrt + t ωt × (Jrc ⋅t ωrt ) in the above equations are the additional gy­
t ′

roscopic moments produced by the rotor rotation.

2.3.2. Environmental forces


The external forces and corresponding moments acting on the spar
FOWT arise from wind and wave excitation forces, and gravity. The
wave excitation forces include hydrodynamic force and hydrostatic
forces.

2.3.2.1. Wind force. For simplicity, an approximate method proposed


by Nielsen (Nielsen et al., 2006) is used to calculate the wind thrust
forces. In this simplified method, the thrust is mainly defined by the
thrust force coefficient CT . The thrust coefficient only depends on the
relative wind velocity and is based on a look-up table, illustrated in
Fig. 3. The equation to calculate the wind thrust force is written as
1
F w = ρa AUr2 CT (37)
2

where ρa is the density of air; A is the total area swept by the rotor and Ur
is the relative velocity between the incoming wind and the wind turbine.
First, the wind velocity in the global coordinate frame e(g) is transformed Fig. 10. The sensitivity analysis for element number.
to the body-fixed frame e(r) by RTM T g→b , thus, the relative velocity is

7
Y. Zhang et al. Ocean Engineering 262 (2022) 112221

Fig. 11. The comparison of free decay motion between FAST and VFIFE model, (a) surge; (b) heave; (c) pitch; (d) yaw.

cylinders and one cone, however, the cone is treated as a cylinder


Table 1 (Fig. 4) to be able to use the Morison’s equation and strip theory
The definitions of three load cases.
approach to estimate the hydrodynamic force (Nielsen et al., 2006).
No. Wave height (m) Wave period (s) Wind speed (m/s) Wave type Hydrostatic forces (Fhs ) and added mass and radiation damping, are
LC 1 6 10 0 Regular all linearized. i.e. they are computed for the initial equilibrium position
LC 2 6 10 8 Regular of the hull. The added mass could be considered frequency independent
LC 3 6 10 0 Random
for the actual hull parameters, and radiation damping could be neglec­
ted (Nielsen et al., 2006). Therefore, the equations of hydrostatic forces
[ ]T and added mass as well as corresponding moments can be written as
system, with overturning moment being r Mw = 0 t
rOt →Or ⋅Fxwr 0 .
follows:
[ t hs ] [ t ] [ g→b ] [g ]
2.3.2.2. Wave excitation forces. The main objective of this research was F buoy
T 03×3 rog →ot
= F − g→b C
hs
(38)
to develop a structural model of FOWT, thus the hydrodynamic forces
t
M hs
03×1 03×3 T θ
(Fhd ) of the slender spar platform are approximately calculated by
Morison’s equation (Sarpkaya et al., 1982). The possibility of using the where Chs is the linear hydrostatic-restoring matrix from the effects of
Morison method to calculate the hydrodynamic of spar type structure water-plane area and the center of buoyancy; ρw is the water density; V0
has been verified (Karimirad et al., 2009). For the spar, the second-order is the displaced volume of fluid when the platform is in its undisplaced
forces are very small compared to the first-order forces. Therefore, only t
position; Fbuoy is the buoyancy which is defined as
the linear wave is considered in the present paper. The model will be
extended to include the higher-order wave forces in future work (Roald
et al., 2013). The spar platform shown in Fig. 1 is composed of two

8
Y. Zhang et al. Ocean Engineering 262 (2022) 112221

Fig. 12. The comparison of VFIFE model and FAST for LC1, (a) platform surge response; (b) platform pitch response; (c) mooring tension of line 1; (d) mooring
tension of line 2.

⎡ ⎤
0 where Ca is the coefficient of additional mass; Cd is the coefficient of
(39)
t g→b ⎣
F buoy
=T 0 ⎦ water
ρw gV0 viscous force; D is the diameter of spar; V̇ n is the acceleration of water
particles; Vnrel is the relative velocity.
In this research, the OC3 Hywind is used to verify the numerical Based on the strip theory approach, the total forces and moments in x
model, and the parameters of the model are obtained from the technical direction can be obtained by integrating over the submerged length ls
report of NREL (Jonkman, 2010). ∫
Hydrodynamic forces (Fhd ) can be expressed as a sum of the inertial Fxhd* = fxhd dz (42-1)
terms and combined drag and Froude-Krylov terms (Fhd* ). Hence, the ls

additional mass forces and corresponding moments can be written ∫ ( )


separately from the hydrodynamic forces Mxhd* = z ⋅ fyhd dz (42-2)
ls
⎡ ⎤
[ t hd ] [ g→b ] [ t hd* ]
Similarly, the expressions for the forces in the y direction can also be
g ¨
F rog →ot
=−
T 03×3
g→b ⋅μ(∞)⋅
⎣ ⎦ + tF (40)
t
M hd 03×3 T θ̈ Mhd* written. Hence, the wave forces and corresponding moments calculated
in the body-fixed coordinate frame e(t) are expressed
where μ(∞) is the added mass coefficient matrix from wave radiation ⎡ ⎤
loads at the infinite frequency. The drag and Froude-Krylov forces and Fxhd*
⎢ hd* ⎥
(43-1)
t hd*
corresponding moments can be written by Morison’s equation. Here we F =⎢ ⎣ Fy ⎦

just consider the forces acting normal to the axis of the leaning cylinder, 0
and the wave forces per unit length are

π D2 1 ⃒ ⃒
(41)
water
fnhd = ρw (1 + Ca ) V̇ n + Cd ρw DVnrel ⃒Vnrel ⃒
4 2

9
Y. Zhang et al. Ocean Engineering 262 (2022) 112221

Fig. 13. The comparison of VFIFE model and FAST for LC2, (a) platform surge response; (b) platform pitch response; (c) mooring tension of line 1; (d) mooring
tension of line 2.

