Energy 154 (2018) 442e454

Contents lists available at ScienceDirect

Energy
journal homepage: www.elsevier.com/locate/energy

An open-source comprehensive numerical model for dynamic

response and loads analysis of floating offshore wind turbines
M. Barooni a, *, N. Ale Ali a, T. Ashuri b
a
Department of Ocean Engineering, Khorramshahr Marine Science and Technology University, Khoramshahr, Iran
b
Department of Mechanical Engineering, Arkansas Tech University, Russellville, AR 72801, USA

a r t i c l e i n f o a b s t r a c t

Article history: This paper presents the development of a comprehensive open-source numerical model to study the
Received 6 March 2017 dynamic response and load analysis of floating offshore wind turbines. The model accounts for the wind
Received in revised form inflow, rotor aerodynamics, multibody structural model of the system, wave and current kinematics,
2 April 2018
hydrodynamics, and mooring-line dynamics. This coupled simulation tool can be used for analysis,
Accepted 27 April 2018
Available online 30 April 2018
optimization and preliminary design to determine the technical and economic feasibility. Several veri-
fication and validation cases are performed to show the correctness of the numerical simulations. The
results show that the proposed approach provides an accurate estimate of the wind turbine dynamics
Keywords:
Floating offshore wind turbine
and loads. The simulation tool is then applied in the analysis of a 5 MW wind turbine aimed to char-
Dynamic response analysis acterize the dynamic response and to identify potential loads and instabilities resulting from the dy-
Load analysis namic couplings between the turbine and the external conditions. This open-source fully coupled aero-
Wind turbine design hydro-elastic model provides a modular framework to enable investigating a variety of wind turbine
Numerical model configurations, support systems, and mooring lines. Therefore, it is expected that researchers and design
engineers worldwide use the model to study, investigate and analyze different aspects of floating
offshore wind turbine design, which results is the promotion and advancement of science and tech-
nology for floating offshore wind turbines.
© 2018 Published by Elsevier Ltd.

1. Introduction allows the generation of electricity in deep water depths where

fixed bottom-mounted towers are not economically feasible.
Global warming and climate change are among the grand However, FOWTs are one of the most challenging ocean structures
challenges of the new millennium [1e3]. Renewable energies have to design, since they require a multidisciplinary approach to
the potential to address these challenges by providing a clean analyze different coupled disciplines such as aerodynamics, hy-
resource of sustainable energy [4,5]. Among all sources of renew- drodynamics, structural dynamics and controls [21,22]. Therefore,
able energy, offshore wind is promising, since it allows bulk gen- FOWTs are still in their infancy stage of technology readiness, and
eration of electricity [6]. However, the cost of offshore wind continuous research and development is needed to better under-
generated electricity is still on average higher than that of fossil stand the system's dynamics and propose cost effective designs
fuels. This encourages the development of science and technology [23,24].
to reduce the costs by developing larger offshore wind turbines In the recent years, several researchers investigated different
[7e10], novel support structures [11e13], advanced control algo- aspects of modeling FOWTs to better understand their complex
rithms [14,15], and new design techniques [16e18]. dynamic behavior. Jonkman [25] presented a computational code
A floating offshore wind turbine (FOWT) is among the concepts to model the aerodynamics of the rotor, the structural motion, the
that has the potential to be effective and economic in extracting hydrodynamics of the wave, and a blade-pitch and generator
energy from the vast offshore wind resources in deep waters controller for time-domain simulation. Sclavounos et al. [26]
[19,20]. A FOWT is a turbine mounted on a floating structure that developed two low-weight, motion resistant stiff floating wind
turbine concepts for deployment in water depths ranging from 30
to several hundred meters. They obtained linear and nonlinear
* Corresponding author. wave loads on the floater using uncoupled computational methods
E-mail address: [email protected] (M. Barooni).

https://doi.org/10.1016/j.energy.2018.04.163
0360-5442/© 2018 Published by Elsevier Ltd.
 M. Barooni et al. / Energy 154 (2018) 442e454 443

developed for the design of oil and gas offshore platforms. github.com/tashuri.
Karimirad and Moan [27] developed a stochastic dynamic model In BA-Simula, the dynamic response due to external and inertial
of a tension leg spar-type wind turbine subjected to wind and wave loads is obtained using a fully coupled numerical model imple-
action. They implemented the model using the HAWC2 aero-hydro- mented in MATLAB®. Blade element momentum (BEM) theory is
elastic code and analyzed the dynamic motion of the structure, used to determine aerodynamic loads on the rotor [35]. Panel
power production and tension leg responses. Skaare et al. [28] method and Morison's equation are used to calculate the hydro-
performed an analysis between full-scale measurements from the dynamic loads considering the instantaneous position of the wind
floating wind turbine Hywind demo, and corresponding numerical turbine system [36]. Mooring loads are calculated by a quasi-static
simulation. Wang et al. [29] presented a stochastic dynamic equilibrium at different time steps [37]. The fully coupled equations
response analysis of a 5 MW Floating vertical-axis wind turbine of motion are solved using Runge-Kutta method [38]. The accuracy
(FVAWT) based on fully coupled nonlinear time domain simula- of the code is tested using several model-to-model comparisons,
tions. The turbine has a Darrieus rotor, and a semisubmersible and validation with experimental data as found in the literature.
floater subjected to various wind and wave loads. This gives the confidence to perform more thorough analyses, and
Vaal et al. [30] investigated the effect of a periodic surge motion it provides the users the necessary insight needed to analyze the
on the integrated loads and induced velocity on a wind turbine dynamic behavior and design of FOWTs. This contributes to
rotor. Through the analysis of the integrated rotor loads, induced advancement of science and technology for FOWTs, which in turn
velocities and aerodynamic damping, it is concluded that typical reduces the costs.
surge motions are sufficiently slow to affect the wake dynamics The remainder of the paper is organized as follows. First, the
predicted by engineering models. methodology and the mathematical formulation to model the fully
Vorpahl et al. [31] presented a benchmark study on aero-servo- coupled nonlinear FOWT is presented. Second, the verification and
hydro-elastic codes for offshore wind turbine dynamic simulation. validation of the code is illustrated. Third, numerical simulations
The verified codes account for the coupled dynamic systems are performed to investigate the influence of several input condi-
including the wind inflow, aerodynamics, elasticity and controls of tions on a spar type FOWT dynamic response and loads. Finally,
the turbine, along with the incident waves, sea current, hydrody- conclusions and future works are drawn and presented.
namics and foundation dynamics of the support structure.
As the literature shows, limited efforts have been made to 2. Methodology
develop a comprehensive open-source numerical model to be
shared publicly for investigating different aspects of the design and This section presents the development of a comprehensive nu-
analysis of FOWTs. Therefore, we still lack numerical models, and merical model to study the dynamics and load analysis of FOWTs.
their corresponding mathematical formulation [32e34]. To address As Fig. 1 shows, there are several support structure configurations
this shortcoming, this paper presents the development of the used for FOWTs. A catenary moored spar type is the commonly
governing equations of a fully coupled nonlinear FOWT, and it used configuration that consists of a single floating cylindrical spar-
presents a public open-source computational code for analysis, buoy moored by catenary cables. This research uses the 5 MW
optimization and preliminary design. This new simulation code is National Renewable Energy Laboratory (NREL) wind turbine that
named BA-Simula, and it is available to download from https:// has a catenary moored spar type support structure. As Fig. 2 shows

