Ocean Engineering 257 (2022) 111320

Contents lists available at ScienceDirect

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

Fully nonlinear dynamics of floating solar platform with twin hull by

tubular floaters in ocean waves
Arash Abbasnia a, b, *, Madjid Karimirad a, Dallán Friel a, c, Trevor Whittaker d
a
School of Natural and Built Environment, Queen’s University Belfast, Belfast, UK
b
Centre for Advanced Sustainable Energy, Queen’s University Belfast, Belfast, UK
c
The Bryden Centre, Letterkenny Institute of Technology, Donegal, Ireland
d
Queen’s University Belfast, Belfast, UK

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

Keywords: Sustainability in harvesting solar power offshore depends on the survival of their platforms under environmental
Floating solar platform loads such as ocean waves. Therefore, a framework is established in this study to estimate the dynamics of
Boundary element method floating solar platforms under ocean waves experimentally and numerically. Hence the platform is taken twin
Stream function
hull with double circular cylinders which exposed to a broad range of waves in the calibrated physical wave tank.
Acceleration potential
Twin hull
On the other hand, the high order boundary element method and mixed Eulerian-Lagrangian approach are used
Floating photovoltaics in the numerical model to solve the fully nonlinear fluid field and the acceleration potential method is employed
to approximate the response of the platform implicitly. To validate the framework, numerical and experimental
results are compared for a single cylinder under the incident waves. Then, the comparison is extended for a twin
hull platform. The effect of nonlinear free surface flow amongst the twin cylinders on the dynamics of the
platform is studied parametrically. Afterwards, the performance of the present framework is assessed for a near
breaking incident wave and an irregular wave. The stream function theory and flux of superposed wave com­
ponents are used to emulate the high steep wave and the irregular wave respectively.

1. Introduction platforms therefore the determination of the exciting loads plays a vital
role in the design of a robust mooring system for preserving the sur­
The technology of Floating Photovoltaics (FPV) platforms has been vivability of the platforms. In this regard, a numerical simulation was
developed progressively in the recent decade. Even though land-based carried out by Mignone et al. (2021) to calculate the wind loads on the
photovoltaics (PV) farms have been installed widely as a global power floating solar panel using a viscous model. Moreover, the performance of
supply for several decades, offshore and nearshore floating solar plat­ land-based and offshore PV platforms has been studied under environ­
forms systems have been gradually installed to avoid using valuable mental conditions by Golroodbari and Sark (2020). Incident wave ef­
land, to address cooling challenges, and avoid the environmental impact fects, air humidity and wind speed were taken into account in their study
of land-based solar systems. An overview of various designs of floating and the induced tilted surface of the PV panels due to the motion
solar platforms developed in recent years is summarized in Friel et al. response of the floating platform under ocean wave conditions were
(2019). Moreover, a comprehensive review of basic design aspects for emulated using a linear potential model. Nonlinear hydrodynamic
the floating solar platforms such as environmental loads, mooring sys­ analysis was conducted by Friel et al. (2021) to calculate the responses
tems, environmental compatibility is outlined by Rosa-Clot and Giu­ of a moored FPV platform under regular waves. They used cylindrical
seppe (2020). The environmental impacts and the implication of the floaters to supply the buoyancy for the platforms as well as a mooring
floating solar systems in the marine environment were investigated by system based on elastic catenary equations to approximate the natural
Hooper et al. (2021). The sustainability of harvesting solar power by frequency of the dynamic system in the free decay test. Additionally, a
using floating solar platforms substantially depends on the survivability spectral analysis was performed upon the dynamic response of the
of the platforms subjected to environmental loads. Wind, current and platform to analyse the nonlinearities of the response amplitude oper­
wave are identified as sources of environmental loads on the floating ators under different wave load cases. Chasing the complexity of free

* Corresponding author. Centre for Marine Technology and Ocean Engineering (CENTEC), Instituto Superior Técnico, Universidade de Lisboa, Lisbon, Portugal.
E-mail address: [email protected] (A. Abbasnia).