⎡ ⎤
⎡ ⎤
Mxhd* t [t ] [t ] [t ] [t ] [t ] [t ]
⎢ hd* ⎥ Fall Fw Fhs Fhd FG FB FL
(43-2)
t ex
Mhd* = ⎢
⎣ My ⎦
⎥ ⎣ ⎦= t + t + t + t + t + t
t
Mall Mw M hs M hd MG MB ML
ex
0
(46)
Other External Forces
t t
The gravity force and moment around the origin of the body-fixed where FL and ML are the mooring tension and corresponding moment,
coordinate frame Ot are obtained and expressed in the frame e(t) are the calculate method will be introduced in detail in section 3. The
t t
⎡ ⎤ expression of FL and ML are
0
(44-1)
t G g→b ⎣
F =T 0 ⎦
(47-1)
t g
FL = T g→b ⋅ FL
− mall g

(47-2)
t t
ML = t̃rFair ⋅ FL
(44-2)
t t
MG = t̃rcall ⋅ FG
The mooring tension calculated in section 3 is defined in global co­
In addition, an additional linear damping matrix (Bdamp ) is recom­
ordinate frame e(g) , t̃rFair is the radius vector defined in the frame e(t)
mended in the technical report of NREL (Jonkman, 2010), it will gain a
point from Ot to fairlead position.
good agreement in free decay motion compared with the prototype test.
[t
FB
] [ g→b ] [g ] 2.3.3. Central difference solution of EOMs
T 03×3 ṙog →ot
t B
=− g→b B
damp
(45) The linear hydrostatic-restoring force in Eq. (38), the inertial terms
M 03×3 T θ̇
in Eq. (40) and the additional linear damping force in Eq. (45) are the
All external forces and corresponding moments acting on the spar functions of the platform kinematics and are often shifted to the left-
FOWT are hand side of the EOMs as in Eq. (36). Then, we rewrite the EOMs as a

10
Y. Zhang et al. Ocean Engineering 262 (2022) 112221

Fig. 14. The comparison of the VFIFE model and FAST for LC3, (a) platform surge response; (b) platform pitch response; (c) mooring tension of line 1; (d) mooring
tension of line 2.

simplified form [ ⎤[ ] [ ]
mall I3 − mall t̃rcall g→b
T g→b
g g
Ms = ⎦ T 0
+
03×3
μ(∞) (49-2)
Ms Ẍ + Cs Ẋ + K s g X = t Fall (48) t all
mall ̃rcall Jc 0 L 03×3 T g→b

in which [ ]
T g→b 03×3
Cs = Bdamp (49-3)
03×3 T g→b

[t ] [t ] [ t hd* ] [ t G ] [ t L ] [ ] [ ( ) ]
FW + Fbuoy + F F F − mall t̃rcall Y mall t ωt × t ωt × t rcall
t
Fall = t t + t + t − − ( ) t ′ ( ) (49-1)
Mw 03×1 Mhd* MG ML Jall
c Y
t
ωt × Jall t t t
c ⋅ ωt + Jrc ⋅ ωrt + ωt × Jrc ⋅ ωrt

11
Y. Zhang et al. Ocean Engineering 262 (2022) 112221

Fig. 15. Statistical comparison of the VFIFE model, FAST(MoorDyn) and FAST(MSQS) of LC1, (a) platform surge motion; (b) platform pitch motion; (c) mooring
tension of line 1; (d) mooring tension of line 2.

[ ]
T g→b 03×3 ( ) ( )
Ks = g→b C
hs
(49-4) hCs hCs g n− 1 ( 2 )
X + h Ks − 2Ms g X n = h2t Fall (51)
g
03×3 T Ms + X n+1 + Ms −
2 2
[g ] ( ) ( )
g
X=
rog →ot
(49-5) If we define three coefficients: P1 = 1/ Ms + hC2 s , P2 = P1 Ms − hC2 s ,
θ
P3 = P1 (h2 K s − 2Ms ), then the final iteration format represent by the
There are many numerical methods to solve the EOMs, in this central difference scheme can be obtained
research, the central difference scheme is used to solve Eq. (48). In the
(52)
g g
Xn+1 = P1 h2 t Fall − P2 X n− 1 − P3 g Xn
central difference scheme, the velocity and acceleration of the structure
g
can be calculated by the position vector X n+1 at step n+1, the position If the initial conditions are given, the position and forces of the
vector X at step n and the position vector X
g n
at step n-1. Therefore,
g n− 1 structure at any time can be calculated iteratively through the above
the expressions of velocity and acceleration are given by (Ting et al., steps.
2012)
3. Dynamic modeling of mooring line
g
1 ( g n+1 )
(50-1)
g
Ẍ = X − 2g X n + Xn− 1
h2 A lower-fidelity quasi-static approach that neglects inertia and hy­
drodynamics is often used for the mooring line model (Hall et al., 2014).
1 ( g n+1 )
(50-2) For small sea states, or minimal platform and mooring line velocities,
g g
Ẋ = X − X n− 1
2h
quasi-static models can provide a good approximation to the true
where h is the time increment. Substituting equations (50-1) and (50-2) mooring tension. However, for extreme sea states, the hydrodynamic
into equation (48), the governing equation becomes drag, added mass, and mooring line inertia are omitted in quasi-static

12
Y. Zhang et al. Ocean Engineering 262 (2022) 112221

Fig. 16. Statistical comparison of the VFIFE model, FAST(MoorDyn) and FAST(MSQS) of LC2, (a) platform surge motion; (b) platform pitch motion; (c) mooring
tension of line 1; (d) mooring tension of line 2.

models. This will result in a significant increase in error. Especially, the ⃒ ⃒ ⃒ ⃒


⃒g ⃒ ⃒ ⃒
(54)
g
quasi-static models cannot truly reflect the platform-line coupling mo­ X bj ⃒ − ⃒g X aj+1 − g
Δli = ⃒ X bj+1 − X aj ⃒
tion (Hall and Goupee, 2015). To overcome the defects of the
quasi-static method, a dynamic model of the mooring system based on where Mi is the mass matrix of element i, Mi =
π d2l l0i ρl
I3×3 ; I is the identity
4
the VFIFE method is established in our previous work (Zhang et al.,
matrix; ρl is the density of the mooring line; l0i is the element length and
2022), and the accuracy of this model has been validated with
dl is the diameter of the line before extension; g Xaj and g Xaj+1 are the
experiment.
position vectors of two nodes j and j+1 of element i at time ta , respec­
tively; X bj and X bj+1 are the position vectors of two nodes j and j+1 of
g g