Fig. 1. Different FOWT support structure used as the floater in large water depths [40].
444 M. Barooni et al. / Energy 154 (2018) 442e454

The first coordinate system is global, and it is fixed at sea bed. The
three other coordinate systems are local. One is located at the
center of gravity (CG), the other at hub height, and another in the
blade root.

2.1. Aerodynamic loads

Blade aerodynamic loads are calculated using BEM theory [35].

BEM combines momentum theory, and blade element theory to
compute the loads on the blade iterativly. First, the momentum
balance on a rotating annular stream tube passing through a tur-
bine is computed. Second, lift and drag forces on spanwise cross
sections of the blade are computed. These two separate theories
run iterativly until the desired convergence on the aerodynamic
loads is achieved [41]. In BEM theory, the perpendicular force (Fp ),
and tangential force (Ft ) are expressed as a function of the drag, D,
and lift, L, forces [42]. That is:

Fp ¼ Nb ðL cos f þ D sin fÞ
¼ 1=2rair W 2 Nb cðCl cos f þ Cd sin fÞDr (1)

Ft ¼ Nb ðL sin f  D cos fÞ
¼ 1=2rair W 2 Nb cðCl sin f  Cd cos fÞDr (2)

Here, rair is air density, W is resultant wind velocity, Nb is number of

blades, c is chord length, Cl is lift coefficient, Cd is drag coefficient, f
is inflow angle, and Dr is the length of blade element.
As shown in Fig. 3, the resultant wind velocity W is the vector
sum of the axial (Vax ), and tangential (Vt ) wind velocities on the
blade element, and it can be expressed as:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
W¼ V 2ax þ V 2t (3)

In this figure, a is the angle of attack, and g1 and g2 are the blade
twist and pitch angle, respectively. Due to the presence of the rotor,
the axial and tangential wind velocities are not identical to free
stream and rotational speed. Therefore, axial (a), and tangential (a0 )
induction factors are used to represent the fractional change in
Fig. 2. Configuration of the 5 MW NREL floating wind turbine, and the used coordinate wind velocity due to rotor presence as:
systems.
Vax ¼ V∞ ð1  aÞ (4)

the turbine has a draft of 120 m at a water depth of 320 m [39]. Vt ¼ Urð1 þ a0 Þ (5)
Table 1 presents the main properties of the 5 MW NREL FOWT.
To describe aerodynamic, hydrodynamic and structural loads, where V∞ is free stream velocity, U is angular velocity of the rotor,
four different coordinate systems are used as presented in Fig. 2. and r is blade element radius.

Table 1
Overall system properties of the 5 MW NREL wind turbine [39].

Property Value

Rotor configuration () 3 blades, upwind

Rotor and hub diameter (m) 126; 3
Hub height (m) 90
 
Cut-in, rated, and cut-out wind speed ms = 3; 11:4; 25

Cut-in, and rated rotor speed (rpm) 6:9; 12:1

Draft (m) 120
Diameter above taper (m) 6.5
Diameter below taper (m) 9.4
Total mass (kg) 8.06E8
Overall center of gravity 78.0
(along the centerline of the platform) (m)
Pitch inertia about the center of gravity (kg m2 ) 6.80E10
Yaw inertia about the centerline (kg m2 ) 1.92E8
Fig. 3. Velocity and force components at the rotor plane.
 M. Barooni et al. / Energy 154 (2018) 442e454 445

Knowing the induced wind velocity at each blade element in

time, and taking into account velocity of blade elements induced by
blade motion, the loads on each blade element can be calculated as:
 2 
L ¼ 1=2rair c Cl Vax  x_p þ ðVt  x_t Þ2 Dr (6)

 2 
D ¼ 1=2rair c Cd Vax  x_p þ ðVt  x_t Þ2 Dr (7)

where x_p and x_t are velocities of blade element in perpendicular

and tangential directions to rotor plane, respectively.
In the above formulation, the free stream velocity is computed
at the hub-height of the wind turbine, and it can be either steady or
turbulent [43,44]. For the simulation of the turbulent wind, the von
Karman spectrum is used [45].