https://doi.org/10.1016/j.oceaneng.2022.111320
Received 9 September 2021; Received in revised form 16 January 2022; Accepted 10 April 2022
Available online 11 June 2022
0029-8018/© 2022 Elsevier Ltd. All rights reserved.
A. Abbasnia et al. Ocean Engineering 257 (2022) 111320

Fig. 1. Scheme of the numerical wave flume for a FPV structure with double tubular floaters.

surface elevation around the floaters, which impacts the response of the boundary integral equation significantly effects the accuracy of the so­
platform particularly in short-crested waves, needs a fully nonlinear lution. For instance, curvilinear basis functions such as the B-spline and
simulation in the time domain to reveal the resonant free surface fluid Non-Uniform Rational B-Spline (NURBS) functions have been imple­
particle motion. Indeed, this matter becomes critical once the air is mented into the boundary integral equation by Abbasnia et al. (2017),
trapped as this has a greater influence on the response of these struc­ Datta and Sen (2006) and Gao and Zou (2008) to describe the
tures. Therefore, fully nonlinear free surface boundary conditions need complexity of the boundaries.
to be satisfied in the boundary value problem to take account of the In the present study, nonlinear loads on the floating solar platform
exact free surface oscillation and in the computation of the hydrody­ are estimated experimentally and numerically. The submerged surface
namic pressure. On the other hand, the instantaneous body boundary of the platform consists of two tubular floaters embedded apart at a
condition, particularly for non-wall side floaters such as cylinders, needs distance. Thus, the platform is considered a twin hull body in the wave-
to be accommodated in the solution of the boundary value problems. To body interaction. The nonlinear free surface flow between the tubular
meet these prerequisites, fully nonlinear potential Numerical Wave floaters might affect the dynamics of the platform under the ocean wave,
Tanks (NWT) have been developed in the past two decades to solve the which is dependent on the wave characteristics. Therefore, a wide range
fluid flow equations and the dynamic equations of floating platforms. of incident waves is chosen to interact with the platform in the physical
NWTs haves been used extensively for different applications of ocean wave tank. The range of wave characteristics extends from second order
engineering and marine renewable energy (Abbasnia and Guedes Stokes wave as a weak nonlinear wave to the near breaking wave as a
Soares, 2018, 2019; Ning et al., 2016, 2017). Artificial wave generators strong nonlinear wave. On the other hand, a fully nonlinear wave flume
and damping zones make NWTs capable of simulating the response of is established to simulate the floating solar platform with a twin hull
floatation under a broad range of wave steepness. Implementing the floater interacting with the incident wave. According to the prior
relaxation zone of the wave generator and damping zones adjacent to studies, the application of the fully nonlinear wave flume in the wave
the impermeable walls over the exact free surface triggers an open-sea interaction with twin hull bodies is a novelty in the present paper. The
condition during long-term numerical simulation. A summary of the potential theory is applied to solve the fluid flow and the MEL method
various wave generators and damping zones can be found in Tanizawa based on the material node approach is implemented to approximate the
(2000). The Mixed Eulerian-Lagrangian (MEL) approach coupled with a kinematics of the fluid particles on the exact free surface along with time
high-order time integration scheme has been employed for time marching. The fourth order Runge-Kutta time integration scheme is
marching the fluid particle on the free surface and updating the fully employed to update the computational domain. An artificial wave
nonlinear free surface boundary conditions for the next time step. The generator is adopted on the inlet boundary which is fused to a relaxation
MEL method has been implemented using the material node approach zone to preserve the open sea condition of the incident wave. To avoid
discussed in (Abbasnia and Guedes Soares, 2020) and the downstream reflections an artificial damping zone is stretched over the
semi-Lagrangian scheme (Kim and Koo, 2019). To solve the equation of free surface adjacent to the outlet boundary. Indeed, to retain the open
body motions, various methodologies have been proposed to calculate sea condition throughout the long-term simulation, transmitted incident
the instantaneous hydrodynamic pressure on the exact wet surface. An waves over the FPV structures as well as radiated waves due to dynamics
acceleration potential formulation (Vinje and Brevig, 1981) based upon of the structures toward the downstream, need to be diminished grad­
the modal decomposition method can obtain the time derivative of the ually along with the damping zone. The platform dynamics due to the
velocity potential field on the instantaneous wet surface implicitly; for excitation loads of the incident waves are computed using the acceler­
instance Coslovich et al. (2021); Ganesan and Sen (2019); Kim and Koo ation potential. To couple the solution of the velocity potential field with
(2019). The implicit body boundary condition algorithm was imple­ the dynamic equation of the platform, the modal decomposition scheme
mented by Abbasnia and Guedes Soares (2017) to calculate the re­ is implemented into the time marching scheme to calculate the kine­
sponses of a ship-shaped freely floating body under strong nonlinear matics of the twin hull platform and simultaneously to update the wet
incident waves. Since the solution of the velocity potential field is surface of the structure. At each time step re-gridding of the computa­
required to solve the acceleration potential, the Boundary Element tional boundary is performed over the updated moving boundary,
Method (BEM) has been employed widely to solve the potential flow on therefore the NURBS basis function, as a higher-order interpolation
the wet surface (Kim et al., 2021; Tang et al., 2021). The order of basic function, is implemented into the numerical algorithm. Furthermore, to
functions drastically influences the accuracy of the boundary integral define the distribution of boundary value accurately and determine the
equations. In other words, defining the distribution of boundary values spatial derivatives of the BEM solution the advantage of the NURBS basis
as well as calculating the geometry characteristics of the boundary in the function is taken. The free decay test of a single cylinder is simulated to