3.1. Forces calculation of mooring line element i at time tb , respectively; the symbol ‘||’ refers to the norm of the
vector.
In the VFIFE method (Ting et al., 2012), the structure is discretised in Then the equivalent internal force of node j is calculated based on the
the space and time domains by point value description and the path unit pure deformation
(ta ≤ t ≤ tb ), respectively shown in Figs. 5 and 6. The fictitious reverse
(55)
g
motion calculates the pure deformation of element during a path unit. FTj = T bi − T bi− 1
The processes of these conducts are shown in Fig. 7. The mass of particle
⎧ ( )
j and the pure deformation of element i in a path unit are ⎨ T bi = σ ai Ai + EAi Δla i ebi
⎪ εi > 0
1
li (56)
Mj = [Mi− 1 + Mi ] (53) ⎪
⎩ b
2 T i = 0 εi ≤ 0

13
Y. Zhang et al. Ocean Engineering 262 (2022) 112221

Fig. 17. Statistical comparison of the VFIFE model, FAST(MoorDyn) and FAST(MSQS) of LC3, (a) platform surge motion; (b) platform pitch motion; (c) mooring
tension of line 1; (d) mooring tension of line 2.

where T bi is the axial tension at time tb ; εi is the strain of element i; σ ai is [ ]T


(59)
g
FRj = FjRx FjRy FjRz
the axial stress of element i at time ta ; lai is the element length at time ta ;
ebi is the direction vector of element i at time tb . The expression of FjRx , FjRy and FjRz are
The internal damping force is defined as the product of mass and ( )
velocity in the VFIFE method (Ting et al., 2012): FjRx = − μ⋅FjRz ⋅sign g ẋj (60-1)
g
( )
(57)
f
FDj = − ξ⋅Mj ⋅ Ẋ j FjRy = − μ⋅FjRz ⋅sign g ẏj (60-2)

where ξ is the internal structural damping coefficient. li + li− 1 [ ( ) ]


The equivalent node external force contributed from wet gravity are FjRz = Kb zbot − g
zj − Cb g żj (60-3)
2
given by
where Kb represents the stiffness per unit area of the seabed; Cb repre­
1( w ) π d2
(58) sents the viscous damping per unit area; zbot is the vertical coordinate of
g
FW
j = m + mwi ez , mwi = l l0i (ρw − ρl )g
2 i− 1 4
the seabed; g zj is the vertical coordinate of the node j of line; μ is the
where ρw is the density of water, and ez is the direction vector of the z- friction coefficient.
axis. The hydrodynamic forces on the mooring line are calculated with
The seabed forces of node j are Morison’s equation. The hydrodynamic drag force and the additional

14
Y. Zhang et al. Ocean Engineering 262 (2022) 112221

⎡π ⃒ ⃒√̅̅̅̅̅̅̅̅̅̅̅̅ ⎤
CDτ ρw dl li b V τi ⃒b V τi ⃒ 1 + εi
⎢2 ⎥
⎢ √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅√̅̅̅̅̅̅̅̅̅̅̅̅ ⎥
⎢1 ( )2 ( ) 2 ⎥
⎢ 1 + εi ⎥ (65)
b n b n
b H
Fi = ⎢ 2CDn ρw dl li V i V i + b V vi ⎥
⎢ ⎥
⎢ √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ⎥
⎣1 ( ) ( ) 2 √ ̅̅̅̅̅̅̅̅̅̅̅̅ ⎦
b b n 2 b
CDb ρw dl li V bi V i + V bi 1 + εi
2
( )
( a )
(66)
b g ˙
FM
i = M i + I 3×3 Q i ⋅ u i

⎡ / ⎤
Caτ ρw li πdl2 4 0 0
⎢ / ⎥
Mai = ⎢
⎣0 Can ρw li πdl2 4 0 ⎥
⎦ (67)
2
/
0 0 Cab ρw li π dl 4

where (θ, φ) are the Euler attitude angles, the rotation order is z-y-x, θ
and φ corresponding to the z and y axis, respectively; where CDτ , CDn and
CDb are the drag coefficients in the tangential, normal and binormal
Fig. 18. Comparison of yaw response for parked and spinning rotor. directions, respectively; Caτ , Can and Cab are the coefficients of added
b
mass; b V τi , b V ni and V bi represent the components of the relative velocity
mass force as well as added mass for node j can be written as
1 (b )
(61)
g H b H
Fj = QTi FH i− 1 + Fi
2

1 T (b M )
(62)
g b
FM
j = Qi Fi− 1 + FM
i
2

1( T )
Madd
j = Q ⋅ Mai ⋅ Qi + QTi− 1 ⋅ Mai ⋅ Qi− 1 (63)
2 i

where Q is the rotational transformation matrix (RTM) from global co­


ordinates (X, Y, Z) to element principal coordinates (xl , yl , zl ) and used
b b
only in the line model, as shown in Fig. 1; FH M
i and F i are the hydro­
dynamic drag force and the additional mass force in the element prin­
cipal coordinates, respectively; Mai is the added mass in the element
principal coordinates. The formulas to calculate these four variables can
be written as (Chen et al., 2019)
⎡ ⎤
cosφcosθ cosφsinθ sinφ

Q = − sinθ cosθ 0 ⎦ (64)
− sinφcosθ − sinφsinθ cosφ
Fig. 20. Pattern of platform yaw vary with rotor speeds.

Fig. 19. Comparison of platform motion responses with and without consider gyroscopic moments, (a) surge motion; (b) pitch motion.