2.2. Hydrostatic and restoring loads

Hydrostatic loads are found using Archimedes' principle [46].

Using this principle, the total weight of the wind turbine, platform,
and mooring cables in water is equal to the weight of the displaced
water volume. Buoyancy force (Fb ) is calculated at the center of
buoyancy (CB). The general form of buoyancy force equation can be
expressed as:
 
Fb ¼ rsea g V0  Asurf uZ þ Asurf hsurf (8)

Here, rsea is water density, g is the gravitational acceleration, V0 is

displaced water volume, Asurf is spar area at water surface, uZ is
translation in Z-direction in global coordinate system, and hsurf is
wave elevation. The restoring moment arm of the buoyancy force is
given by:
 
 
CB  CG ¼ ðV1 r1 þ V2 r2 þ V3 r3 Þ (9)
  ðV þ V þ V Þ
1 2 3

where, V1 , V2 and V3 are displaced water volumes by spar upper,

middle and lower sections, respectively, and r1 , r2 and r3 are the Fig. 4. Spar platform diagram to compute the buoyancy force.
corresponding vertical distances between each volumetric section
CG and the entire system CG. These are shown in Fig. 4.
Wave particle kinematics is obtained for each time step based
on the Airy linear wave theory [50], and used for calculating the
2.3. Hydrodynamic loads
hydrodynamic forces for each section considering the instanta-
neous position of each strip. Due to deep draft of the spar, an
Hydrodynamic loads are calculated by combining Morison
alternative approach for calculating hydrodynamic forces on the
equation and strip panel method. Morison equation is widely used in
bottom section of the spar is used, which consist of Morison inertia
the analysis of fixed bottom offshore structures such as spars [36]. In
term and a fluid damping term [51] as:
strip panel method, the structure is split into a number of elements
[47], and Morison equation is used to compute linear wave, and  
Z
Fhydro ¼ Bm z€s  Cvd z_s z_s (11)
nonlinear viscous drag loads on each element [39]. In this paper,
Morison equation is modified to take into account the moving
where, Bm is the vertical added mass that is assumed to be equal to
element due to the motion of the entire structure [36]. That is:
half of a sphere of water below the spar [36], and z_s and z€s are
1   vertical velocity and acceleration components of spar, respectively,
fhydro ¼ rsea Cd ds ðu þ c  q_ s Þu þ c  q_ s dz þ rsea As ðcm h
€ and Cvd is fluid damping.
2
 ca q€s Þdz (10)
2.4. Mooring loads
Here, ds is the diameter of spar, u and c are wave particle and
current velocities, q_ s and q€s are the body velocity and acceleration, The mooring lines are modeled using a quasi-static method that
As is cross-sectional area of spar, h € is the local acceleration of wave finds the tension of the cables by assuming its static equilibrium in
particles, and dz is the unit length of spar section. Cd , Cm and Ca are each time step. The model accounts for the apparent weight in
viscous-drag, inertia and added mass coefficients that are found fluid, elastic stretching and seabed friction of each line, but it ne-
empirically [48]. These parameters are functions of Reynolds glects the individual line bending stiffness, and the inertia and
number, Kaulegan-Carpenter number, relative current number, and damping of the mooring system. However, the nonlinear geometric
surface roughness ratio [49]. restoration of the whole mooring system is considered [25].
446 M. Barooni et al. / Energy 154 (2018) 442e454

Fig. 5. Mooring line in a local coordinate system. Fig. 6. Mooring line positions and net forces.

Knowing the fairlead position at any instant of time, the located at the system's CG with unit vectors bi;bj; b
k. The second local
following two nonlinear equations are solved using Newton- coordinate system is positioned at the hub center with b q and bl
p; b
Raphson technique [52]. As Fig. 5 presents, the horizontal and unit vectors. Unit vector b p is aligned with the main shaft of the
vertical components of the effective tension in the mooring line turbine, and b q and bl form a perpendicular plane to it. The fourth
denoted by HF and VF are computed as: coordinate system is defined at the blade root with unit vectors b1 ,
2 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 3 b2 and b3 . Transformation between different coordinate systems is

VF V HF VF 2 5 HF L accomplished with a transformation matrix. As an example, the
xF ðHF ; VF Þ ¼ L  þ ln4 F þ 1þ þ transformation between the CG coordinate, and the global coordi-
u u HF HF EA
nate system is given by:
"    #
C u VF 2
HF VF VF HF 2
bI
3 2 3
bi
þ B  L þ L  max L   ;0
2EA u u CB w u CB u 4 bJ 5 ¼ R4 bj 5 (17)
(12) b
K b
k
2sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 3
  With:
HF 4 VF 2 VF  wL 2 5
zF ðHF ; VF Þ ¼ 1þ  1þ 2      3
u HF HF cos bI; bi cos bI; bj cos bI; b
k
(13) 6      7
 6 7
1 WL2 R ¼ 6 cos bJ; bi cos bJ; bj cos bJ; b
k 7 (18)
þ VF L  4      5
EA 2 cos Kb ; bi cos Kb ; bj cos Kb; b
k

where, L is the total unstretched length, u is the apparent weight in A similar matrix is used to make the transformation across
fluid per unit length, EA is extensional stiffness, CB is coefficient of different coordinates. As an example, a chain of coordinate system
seabed static-friction drag, and xF and zF represent the fairlead transformation is used to first compute the aerodynamic loads in
location relative to the anchor. the blade coordinate system, and then transform it to the hub, and
As illustrated in Fig. 6, the mooring reaction forces in global then CG coordinate system.
coordinate system are found by:
2.6. Kinematics of FOWT
X
Fmooring ¼ HF1 cosðbline1 Þ þ HF2 cosðbline2 Þ þ HF3 cosðbline3 Þ (14)
The rotational velocity, u, of the FOWT can be determined with
Y
Fmooring ¼ HF1 sinðbline1 Þ þ HF2 sinðbline2 Þ þ HF3 sinðbline3 Þ (15) respect to CG as:
h i h i h
Z u ¼ q_ cos f cos j þ fsin
_ j bi  q_ cos f sin j þ fcos
_ j bj þ j_
Fmooring ¼ VF1 þ VF2 þ VF3 (16)
i
þ q_ sin f bk
Here, bline is the mooring line angle as shown in Fig. 6.
(19)