2
A. Abbasnia et al. Ocean Engineering 257 (2022) 111320

system O x z is assigned at the centre of rotation of the structure, which

′ ′ ′

moves with the body in translational directions. It is assumed that the

fluid is homogenous, incompressible, and inviscid and the flow is irro­
tational then the potential theory properly describes the flow field.
Accordingly, the Laplace equation of the velocity potential function
φ(x, z, t) formulated as

∇2 φ(x, z, t) = 0 (1)
needs to be solved over the fluid volume Ω(t) and the solution sat­
isfies the boundary conditions. The normal vector is directed out of the
Fig. 2. Schematic scheme of the hydrodynamic coordinate system allocated on fluid and the boundary conditions are designated as
the body. ∂φ ∂φwave ​
=− ​ on Γ1
∂n ∂x
perform the convergence test and accuracy test. Sets of experimental ∂φ
data of wave interaction with a single cylinder as well as the platform = 0 on Γ2 , Γ3 (2)
∂n
are compared with the numerical results to verify the present frame­
∂φ → →
work. A parametric study is conducted to find out the effect of nonlinear = V body ⋅ n on Γ4
∂n
flow between the floaters on the dynamics of the platform exposed to a
wide range of nonlinear waves. To demonstrate the capability of the φwave is an arbitrary theoretical velocity potential of the incident
model for simulating the nonlinear wave-body interaction, a steep wave waves corresponding to a broad range of the wave’s nonlinearities. Γ4
is chosen to interact with the solar platform. Hence the boundary con­ represents the wet surface of the tubular floaters and
dition on the wavemaker is manipulated to emulate the high steep → →
V body = V 0 + ϖ × →
r (3)
incident wave using the stream function theory. It is taken a state-of-the-
art contribution in developing potential numerical wave tanks. Then the →
in which V 0 is the respective translational velocity of the structure to
numerical results and the experimental data are compared. It is aimed to Oxz, ϖ is its angular velocity around the centre of rotation. The centre of
manifest the validity of the present framework for considering the
gravity is taken as the centre of rotation in this study and → r is the
complex hydrodynamic phenomena in the interaction of nonlinear
perpendicular distance of the arbitrary computational node (P) on the
ocean waves and the floating solar platform into the numerical simu­
body to the centre of rotation as shown in Fig. 2.
lation. In addition, the feasibility of the present model for the interaction
Fully nonlinear kinematic and dynamic free surface boundary con­
of the platform with the irregular wave is illustrated.
ditions that are assigned to the exact free surface, are given as
⎧
2. Mathematical formulation ⎪
⎪ ∂η ∂φ ∂φ ∂η
⎨ ∂t = ∂z − ∂x ∂x
⎪
​ on ​ Γ5 (4)
2.1. Governing equation and boundary conditions ⎪
⃒
⃒ 2
⎪ ∂φ = − gz − ∇φ|
⎪
⎩
∂t 2
The scheme of fluid control volume of fluid is illustrated in Fig. 1. A
fixed Cartesian coordinate system Oxz is allocated on the mean water where g is the gravitational acceleration and η is the wave elevation. The
level where z axis is oriented upward and a hydrodynamic coordinates relative air pressure is taken zero on the exact free surface within the gap
of floaters. It is assumed the air is homogenous, incompressible, and
inviscid and then the spring-like behaviour of the air is neglected. Initial
conditions to obtain a unique solution of the boundary value problem is
assumed as
{
φ(x, 0, t ≤ 0) = 0
(5)
η(x, t ≤ 0) = 0
Boundary integral equation is employed based on the second Green’s
identity to solve the boundary value problem as
∫ ∫
∂G(q, P) ∂φ(P)
c(q)φ(q) + φ(P) ds = G(q, P) ds (6)
Γ ∂n Γ ∂n
Fig. 3. Scheme of a corner point.