15
Y. Zhang et al. Ocean Engineering 262 (2022) 112221

along the tangential, normal and binormal directions in the principal weak coupling.
g
coordinates, respectively; u ˙i is the acceleration of fluid particles in the
position of element i in the global coordinates system. The detailed 5. Results and discussion
modeling process can be found in Ref. (Zhang et al., 2022).
The new FOWT model proposed here based on the VFIFE method is
validated by comparing the results with the popular wind turbine dy­
3.2. Equations of motion of mooring line
namics software FAST (Jonkman, 2007). The OC3-Hywind (Jonkman,
2010) and NREL 5 MW wind turbine (Jonkman et al., 2009) are
The torsional and bending deformations of the mooring line and the
considered. The OC3 Hywind is a conceptual design of the Hywind
moment of inertia of the element are neglected in this model. Thus,
system developed to support the NREL 5 MW wind turbine. In FAST,
based on the Newton’s Second Law, and considering the Eq. (57), the
three candidate models can be selected to model mooring system, they
EOM of mooring line for node j can be written as
are MAP++, MoorDyn and FEAM, respectively. MAP++ represents the
( ) g
mooring model developed using the Multi-Segmented Quasi-Static
(68)
g ¨ g g g g g
Mj + Madd Xj + ξ ⋅ Mj ⋅ Ẋj = FTj + FW H M R
j + Fj + Fj + Fj
j
(MSQS) method; MoorDyn represents the mooring model developed
The central difference scheme is also used to solve the EOM, using the lumped-mass method FEAM module, the finite element
substituting Eq. (50) into above, and the external forces are simplified as modeling of the slender rod is used to establish the model of mooring
g g g g H g M g R system. The MoorDyn will be used to verify that the results calculated by
Fj = FTj + FWj + Fj + Fj + Fj . Then the EOM can be rewritten as
the VFIFE model are accurate. The MAP++ will be used as a comparison
( ) 1 (g g
) 1 (g n+1 g
) to highlight the dynamic effects of the mooring system on the calcula­
Mnj + Madd,n X n+1 − 2g Xnj + Xn− 1
+ ξ ⋅ Mnj Xj − Xn− 1
= g Fnj
j
h 2 j j
2h j
tion of mooring tension. The related dimensions and mass and inertial
(69) properties of the platform, tower, nacelle, and rotor are available in
NREL’s technical reports (Jonkman, 2010; Jonkman et al., 2009). The
Through simple mathematical operations, the final iteration format
moment of inertia of rotor about spin axis is 3.54 × 107 kg m2 (Wang
of EOM can be obtained as
and Sweetman, 2012), the moment of inertia of wind turbine are Ixx =
(70) 3.77 × 109 kg m2, Iyy = 3.66 × 109 kg m2, Izz = 1.12 × 108 kg m2
g 1g g
Xn+1
j = 2h2 (M1 )− Fnj + (M1 )− 1 M2 g Xnj − (M1 )− 1 M3 Xn−
j
1

(Karimirad and Michailides, 2015).


in which
[ ]
M1 = (2 + ξh) ⋅ Mnj + 2Madd,n
j (71-1) 5.1. Free decay motion

( ) The schematic diagram of the mooring system and spar platform of


M2 = 4 Mnj + Madd,n (71-2)
j
the VFIFE model in MATLAB is illustrated in Fig. 9. To calculate the
[ ] hydrodynamic loads acting on the mooring system and spar platform,
M3 = (2 − ξh) ⋅ Mnj + 2Madd,n
j (71-3) every mooring line is discretised into 30 segments while the platform is
discretised into 58 segments. The element length for cone and the upper
The n, n-1 and n+1 represent the variables belong to the time step n, cylinder of platform is considered as 1 m, whereas the element length for
n-1 and n+1. The time increment can be roughly estimated by h = lower cylinder of platform is considered as 3 m. The sensitivity analysis
2π li
√̅̅̅̅̅̅̅̅̅
3
ρl /E (Zhang et al., 2022). The function of this section is to calculate for the number of discrete elements was carried out. From Fig. 10, it
the mooring tension. shows that the number of discrete elements defined as above is sufficient
to ensure the accuracy of calculation.
4. Framework of coupled model Then, a free decay test is performed. Firstly, the VFIFE model moves
from the static equilibrium position to the specified initial perturbation
In this research, a new framework for modeling FOWTs is developed position, then the VFIFE model is released from the specified initial
based on the VFIFE method together with coupling the multi-body dy­ perturbation position in still water. Only the response of surge, heave,
namics theory. In this framework, the MBD method is used to handle the pitch, and yaw are presented here for brevity. The initial perturbation of
rigid body motion, and the calculation of structural deformation and the surge is 18 m, the initial perturbation of heave is 5 m, the initial
solution of governing EOMs are implemented based on the VFIFE perturbation of pitch is 10◦ , and the initial perturbation of yaw is 6◦ . The
method. Hence, this framework is capable of handling the rigid-flexible free decay test is a quite vital way to examine the function of hydro­
coupling problem. dynamics, hydrostatic, and mooring system, which are the primary
As illustrated in Fig. 8, the program of FOWT is divided into two sources of stiffness and damping in the system. The decay test results are
parts, the mooring system and the main structure (including rotor, na­ illustrated in Fig. 11. The comparison of surge, heave and yaw calculated
celle, tower and platform). The mooring system is regarded as flexible by VFIFE model exhibits excellent agreement with those obtained from
and deformable, while the main structure is regarded as an assembly of FAST. However, there are considerable variations in the pitch decay
multiple rigid bodies. The procedure of this framework includes a total results shown by the blue line compared to the FAST results shown by
of 6 modules and is realized in MATLAB. It should be pointed out that black line. This is because the hydrodynamic is calculated by a simple
the module of the mooring system and the module of the main structure approach of Morison’s equation. Therefore, the radiation damping is
are bilaterally weak coupled, that is: firstly, the fairlead position vector neglected in the VFIFE model, while this effect is considered in FAST.
calculated in the module of the main structure is transmitted to the The effect of radiation damping becomes evident at low platform ve­
module of the mooring system; then, the fairlead position vector is used locity, which is more visible for the pitch decay test (Al-Solihat et al.,
to calculate the mooring tension in the model of mooring line; then, the 2019). This difference will cause a slow decay of amplitude, the blue line
mooring tension is transmitted to the module of main structure; at last, in Fig. 11 (c). In order to obtain a same decay pattern like FAST, we
the module of main structure using the mooring tension to update the added a linear damping for roll and pitch motions in the additional
fairlead position vectors in next time step. Repeating the above steps, linear damping matrix (Bdamp ) in Eq. (45). The values of additional linear
cycling the calculation until the pre-set analysis time is completed. It can damping are B44 = B55 = 1 × 109 Nm/(rad/s) (Coulling et al., 2013).
be noted that there are no iterative corrections in the process of data The modified result exhibits an excellent agreement with that obtained
exchange between the two modules; thus, the modules have bilaterally from FAST, which is shown in Fig. 11 (c) using the red dash line.