2.5. Coordinate systems Here, f, q and j are relative angles between the unit vectors of the
global coordinate system and CG, and the dot overhead shows the
As mentioned earlier, four different coordinate systems are used derivative with respect to time.
to describe the equations of motion. One global coordinate system To account for relative velocities experienced by the rotor due to
that is fixed at the sea bed and defined by bI; bJ; K
b unit vectors, and the motion of the structure, the system's rotational velocity
three local coordinate systems. The first local coordinate system is computed at CG is transferred to the blade as:
 M. Barooni et al. / Energy 154 (2018) 442e454 447

  
u ¼ ux b1 þ uy cosUt þ uz sinUt b2 þ  uy sinUt ab ¼ u_ b1 b1 þ u_ b2 b2 þ u_ b3 b3 þ !
u  ðUb1 Þ (29)

þ uz cosUt b3 (20)

Here, U is the rotational velocity of the rotor, and t is time.

2.7. Equations of motion
Thus, the blade's rotational velocity vector can be expressed as:

! The equations of motion for the blade are defined as:

u b ¼ Ub1 þ !
u (21)
M1 ¼ Me  I1 ab1
Therefore, the components of !
(30)
u b are:

ub 1 ¼ ux þ U (22) M2 ¼ I2 ab2  ðI2  I1 Þub1 ub3 (31)

ub2 ¼ uy cosUt þ uz sinUt (23) M3 ¼ I3 ab3  ðI3  I1 Þub1 ub2 (32)

where, M1 , M2 and M3 are moments around each blade axis, I1 , I2

ub3 ¼ uy sinUt þ uz cosUt (24) and I3 are the mass moment of inertia around each blade axis, and
Me is the blade torque defined as:
Similarly, the acceleration vector of the FOWT with respect to
the CG coordinate system is:
Me ¼ Ftaero r (33)

a ¼ axbi þ aybj þ az bk (25) Here, Ftaero is the tangential aerodynamic force on the blade, and r is
the moment arm.
In this formulation, the components of a are given by:
The modified Euler equation is used for deriving the dynamic
equations of motion for translational modes by Ref. [53]:
ax ¼ €qcos f cos j þ fsin
€ j  q_ j_ sin f cos j  q_ j_ cos f sin j
! ! !aero !hydro !mooring
þ f_ j_ cos j M u€ ¼ W þ F þ F þ F (34)
(26) !
Here, M is the mass matrix of the entire system, u€ is translational
!
acceleration vector and W is weight vector. This equation is solved
ay ¼ €qcos f sin j þ fcos
€ j þ q_ j_ sin f sin j  q_ j_ cos f cos j in time to obtain the dynamics of the FOWT using the 4th -order
 f_ j_ sin j Runge-Kutta method. Dynamic equations of motion for rotational
modes are given as:
(27)
 
Ixx ax þ Izz  Iyy uy uz ¼ Mx þ M1 (35)
€ þ€
az ¼ j qsin f þ q_ fcos
_ f (28)
Iyy ay  ðIzz  Ixx Þux uz ¼ My þ M2 cosUt  M3 sinUt (36)
This allows the computation of the relative rotational accelera-
tion vector of the blade as:

Fig. 7. BA-Simula code structure. In this figure, Hwave and Twave are wave high and period, respectively, and S is the output list of all translational and rotational modes of the wind
turbine.
448 M. Barooni et al. / Energy 154 (2018) 442e454

for this purpose. First, the aerodynamic module of BA-Simula is

verified with QBlade [54]. QBlade is an open-source wind turbine
code developed by Hermann Fo €ttinger Institute of TU Berlin, and
distributed under General Public License.
Figs. 8e11 illustrate important parameters in aerodynamic
analysis module such as axial and tangential induction factors, and
axial and tangential forces on individual blades. These results are
for a constant wind velocity of V ¼ 11:4 m=s. As the figures show, a

Fig. 8. Axial induction factor.

 
Izz az  Ixx  Iyy ux uy ¼ Mz þ M2 sinUt þ M3 cosUt (37)

where Ixx , Iyy and Izz are system's mass moments of inertia with
respect to CG, and Mx , My and Mz are the total excitation moments.

2.8. Block diagram of the code

As Fig. 7 shows, BA-Simula consists of five function blocks. Each Fig. 10. Tangential induction factor.

block calculates the loads acting on the FOWT at different time

steps. The equations of motion function block solves iteratively the
translational and rotational modes for the user defined initial
conditions, and inputs.
Inputs of the code are: wind data, wave and sea data, rotor
aerodynamic chord and twist data, rotor polar data, turbine oper-
ational data, mass and geometry of the turbine, blade-pitch and
generator controller data, mooring and spar data, as well as the
initial condition and simulation data. Outputs of the code are: time-
series of the loading on the structure at locations such a blade root,
tower, and mooring lines, displacements, velocities and accelera-
tions, as well as the plot of important information as presented in
the results section of the paper.

3. Verification and validation

This section presents the verification and validation of different

function blocks of BA-Simula. The 5 MW NREL wind turbine is used Fig. 11. Tangential force.

Fig. 9. Axial force. Fig. 12. Rotor thrust comparison.

 M. Barooni et al. / Energy 154 (2018) 442e454 449

Fig. 13. Mooring line tension comparison.