Fig. 4. Scheme of the experimental setups.

3
A. Abbasnia et al. Ocean Engineering 257 (2022) 111320

where q represents the source and P demonstrates computational point Table 1

Characteristics of the incident waves.
on the boundary Γ(≡ Γ 1 ∪Γ2 ∪Γ3 ∪Γ4 ∪Γ5 ). The Green’s function is
given as Sea state (
ω 2π/
) (
k 2π/
) H (m) H/
λ
T λ
( ) ( )
1 1 1 3.141 1.311 0.056 0.011
G P, q = ln (7)
2π R(q, P) 2 4.724 2.389 0.063 0.024
3 6.283 4.002 0.049 0.031
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ 4 7.853 6.160 0.055 0.053
(xq − xP )2 + (zq − zP )2 and c(q) = θq
/
where R(q, P) = in which θq is
2π
the internal angle of the source point.
⃒ 2
∂φ ⃒∇φ|
Φ〈= − −p − gz〉 = + (12)
2.2. Time marching of free surface ∂t 2
which represents the acceleration fluid particles by −a = ∇φ.
To satisfy free surface boundary conditions, the Eulerian form of Eq.
Nevertheless, the acceleration potential function is not satisfied in the
(4) is transformed into the Lagrangian manner using the total derivative
( →) Laplace equation due to the presence of the second term on the right-
formulation δ/δt = ∂/∂t + → v . ∇ . In the material node approach →
v = hand side of Eq. (12). Whereas the first term of the right-hand side
→ satisfied in the Laplace equation (∇2 φt = 0), is computed through the
∇ φ that means the computational nodes on the exact free surface are
allowed to move with the fluid particle. Therefore, the Lagrangian form modal decomposition method. Hence the time derivative of the velocity
of the free surface boundary conditions is written as potential is decomposed into four modes as,
⎧ ̅→ ∂φ ∑3
⎪
⎪
⎪ δL → = ai φi + φ4 (13)
⎨ = ∇φ ∂t i=1
δt
⃒ ​ on ​ Γ 5 (8)
⎪
⎪ δφ 1⃒ where ai is the ith mode component of the body acceleration. i = 1 and 2
⎪
⎩ = − gη + ⃒⃒∇φ|2
δt 2 are marking the translational acceleration in the x-direction and the z-
direction respectively and i = 3 is the rotational acceleration. φ4 is
̅→
where L = (xP , zP ) represents the coordinates of the point P on the corresponds to the body boundary condition for the time derivative of
exact free surface. According to Fig. 1, a relaxation zone and a damping the velocity potential as
zone are embedded over the free surface in the vicinity of the inflow {
∂φ N , i = 1, …, 3
boundary (wave generator) and the outflow boundary respectively. To = − i ​ on ​ Γ4 (14)
∂n q, i = 4
address their functionalities, artificial damping terms (Cointe, 1990) are
added to the fully nonlinear free surface boundary condition as ̅→
where N = (→ n ,→
r ×→ n ) and
⎧ ̅→
⎪ δL ( ̅→ ̅→ ) ( )2 ( )
⎪
⎪ →
⎨ = ∇ φ − μ(x) L − L e → →
δt −q = − K →∇ φ − V body + → n ⋅→ →× →
ϖ × (ϖ r )+→ →× →
n ⋅ 2ϖ ∇ φ − V body
⃒ ​ on ​ Γ 5 (9)
⎪
⎪ δφ 1⃒
⎪ = − gη + ⃒⃒∇φ|2 − μ(x)(φ − φe ) → →
⎩
δt 2 ∂ ( ∇ φ⋅ ∇ φ)
−
∂n 2
where subscript e is marking the reference values and (15)
( )2 the last term of Eq. (15) is found as
k(x − x0 )
μ(x) = αω , x0 ≤ x ≤ x0 + βλ (10)
2π
in which k and ω shows respectively the wavenumber and the
angular frequency of the incident wave. λ is the incident wavelength that
correlates to the wave period (T) through the dispersion relation. α and β
are used to tune up the strength of the damping function and its effective
length respectively. The designation of the reference values allows Eq.
̅→
(9) to act either as a relaxation zone for L e = (x, ηwave ), φe = φwave or as
̅→
a damping zone when L e = (x, 0), φe = 0.