16
Y. Zhang et al. Ocean Engineering 262 (2022) 112221

5.2. Response of platform motion and mooring tension can be obtained that the surge and pitch responses calculated by these
three models are basically agreement, only the mooring tension calcu­
The motion responses of FOWT under the excitations arising from a lated by the FAST(MSQS) is obvious different from the other two
linear wave and steady uniform wind are evaluated. Three load cases models.
(LCs) are defined in Table 1, in which the random wave is generated by The relative errors of above three cases of the three models can be
Jonswap spectrum (Al-Solihat and Nahon, 2018). The wave and wind seen by comparing their statistical performance in Figs. 15–17. These
are assumed collinear and moving in the global X-direction. Given that, results illustrate the maximum (max), minimum (min), mean, and
the surge and pitch motion of platform responses are expected to standard deviation (STD) values of the surge, pitch, mooring tension
dominate, therefore these responses are used to compare the proposed predicted by the VFIFE and FAST. Though the differences of mooring
VFIFE model and FAST. Meanwhile, the mooring tensions of line 1 and 2 tension between FAST(MoorDyn) and FAST(MSQS) are significant in
are also compared. Note that the transient effects are ignored in the Figs. 12–14, the maximum relative error does not exceed 5.0% for cases
simulation results, and only steady-state responses post-processed for 1 and 2; the relative errors between FAST(MoorDyn) and VFIFE model
comparison (Chen et al., 2019). are even smaller. For case 3, the maximum relative error of mooring
The steady-state responses for platform surge and pitch motion ob­ tension between FAST(MoorDyn) and FAST(MSQS) is 8.2%, while the
tained by the VFIFE model and the FAST for LC1 are plotted in Fig. 12 maximum relative error between FAST(MoorDyn) and VFIFE model is
(a)-(b). In the figures, ‘FAST(MoorDyn)’ represents the results calculated less than 1.5%. The significant difference of platform surge and pitch
by the MoorDyn model in FAST; ‘FAST(MSQS)’ represents the results between FAST(MoorDyn) and VFIFE model for case 3 are occurred in
calculated by the MSQS model in FAST; ‘VFIFE’ represents the results min-value, but the relative error still less than 5.0%. In general, the
calculated by the VFIFE model. The results clearly show that the plat­ VFIFE model proposed in this research displays a high performance and
form displacement in surge and pitch direction calculated by VFIFE accuracy in calculating platform motion response and the mooring
method agree well with both the models from FAST. Note that for LC1, tension, especially the dynamic characteristics of mooring system are
there are no significant differences in the platform displacement in surge reflected clearly, with accurate estimation of mooring line tension.
and pitch due to quasi-static and dynamic formulation of the mooring Overall, the VFIFE model can be used to conduct future scientific
line. However, a detailed investigations for different extreme load cases research, such as the coupled motion analysis of the floating platform
and the degree of influence of quasi-static and dynamic formulation of and mooring system.
the mooring line on the platform response will be dealt in future
research. 5.3. Rotor gyroscopic effect
Furthermore, the comparisons of mooring tension for line 1 and line
2 are shown in Fig. 12 (c) and (d), which illustrates that the maximum From Eq. (36) we know that the rotor spin will induce additional
and minimum values of mooring tension from VFIFE model are in gyroscopic moments to the system. This section considers a more spe­
excellent agreement with the FAST(MoorDyn) results. However, the cific expression of the additional gyroscopic moments. The rotational
mooring tension from FAST(MSQS) varies from the other two models. inertia of rotor Jrc and t ωt can be written as matrix form
Thus, it can be concluded that the dynamic characteristics of the ⎡ ⎤
mooring system can significantly affect the calculation of mooring ten­ Jrcxx Jrcxy Jrcxz
⎢ xy ⎥
sion. Note that an accurate estimate of the mooring tension is critical for Jrc = ⎢ yy yz ⎥
(72)
⎣ Jrc Jrc Jrc ⎦
accurate estimation of fatigue life of the mooring lines (Peng et al., xz yz
Jrc Jrc Jrc zz
2021). Therefore, it is essential to establish the dynamic model of the
mooring line of FOWT. ⎡ ⎤
Further, Fig. 13 presents the steady-state time response obtained by
t
ωx
⎢ t ty ⎥
the VFIFE model and the FAST for LC2. The comparison of LC2 is to t
ωt = ⎢
⎣ ωt ⎦
⎥ (73)
verify the VFIFE model has the capability to deal with complex load t
ωzt
conditions. Similar results for platform surge and pitch responses are
obtained where the results from VFIFE and FAST are found in excellent Further, combining the equations (24) and (25), the additional gy­
agreement. In addition, with the increase of platform motion amplitude, roscopic moments can be written as
the difference between dynamic model and quasi-static model also in­ ⎡ ⎤ ⎡ ⎤
creases. Meanwhile, the amplitude of platform motion for LC2 is larger Jrcxx − t ωzt Jrcxy + t ωyt Jrcxz
( ) ⎢ xy ⎥ ⎢ t z xx t x xz ⎥
than LC1, and the deviation in the maximum value of mooring tension (74)
t ′
Jrc ⋅ ωrt + t ωt × Jrc ⋅ t ωrt = φ̈⎢ ⎥ ⎢
⎣ Jrc ⎦ + φ̇⎣ ωt Jrc − ωt Jrc ⎦