Fig. 16. Free decay heave motion time history.

Figs. 12 and 13 show the rotor thrust and tension in the mooring
line. A code-to-code comparison is made with FAST [56]. FAST is an
aero-hydro-elastic code developed by NREL. The shown distance in
Fig. 13 is the horizontal distance between fairlead and anchor. This
good agreement indicates the correct implementation of the
aerodynamic and mooring line modules.
Figs. 14e16 compare the platform surge, sway and heave free
decay time series in BA-Simula and FAST. For this comparison, all
external forces are eliminated, and an identical initial displacement
in the surge motion is considered. As the results show, BA-Simula
and FAST exhibit comparable free decay motion.
To investigate the forced dynamic response of BA-Simula model,
two load cases are considered and compared with FAST. First, a
regular wave with a wave height of H ¼ 1:4 m, and period of
T ¼ 10 s is considered. Figs. 17 and 18 show the surge and heave
motions. The results show in general a good agreement. Small
Fig. 14. Free decay surge motion time history. variations in these results relate to the amount of damping used in
each code. In BA-Simula, we added a constant damping for all
modes. To confirm this, we varied the damping value in BA-Simula
good agreement exists between a model-to-model comparison in
to see the sensitivity of the results, and if we can obtain closer re-
BA-Simula and QBlade. Minor differences can be explained by the
sults to FAST predictions. We observed that the amount of damping
way QBlade models the airfoil polar data. In QBlade, airfoil polar
can result in a better match of the dynamic response outputs.
data are modeled using the XFLR5 code [55]. XFLR5 is integrated
However, it is not clear which code uses a more realistic value for
into QBlade to generate airfoil lift and drag coefficients for turbine
damping and further investigation is needed to fully address this
aerodynamic simulations. In BA-Simula, the user has the freedom
issue. Therefore, no conclusion can be made about the accuracy of
to use any airfoil polar data as long as they follow the format used in
each code. This shows the importance and the need for conducing
BA-Simula. Therefore, small changes in the airfoil polar data are
experiments to advance this field of science and technology.
responsible for the minor differences in these figures.

Fig. 15. Free decay sway motion time history. Fig. 17. Surge motion time history (only wave).
450 M. Barooni et al. / Energy 154 (2018) 442e454

Fig. 18. Heave motion time history (only wave). Fig. 20. Heave motion time history (wind & wave).

Second, a regular wave with a wave height of H ¼ 1:4 m and replacing mooring delta line with an equivalent spring stiffness
period of T ¼ 10 s is considered, while the turbine is in operation at that can be adjusted to agree with other codes. However, this
a wind speed of V ¼ 11:4 m=s. The results are shown in Figs. 19 and parameter is not changed intentionally here to show the impact of
20 for surge and heave modes, respectively. Again, a good agree- making different modeling assumptions. Fig. 21d shows the varia-
ment is found between BA-Simula and FAST. Here, damping terms tion in the fairlead tension, and it can be concluded that BA-Simula
are added to BA-Simula to make up for diffraction and radiation shows comparable results to other codes.
that is not represented in Morison equation. These added damping The main reason for having slightly different dynamic response
terms can cause instability in some load cases, since they are sen- among these codes is the hydrodynamics model. Two different
sitive to initial conditions. approaches are used to model offshore spar-buoy floating wind
BA-Simula is also compared with a code-to-code comparison turbines. In the first approach, hydrodynamics loads are modeled
study present in the literature [31]. The comparison is performed using Morisons equation that is augmented with hydrostatics and
using a regular wave with a wave height of 6 m, a wave period of wave excitation heave forces. In the second approach, hydrody-
10 s, and a constant wind speed of 8 m=s. namic loads are modeled using potential-flow theory that is
Fig. 21aed shows the results for surge, heave and yaw motion, as augmented with the nonlinear viscous drag term from Morisons
well as fairlead loads. Variation in the results is related to different equation. Both hydrodynamic models are valid to predict equiva-
damping values, different equation solvers and time steps, and lent hydrodynamic loads on the spar, since in most of the condi-
different assumptions and simplifications used in each code. tions the radiation damping is negligible.
Fig. 21a shows the platform surge displacement, and all codes For codes such as BA-Simula that employ the first approach, the
except HAWC2 agree on the amplitude of oscillation. For the heave heave force can be calculated based on the variation in buoyancy
motion depicted in Fig. 21b, BA-Simula shows a heave over- force, and using a direct integration of the time-dependent hy-
estimation as the result of having less hydrodynamic restoring drostatic pressure and wave elevation. There are some additional
forces and damping in heave motion. This high amplitude motion is linear hydrodynamic damping based on measurements to be
related to the use of Morison equation for the calculation of hy- included in the model. Codes that neglect this additional damping
drodynamic excitations in vertical direction that is not well exhibit less overall damping in their responses, as is the case for BA-
damped. Simula as well [31].
The yaw oscillation shown in Fig. 21c agrees reasonably well Another difference among these models is the physical repre-
among all codes and the low amplitude in BA-Simula is the result of sentation of the mooring system. The two main approaches for
modeling the mooring system dynamics are the dynamic formu-
lation and the quasi-static formulation. In the dynamic formulation,
the mooring system is modeled using time-accurate Finite Element
Methods. The quasi-static formulation is time-dependent but is
slow enough to ignore the inertial effects. All of the compared codes
use the quasi-static formulation of the complete mooring system,
and are therefore only valid for small time-increments and small
displacements. Depending on the amount of damping used in each
code, some load cases may exhibit numerical instability if large
time-steps are selected [31].
In general, it is difficult to conclude which one of these codes
predict more accurate results. This is due to the fact that the limited
existing numerical codes are at the stage of infancy when it comes
to modeling harsh environmental conditions. Therefore, they
require more revision and future work for validating their under-
lying model assumptions with experimental data as it is evident
from the comparisons presented in this work.
BA-Simula is also validated against the limited publicly available
experimental data in the literature. The experimental validation
Fig. 19. Surge motion time history (wind & wave).
 M. Barooni et al. / Energy 154 (2018) 442e454 451