between VFIFE and FAST(MoorDyn) increase with the complex combi­ xz
Jrc − t ωyt Jrcxx + t ωxt Jrcxy
nation of external loads (LC2 is a joint load of wind and wave). The most
likely reason for this deviation is that we assume the position of the three Equation (74) reveals the internal mechanism of the coupling be­
fairleads on the FOWT are always at the same level. In addition, a tween the rotor spin and platform motion. Usually, the values at non-
bilaterally weak coupling was assumed between structural module and diagonal position of inertia tensor Jrc are very small, thus, the equa­
mooring module in the proposed framework. We hypothesize that a tion (74) can be approximated as
bilaterally strong coupling between different modules of the framework ⎡ ⎤ ⎡ ⎤
will reduce these deviations in the maximum value of the mooring Jrcxx 0
( ) ⎢ ⎥ ⎢ ⎥
tension, as these responses are iterated and corrected before being used (75)
t ′
Jrc ⋅ ωrt + t ωt × Jrc ⋅ t ωrt = φ̈⎢ ⎥ ⎢ t z xx ⎥
⎣ 0 ⎦ + φ̇⎣ ωt Jrc ⎦
for platform position calculations in every time step. However, note that 0 − t ωyrt Jrcxx
this approach will make the calculations computationally expensive.
In LC3, the responses of VFIFE and FAST under the action of random The relationship between the rotor spin and the platform motion is
wave were compared. First, the wave elevation in time domain was explicitly seen from the above expression. The acceleration of the rotor
generate in VFIFE model, then the wave elevation was read-in by FAST, spin only affects the roll motion of platform. If φ̈ = 0, the additional
thus, the random wave is same in VFIFE model and FAST. Fig. 14 is the gyroscopic moments are proportional to the rotor spin speed and tower
comparison of VFIFE model and FAST for LC3 in time domain. angular velocity. The combination of the rotor spin and platform yaw
Compared with regular wave case, the differences between VFIFE and motion will excite platform pitch motion, and the combination of the
FAST are slightly obvious under the action of random wave, but the rotor spin and platform pitch motion will excite the platform yaw
VFIFE model still has good computational accuracy. Similar conclusions motion.