Fig. 21. Time series with regular wave with a wave height of 6 m, a wave period of 10 s, and a constant wind speed of 8 m=s. Platform surge, heave and yaw displacements; and
downstream fairlead tension.

cases are based on a scaled model of the 5 MW NREL spar-type The motion trajectory graph illustrated in Fig. 23 indicates the
floating wind turbine [57]. path of CG in 3D space for 5000 s simulation time for this case
In the first validation case, the heave motion time history of the study. As it can be seen, the system oscillates around an equilibrium
wind turbine is compared with the experimental data as presented region, and the closed elliptical orbits indicate that the 3D motion is
in Fig. 22a. The comparison is performed using an irregular wave stable.
with a wave height of 4 m, a wave period of 13 s, and a constant Fig. 24 shows the Poincare map. This map shows that the system
wind speed of 17 m=s. In the second validation case, the pitch is chaotic, and the 3D motion of the system is n-periodic. When a
motion time history is compared with experimental data as system is exhibiting a chaotic response, it means that the behavior
depicted in Fig. 22b. In this case, an irregular wave with a wave of the dynamical system is highly sensitive to initial conditions. In
height of 15.3 m, a wave period of 15.5 s, and a constant wind speed such a situation, each point in a chaotic system can arbitrarily be
of 30 m=s is used. approximated by other points, with significantly different future
As the figures show, the results of BA-Simula match well with paths, or trajectories. Therefore, an arbitrarily small change, or
the experimental data. It should be noted that the stochastic wave perturbation of the current trajectory may lead to significantly
kinematics used in BA-Simula are imported from ANSYS AQWA® for different future behavior [58].
this study. Fourier fast transformation (FFT) method is used to derive the
power spectrum of motion modes in frequency domain. In Figs. 25
and 26, the excitation frequency spectrum, and spectrum of surge
4. Numerical simulation results motion are represented as an example.
As Fig. 25 shows, the surge motion response has a maximum
To show some of the capabilities of BA-Simular, and the type of amplitude at zero frequency that indicates non-vibrating motion.
analysis the user can perform, this section presents few case studies The surge motion response also is centralized in low frequency
using the 5 MW NREL wind turbine. In the first case study, the zone, which coincides with its natural frequency. Eliminating
turbine is analyzed using a constant wind speed of 11:4 m=s, a external excitation forces, and considering initial conditions of 5, 2
single sinusoidal wave height of 1:4 m and a wave period of 6:5 s.
452 M. Barooni et al. / Energy 154 (2018) 442e454

Fig. 22. Validation of BA-Simula against experimental data using a scaled model of the 5 MW NREL wind turbine.

Fig. 23. 3D motion trajectory.

Fig. 25. Surge motion response spectrum.

and 0.5 m for surge, sway and heave, respectively, and 1% damping
for all frequencies, Table 2 presents the natural damped frequencies
loads on the system response. As it can be seen, different wave load
and periods.
cases do not have a significant impact on system response. For
In the second case study, we assumed a steady wind of 11.3 m=s,
these load cases, the dominant response frequency does not
and three wave profiles as presented in Table 3. This is to study the
change, and only a slight variation in response amplitude occurs.
dynamic response of the system for varying wave condition.
Frequency spectrum of roll motion for these load cases is illus-
trated in Fig. 27 to investigate in details the effect of hydrodynamic

Fig. 26. Surge excitation frequency spectrum.

Fig. 24. Poincare map.
 M. Barooni et al. / Energy 154 (2018) 442e454 453

Table 2 5. Conclusion
Natural damped frequencies and periods.

Motion Frequency (Hz) Period (s) This paper presented the mathematical formulation used to
Surge 0.0069 144.1871
develop an open-source numerical model for the analysis of
Sway 0.0069 144.1871 FOWTs. The model is verified and validated using several bench-
Heave 0.0314 31.8362 mark cases with the existing studies in the literature. The results of
Pitch 0.0224 44.5702 the fully coupled nonlinear model showed agreement with results
Roll 0.0224 44.5702
in the literature, and it can be concluded that the model gives
Yaw 0.1140 8.7737
reasonable results.
Several simulation studies were also performed to understand
the dynamics of the integrated FOWT. It was observed that aero-
Table 3
Wave load cases. dynamic loads govern the system response compared to hydrody-
namic and mooring loads. This indicates that for the design of
Load cases Wave properties Wind velocity
FOWT, the rotor design plays a significant role to achieve the
 
H (m) T (s) ms= desired system's response.
The developed open-source coupled aero-hydro-elastic nu-
C1 1.4 6.5 11.4 merical model of this paper allows researchers worldwide to use
C2 2.44 8.1 11.4
C3 3.66 9.7 11.4
the code for the analysis and design of FOWTs. It also enables them
to modify and further develop the source-code to advance the field.
Since the methodology is characterized by a modular integration of
a several function blocks, with little modifications, it is extendable
Fig. 28 shows the effect of rotor gyroscopic moment on turbine for studying other FOWT support structures as well.
power generation. The simulation is performed using a regular However, several challenges still remain when it comes to the
wave with a wave height of 1:4 m, a wave period of 6:5 s, and a analysis and design of FOWTs. For future work, the development of
constant wind speed of 11:4 m=s. As the figure shows, the gyro- a fully-nonlinear mooring dynamic model is recommended. In
scopic moment increases the turbine power out, but this is a addition, improvement of BA-Simula's hydrodynamic module is
negligible amount of 0.1%. This slight increase in power generation required to consider the effect of nonlinear stochastic waves. To
of the turbine is due to the small increase in rotor acceleration, and investigate different operational conditions of the wind turbine,
it is not statistically of any significant importance. Therefore, it can developing a robust dynamic controller should also be considered.
be concluded that the gyroscopic motion does not alter the power If these challenges are resolved using continuous research and
output, and it is only important to consider for the loading and the development, FOWT technology can become an economical way of
motion of the structure. extracting energy from the vast offshore resources in deep waters.
This would enable generating most of the world's energy con-
sumption with a clean and sustainable source of energy.

References

[1] Trenberth KE, Dai A, Van Der Schrier G, Jones PD, Barichivich J, Briffa KR, et al.
Global warming and changes in drought. Nat Clim Change 2014;4(1):17e22.
[2] Wallace JM, Held IM, Thompson DW, Trenberth KE, Walsh JE. Global warming
and winter weather. Science 2014;343(6172):729e30.
[3] Peters GP, Andrew RM, Boden T, Canadell JG, Ciais P, Le Que re C, et al. The
challenge to keep global warming below 2 c. Nat Clim Change 2013;3(1):4e6.
[4] Sun X, Huang D, Wu G. The current state of offshore wind energy technology
development. Energy 2012;41(1):298e312.
[5] Benson CL, Magee CL. On improvement rates for renewable energy technol-
ogies: solar pv, wind turbines, capacitors, and batteries. Renew Energy
2014;68:745e51.
[6] Ashuri T, Zaaijer MB. Review of design concepts, methods and considerations
of offshore wind turbines. In: European offshore wind conference and exhi-
bition. Berlin, Germany: European Wind Energy Association; 2007. p. 1e10.
Fig. 27. Load cases Frequency spectrum of roll rotational mode. [7] Li J, Bhattacharya S, Huang AQ. A new nine-level active npc (anpc) converter
for grid connection of large wind turbines for distributed generation. IEEE
Trans Power Electron 2011;26(3):961e72.
[8] Ashuri T, Martins JR, Zaaijer MB, van Kuik GA, van Bussel GJ. Aeroservoelastic
design definition of a 20MW common research wind turbine model. Wind
Energy 2016;19(11):2071e87.
[9] Geyler M, Caselitz P. Robust multivariable pitch control design for load
reduction on large wind turbines. J Sol Energy Eng 2008;130(3).
[10] Ashuri T, Zaaijer M. Size effect on wind turbine blade's design drivers. In:
Europeain wind energy conference and exhibition. Brussels, Belgium: Euro-
pean Wind Energy Association; 2008. p. 1e6.
[11] Muskulus M, Schafhirt S. Design optimization of wind turbine support
structures-a review. J Ocean Wind Energy 2014;1(1):12e22.
[12] Haghi R, Ashuri T, van der Valk PL, Molenaar DP. Integrated multidisciplinary
constrained optimization of offshore support structures. J Phys Conf
2014;555(1).
[13] Muskulus M. The full-height lattice tower concept. Energy Procedia 2012;24:
371e7.
[14] Boukhezzar B, Siguerdidjane H. Nonlinear control of a variable-speed wind
turbine using a two-mass model. IEEE Trans Energy Convers 2011;26(1):
149e62.
Fig. 28. Turbine generated power with and without gyroscopic moment. [15] Bououden S, Chadli M, Filali S, El Hajjaji A. Fuzzy model based multivariable
454 M. Barooni et al. / Energy 154 (2018) 442e454