17
Y. Zhang et al. Ocean Engineering 262 (2022) 112221

Usually, the aerodynamic thrust will change with the rotor speed CRediT authorship contribution statement
under the action of wind loads. In order to examine the influence of the
rotor spin on the system dynamic response separately from any con­ Yu Zhang: Writing – original draft, Software, Visualization, Inves­
current effects of the aerodynamic loads, the aerodynamic loads are set tigation, Formal analysis. Wei Shi: Conceptualization, Methodology,
to zero. This means that the rotor is idling (without a generator load) Visualization, Investigation, Writing – review & editing, Supervision.
with a constant speed. Two cases are used to validate the rotor gyro­ Dongsheng Li: Conceptualization, Methodology, Investigation, Visual­
scopic effect. In case 1, the wind turbine is parked (φ̇ = 0), and the ization, Writing – review & editing, Supervision. Xin Li: Conceptuali­
platform is excited by a linear regular wave, in which the wave height is zation, Visualization, Writing – review & editing. Yuanfeng Duan:
6 m and peak period is 10 s. In case 2, the rotor idling at a constant spin Conceptualization, Methodology, Investigation. Amrit Shankar
rate of φ̇ = 9 rpm, with wave conditions equivalent to case 1. The dy­ Verma: Conceptualization, Visualization, Writing – review & editing.
namic responses obtained from the VFIFE model are then compared to
examine the rotor gyroscopic effect on the system dynamics.
From the Fig. 18, it is concluded that the gyroscopic moment Declaration of competing interest
resulting from 9 rpm rotor spin generates a significant yaw response
compared to the parked rotor. This result is consistent with our expec­ The authors declare that they have no known competing financial
tations from Eq. (40). In addition, the results of FAST are used to verify interests or personal relationships that could have appeared to influence
the accuracy of VFIFE model which shows excellent agreement. Further, the work reported in this paper.
Fig. 19 shows that the rotor spin has little influence on the platform
surge and pitch motion. This is because the aerodynamic loads and Data availability
hydrodynamic loads mainly dominate the platform surge and pitch
motion, and the component of the additional gyroscopic moments in the No data was used for the research described in the article.
y-axis is minor (Al-Solihat et al., 2019). In addition, for spar type plat­
form, the moment of inertia of yaw is smaller compared to the corre­ Acknowledgement
sponding roll and pitch, thus, the yaw motion can be more easily excited.
Note that the cut-in and rated rotor speed are 6.9 rpm and 12.1 rpm of This research was funded by the National Natural Science Founda­
NREL 5 MW, respectively. Therefore, a variation the platform yaw angle tion of China (Grant No. 52071058, 51939002). This work is also
at different rotor speeds is shown in Fig. 20. It is seen from these results partially supported by Guangdong Basic and Applied Research Foun­
that the platform yaw motion caused by gyroscopic moments becomes dation (Grant No. 2021A1515011770), LiaoNing Revitalization Talents
larger increase in rotor speed. Program (XLYC1807208) and special funds for promoting high quality
development from department of natural resources of Guangdong
6. Conclusions province (GDNRC [2020]016). Co-author Dr. Verma acknowledges the
start-up support provided by the Office of the Vice President for
This paper proposes a novel modeling framework for FOWTs based Research and Dean of the Graduate School, University of Maine.
on the VFIFE method. In this framework, the MBD method is used to
handle the rigid body motion, the analysis of structural deformation and References
the solution of governing EOMs are conducted based on the VFIFE
method. The mooring system is regarded as flexible and deformable, Al-Solihat, M., Nahon, M., Behdinan, K., 2019. Dynamic modeling and simulation of a
while the FOWT is regarded as an assemble of multi-rigid-bodies. The spar floating offshore wind turbine with consideration of the rotor speed variations.
J. Dyn. Syst. Meas. Control 141, 081014.
feasibility and accuracy of the VFIFE model is verified against the FAST.
Al-Solihat, M., Nahon, M., 2018. Flexible multi-body dynamic modeling of a floating
Then the influence of dynamic effects of mooring system on the mooring wind turbine. Int. J. Mech. Sci. 142–143, 518–529.
tension and the gyroscopic moments arise from rotor spin on the plat­ Barthelmie, R.J., Pryor, S.C., 2021. Climate change mitigation potential of wind energy.
Climate 9, 136.
form yaw are investigated. The following conclusions can be drawn:
Chen, J.H., Hu, Z.Q., Liu, G.L., Tang, Y.G., 2017. Comparison of different dynamic
models for floating wind turbines. J. Renew. Sustain. Energy 9 (6), 63304.
(1) The novel framework based on the VFIFE method is feasible and Chen, J., Yang, R., Zhao, Y., 2017. Application of vector form intrinsic finite element on
capable in dealing with the rigid-flexible coupling problem, and integrated simulation of wind turbine. Struct. Des. Tall Special Build. 26 (6), 1347.
Chen, L., Basu, B., Nielsen, S.R., 2019. Nonlinear periodic response analysis of mooring
the comparisons between FAST and VFIFE method validate the cables using harmonic balance method. J. Sound Vib. 438, 402–418.
accuracy of new model can be assured. Coulling, A.J., Goupee, A.J., Robertsonetc, A.N., Jonkman, J.M., Dagher, H.J., 2013.
(2) The results of free decay test and the platform motion as well as Validation of a FAST semi-submersible floating wind turbine numerical model with
DeepCwind test data. J. Renew. Sustain. Energy 5 (2), 23116.
the mooring tension calculated by VFIFE model have a good Ding, H., Han, Y., Le, C., Zhang, P., 2017. Dynamic analysis of a floating wind turbine in
agreement with the FAST(MoorDyn). The dynamic effects of wet tows based on multi-body dynamics. J. Renew. Sustain. Energy 9 (3),
mooring system will generate a significant influence on the ac­ 1–4982742.
Duan, Y.F., He, K., Zhang, H.M., Ting, E.C., Wang, C.Y., Chen, S.K., Wang, R.Z., 2014.
curacy of mooring tension calculation. Thus, the dynamic effects Entire-process simulation of earthquake-induced collapse of cable-stayed bridges by
of mooring system have a non-negligible influence on the fatigue Vector Form Intrinsic Finite Element (VFIFE) method. Adv. Struct. Eng. 17 (3),
life estimation. 347–360.
Duan, Y.F., Wang, S.M., Wang, R.Z., Wang, C.Y., Ting, E.C., 2017. Vector form intrinsic
(3) The gyroscopic moments arise from rotor spin will generate an
finite element based approach to simulate crack propagation. J. Mech. 33 (6),
apparent effect on the platform motion, especially the rotor spin 797–812.
will lead to a coupled motion of platform pitch and yaw. There­ Gu, H., Guo, H., Bai, F., Li, X., Li, F., 2022. Study on the collision probability of vertical
risers considering the wake interference effect. Ocean Eng. 245, 110583.
fore, it is necessary to consider the gyroscopic effects when
Hall, M., Buckham, B., Crawford, C., 2014. Evaluating the importance of mooring line
modeling a simple rigid body model of FOWT. model fidelity in floating offshore wind turbine simulations. Wind Energy 17 (12),
1835–1853.
Future studies will focus on the modeling of a more complex multi- Hall, M., Goupee, A., 2015. Validation of a lumped-mass mooring line model with
DeepCwind semisubmersible model test data. Ocean Eng. 104, 590–603.
body (such as four bodies) model of FOWT, and the tower and three Jonkman, J.M., 2007. Dynamic Modeling and Loads Analysis of an Offshore Floating
blades can be constructed as flexible bodies based on the presented Wind Turbine. NREL National Energy Renewal Laboratory. Technical Report NREL/
framework. TP-500-41958.
Jonkman, J.M., Butterfield, S., Musial, W., Scott, G., 2009. Definition of a 5-MW
Reference Wind Turbine for Offshore System Development. National Renewable
Energy Laboratory. Technical Report NREL/TP-500-38060.

18
Y. Zhang et al. Ocean Engineering 262 (2022) 112221

Jonkman, J.M., 2010. Definition of the Floating System for Phase IV of OC3, Report Shabana, A.A., 2013. Dynamics of Multi-Body Systems, fourth ed. Cambridge University
NREL/TP-500–47535. National Renewable Energy Laboratory, pp. 509–513. Press, London.
Karimirad, M., Gao, Z., Moan, T., 2009. Dynamic Motion Analysis of Catenary Moored Sweetman, B., Wang, L., 2014. Momentum cloud method for dynamic simulation of rigid
Spar Wind Turbine in Extreme Environmental Condition. Offshore Wind Conference, body systems. J. Eng. Mech. 140 (2), 257–267.
pp. 1–10. Stockholm, Sweden. September16-18. Shen, M., Hu, Z.Q., Liu, G.L., 2016. Dynamic response and viscous effect analysis of a
Karimirad, M., Moan, T., 2012. A simplified method for coupled analysis of floating TLP-type floating wind turbine using a coupled aero-hydro-mooring dynamic code.
offshore wind turbines. Mar. Struct. 27 (36), 45–63. Renew. Energy 99, 800–812.
Karimirad, M., Michailides, C., 2015. V-shaped semisubmersible offshore wind turbine: Sarkar, S., Fitzgerald, B., 2021. Use of Kane’s method for multi-body dynamic modelling
an alternative concept for offshore wind technology. Renew. Energy 83, 126–143. and control of spar-type floating offshore wind turbines. Energies 14, 6635.
Leble, V., Barakos, G., 2016. Demonstration of a coupled floating offshore wind turbine Tupling, S.J., Pierrynowski, M.R., 1987. Use of cardan angles to locate rigid bodies in
analysis with high-fidelity methods. J. Fluid Struct. 62, 272–293. three-dimensional space. Med. Biol. Eng. Comput. 25 (5), 527.
Li, Y., Castro, A.M., Martin, J.E., Sinokrot, T., Prescott, W., Carrica, P.M., 2017. Coupled Ting, E.C., Duan, Y.F., Wu, T.Y., 2012. Vector Mechanics of Structure. Science Press,
computational fluid dynamics/multi-body dynamics method for wind turbine aero- Beijing, China.
servo-elastic simulation including drivetrain dynamics. Renew. Energy 101 (FEB.), Ting, E.C., Shih, C., Wang, Y.K., 2004a. Fundamentals of a vector form intrinsic finite
1037–1051. element: Part I. Basic procedure and A plane frame element. J. Mech. 20 (2),
Liu, Y., Xiao, Q., Incecik, A., Peyrard, C., 2019. Aeroelastic analysis of a floating offshore 113–122.
wind turbine in platform-induced surge motion using a fully coupled CFD-MBD Ting, E.C., Shin, C., Wang, Y.K., 2004b. Fundamentals of a vector form intrinsic finite
method. Wind Energy 22 (1), 1–20. element: Part II. Plane solid elements. J. Mech. 20 (2), 123–132.
Li, X., Guo, X., Guo, H., 2018a. Vector form intrinsic finite element method for the two- Tan, J., Zhu, C., Song, C., Li, Y., Xu, X., 2019. Dynamic modeling and analysis of wind
dimensional analysis of marine risers with large deformations. J. Ocean Univ. China turbine drivetrain considering platform motion. Mech. Mach. Theor. 140 (140),
17 (3), 498–506. 781–808.
Li, X., Guo, X., Guo, H., 2018b. Vector form intrinsic finite element method for nonlinear Tan, B., Cao, S., Fang, G., Cao, J., Ge, Y., 2021. openVFIFE: an object-oriented structure
analysis of three-dimensional marine risers. Ocean Eng. 161, 257–267. analysis platform based on vector form intrinsic finite element method. Buildings 11,
Li, X., Wei, W., Bai, F., 2020. A full three-dimensional vortex-induced vibration 505.
prediction model for top-tensioned risers based on vector form intrinsic finite Wang, Y., Shi, W., Michailides, C., Wan, L., Kim, H., Li, X., 2022. WEC shape effect on the
element method. Ocean Eng. 218, 108140. motion response and power performance of a combined wind-wave energy
Li, X., Chen, D., Gu, H., Bai, F., 2021. Three-dimensional vortex-induced vibration converter. Ocean Eng. 250, 111038.
analysis of catenary-type risers under flow with different incident angles. Ocean Eng. Wang, L., Sweetman, B., 2013. Multi-body dynamics of floating wind turbines with large-
240, 109978. amplitude motion. Appl. Ocean Res. 43 (5), 1–10.
Ma, Z., Wang, S., Wang, Y., Ren, N., Zhai, G., 2019. Experimental and numerical study on Wang, L., Sweetman, B., 2012. Simulation of large-amplitude motion of floating wind
the multi-body coupling dynamic response of a novel Serbuoys-TLP wind turbine. turbines using conservation of momentum. Ocean Eng. 42, 155–164.
Ocean Eng. 192, 106570. Wang, S.M., Ni, Y.Q., Duan, Y.F., Yau, J.D., 2021. Vector form intrinsic finite element
Nielsen, F.G., Hanson, T.D., Skaare, B., 2006. Integrated dynamic analysis of floating method for stochastic analysis of train–track–bridge coupling system. Int. J. Struct.
offshore wind turbines. In: Proceedings of the 25th International Conference on Stabil. Dynam., 2140012
Offshore Mechanics and Arctic Engineering, vol. 1. Offshore Technology; Offshore Xu, P., Du, Z., Huang, F., Javanmardi, A., 2021. Numerical simulation of deepwater S-lay
Wind Energy; Ocean Research Technology; LNG Specialty Symposium, Hamburg, and J-lay pipeline using vector form intrinsic finite element method. Ocean Eng.
Germany, pp. 671–679. June 4–9. 234, 109039.
Park, K.P., Cha, J.H., Ku, N., 2016. The flexible multi-body dynamics of a floating Xu, P., Du, Z., Zhang, T., Chen, B., 2022. Vector form intrinsic finite element analysis of
offshore wind turbine in marine operations. Ships Offshore Struct. 12 (4), 1–12. deepwater J-laying pipelines on sloping seabed. Ocean Eng. 247, 110709.
Peng, Z., Zhao, H., Li, X., 2021. New ductile fracture model for fracture prediction Yuan, X.F., Chen, C., Duan, Y.F., Qian, R.J., 2018. Elastoplastic analysis with fine beam
ranging from negative to high stress triaxiality. Int. J. Plast. 145, 103057. model of vector form intrinsic finite element. Adv. Struct. Eng. 21 (3), 365–379.
Roald, L., Jonkman, J., Robertson, A., Chokani, N., 2013. The effect of second-order Zhang, L., Shi, W., Karimirad, M., Michailides, C., Jiang, Z., 2020. Second-order
hydrodynamics on floating offshore wind turbines. Energy Proc. 35, 253–264. hydrodynamic effects on the response of three semisubmersible floating offshore
Ren, Y., Vengatesan, V., Shi, W., 2022. Dynamic analysis of a multi-column TLP floating wind turbines. Ocean Eng. 207, 107371.
offshore wind turbine with tendon failure scenarios. Ocean Eng. 245, 110472. Zhang, Y., Shi, W., Li, D., Li, X., Duan, Y., 2022. Development of a numerical mooring
Sarpkaya, T., Isaacson, M., Wehausen, J.V., 1982. Mechanics of wave forces on offshore line model for a floating wind turbine based on the vector form intrinsic finite
structures. J. Appl. Mech. 49 (2), 466–467. element method. Ocean Eng. 253, 111354.
Shin, C., Wang, Y.K., Ting, E.C., 2004. Fundamentals of a vector form intrinsic finite Zeng, Y., Shi, W., Michailides, C., Ren, Z., Li, X., 2022. Numerical Investigation of the
element: Part III. Convected material frame and examples. J. Mech. 20 (2), 133–143. Hydrodynamics of an Oscillating Water Column (OWC) with Computational Fluid
Dynamics. Energy accepted.

19

You might also like