predictive control of a variable speed wind turbine: Lmi approach. Renew results regarding floating wind turbine modeling. In: 2010 European wind
Energy 2012;37(1):434e9. energy conference and exhibition; 2010. p. 1e10.
[16] Hameed Z, Ahn S, Cho Y. Practical aspects of a condition monitoring system [35] Hansen MO. Aerodynamics of wind turbines. Routledge; 2015.
for a wind turbine with emphasis on its design, system architecture, testing [36] Faltinsen O. Sea loads on ships and offshore structures, vol. 1. Cambridge
and installation. Renew Energy 2010;35(5):879e94. University Press; 1993.
[17] Maki K, Sbragio R, Vlahopoulos N. System design of a wind turbine using a [37] Yilmaz O, Incecik A. Hydrodynamic design of moored floating platforms. Mar
multi-level optimization approach. Renew Energy 2012;43:101e10. Struct 1996;9(5):545e75.
[18] Dhert T, Ashuri T, Martins JR. Aerodynamic shape optimization of wind tur- [38] Kalogiratou Z, Monovasilis T, Psihoyios G, Simos T. RungeeKutta type
bine blades using a Reynolds-averaged NaviereStokes model and an adjoint methods with special properties for the numerical integration of ordinary
method. Wind Energy 2017;20(5):909e26. differential equations. Phys Rep 2014;536(3):75e146.
[19] Castro-Santos L, Filgueira-Vizoso A, Carral-Couce L, Formoso JAF.  Economic [39] Jonkman JM. Dynamics modeling and loads analysis of an offshore floating
feasibility of floating offshore wind farms. Energy 2016;112:868e82. wind turbine. ProQuest; 2007.
[20] Schwanitz VJ, Wierling A. Offshore wind investmentserealism about cost [40] Ng C, Ran L. Offshore wind farms: technologies, design and operation.
developments is necessary. Energy 2016;106:170e81. Woodhead Publishing; 2016.
[21] Ashuri T, Zaaijer MB, Martins JR, van Bussel GJ, van Kuik GA. Multidisciplinary [41] Manwell JF, McGowan JG, Rogers AL. Wind energy explained: theory, design
design optimization of offshore wind turbines for minimum levelized cost of and application. John Wiley & Sons; 2010.
energy. Renew Energy 2014;68(0):893e905. [42] Savenije L, Ashuri T, Van Bussel G, Staerdahl J. Dynamic modeling of a spar-
[22] Nielsen FG, Hanson TD, Skaare B. Integrated dynamic analysis of floating type floating offshore wind turbine. In: European wind energy conference &
offshore wind turbines. In: 25th international conference on offshore me- exhibition. Warsaw, Poland: The European Wind Energy Association; 2010.
chanics and arctic engineering. American Society of Mechanical Engineers; p. 1e10.
2006. p. 671e9. [43] Beltran B, Benbouzid MEH, Ahmed-Ali T. Second-order sliding mode control of
[23] Karimirad M, Gao Z, Moan T. Dynamic motion analysis of catenary moored a doubly fed induction generator driven wind turbine. IEEE Trans Energy
spar wind turbine in extreme environmental condition. In: Proceedings of the Convers 2012;27(2):261e9.
European offshore wind conference; 2009. p. 1e10. [44] Meng W, Yang Q, Sun Y. Guaranteed performance control of dfig variable-
[24] Matha D, Schlipf M, Pereira R, Jonkman J, et al. Challenges in simulation of speed wind turbines. IEEE Trans Contr Syst Technol 2016;24(6):2215e23.
aerodynamics, hydrodynamics, and mooring-line dynamics of floating [45] Karman TV. The fundamentals of the statistical theory of turbulence.
offshore wind turbines. In: The twenty-first international offshore and polar J Aeronaut Sci 1937;4(4):131e8.
engineering conference. International Society of Offshore and Polar Engineers; [46] Bierman J, Kincanon E. Reconsidering archimedes' principle. Phys Teach
2011. p. 1e8. 2003;41(6):340e4.
[25] Jonkman JM. Dynamics of offshore floating wind turbines model development [47] Johnson FT. A general panel method for the analysis and design of arbitrary
and verification. Wind Energy 2009;12(5):459e92. configuration in incompressible flows, vol. 3079. National Aeronautics and
[26] Sclavounos P, Lee S, DiPietro J, Potenza G, Caramuscio P, De Michele G. Space Adminstration, Scientific and Technical Information Office; 1980.
Floating offshore wind turbines: tension leg platform and taught leg buoy [48] Van Der Meulen MB, Ashuri T, Van Bussel GJ, Molenaar DP. Influence of
concepts supporting 3-5 mw wind turbines. In: European wind energy con- nonlinear irregular waves on the fatigue loads of an offshore wind turbine. In:
ference EWEC; 2010. p. 20e3. The science of making torque from wind. Oldenburg, Germany: Citeseer;
[27] Karimirad M, Moan T. Wave-and wind-induced dynamic response of a spar- 2012. p. 1e10.
type offshore wind turbine. J Waterw Port, Coast Ocean Eng 2011;138(1): [49] Sinaiski EG. Hydromechanics: theory and fundamentals. John Wiley & Sons;
9e20. 2011.
[28] Skaare B, Nielsen FG, Hanson TD, Yttervik R, Havmøller O, Rekdal A. Analysis [50] Holthuijsen LH. Waves in oceanic and coastal waters. Cambridge University
of measurements and simulations from the hywind demo floating wind tur- Press; 2010.
bine. Wind Energy 2015;18(6):1105e22. [51] Balachandran B, Magrab EB. Vibrations. Cengage Learning; 2008.
[29] Wang K, Moan T, Hansen MO. Stochastic dynamic response analysis of a [52] Abbasbandy S. Improving Newtoneraphson method for nonlinear equations
floating vertical-axis wind turbine with a semi-submersible floater. Wind by modified adomian decomposition method. Appl Math Comput
Energy 2016;19(10):1853e70. 2003;145(2):887e93.
[30] Jd Vaal, Hansen M, Moan T. Effect of wind turbine surge motion on rotor [53] Baruh H. Analytical dynamics. WCB/McGraw-Hill Boston; 1999.
thrust and induced velocity. Wind Energy 2014;17(1):105e21. [54] Marten D, Wendler J, Pechlivanoglou G, Nayeri C, Paschereit C. Qblade: an
[31] Vorpahl F, Strobel M, Jonkman JM, Larsen TJ, Passon P, Nichols J. Verification of open source tool for design and simulation of horizontal and vertical axis
aero-elastic offshore wind turbine design codes under iea wind task xxiii. wind turbines. Int J Emerg Technol Adv Eng (IJETAE) 2013;3:264e9.
Wind Energy 2014;17(4):519e47. [55] Deperrois A. Xflr5 analysis of foils and wings operating at low Reynolds
[32] Utsunomiya T, Sato T, Matsukuma H, Yago K. Experimental validation for numbers. 2009.
motion of a spar-type floating offshore wind turbine using 1/22.5 scale model. [56] Shin H, et al. Model test of the oc3-hywind floating offshore wind turbine. In:
In: ASME 2009 28th international conference on ocean, offshore and arctic The twenty-first international offshore and polar engineering conference.
engineering. American Society of Mechanical Engineers; 2009. p. 951e9. International Society of Offshore and Polar Engineers; 2011. p. 1e6.
[33] Larsen TJ, Hanson TD. A method to avoid negative damped low frequent tower [57] Gao Z, Moan T, Wan L, Michailides C. Comparative numerical and experi-
vibrations for a floating, pitch controlled wind turbine. J Phys Conf mental study of two combined wind and wave energy concepts. J Ocean Eng
2007;75(1). 01207e3. Sci 2016;1(1):36e51.
[34] Jonkman J, Larsen TJ, Hansen AM, Nygaard T, Maus K, Karimirad M, et al. [58] Dukkipati VR, Srinivas J. Mechanical vibrations. PHI Learning Pvt. Ltd.; 2012.
Offshore code comparison collaboration within iea wind task 23: phase iv