2.3. Acceleration potential

To determine the kinematics of the floating structure instanta­

neously, Bernoulli’s pressure (−p) needs to be computed over the wet
surface of the platform at each time step, that formulate as
⃒
∂φ ⃒∇φ|2
−p = − − gz − (11)
∂t 2
Therefore, the time derivative of the velocity potential and the ve­
locity of fluid flow over the wet surface are required which are obtained
from the solution of the equation of the fluid motion. Although the ve­
locity profile of the fluid flow is obtained from the solution of the ve­
locity potential field, an implicit methodology is required to calculate
the time derivative of the velocity potential. Therefore, the acceleration Fig. 5. Dimensionless underdamped period of free decay test for a single cyl­
potential is defined as inder with different time step sizes and mesh sizes.

4
A. Abbasnia et al. Ocean Engineering 257 (2022) 111320

Fig. 6. Comparing the time history of the free decay test for the single cylinder.

Eventually, φi (i = 1, ..., 4) is calculated using the boundary integral

Table 2 equation (Eq. (6)) for each mode and Eq. (11) can be written para­
Runtime analysis for the decay ratio ε = 10E − 2. metrically as
Runtime per iteration (s) ( ⃒ ⃒2 )
1⃒ ⃒
Nodal Tn 30
/
Tn 40
/
Tn 60
/ −p = − a1 φ1 + a2 φ2 + a3 φ3 + φ4 + gz + ⃒⃒∇φ⃒⃒ (20)
2
Resolution
̅→
40 1.091 1.037 0.91 On the other hand, the hydrodynamic axial forces ( F ) and the
60 1.64 1.87 1.81 ̅→
hydrodynamic moment ( M ) is obtained as
80 2.97 3.09 3.11
∫
120 6.23 6.36 6.66 ⎧ ̅→
⎪
⎨ F 〈 = M(−a1 , −
a 2 + g)〉 = −p→n ds
(21)
Γ4
∫
→ → ⃒ ⃒ ( ) ⎩ ̅→
⎪
−p(→r ×→
∂ ( ∇ φ⋅ ∇ φ) ⃒ ⃒2 ∂φ ∂2 φ ∂φ ∂ ∂φ M 〈 = Mℜ2g −a3 〉 = n )ds
= − K ⃒⃒∇φ⃒⃒ − + (16) Γ4
∂n 2 ∂n ∂s2 ∂s ∂n ∂s
where M is the mass and ℜg is the gyration radius of the body relative to
in which K is the normal curvature of the body surface and → s is the
the centre of rotation. To find the generalized body acceleration, the
unit tangential vector on the body. The free surface boundary conditions
motion equation (Eq. (21)) is converted to a linear system of equations
for each mode are defined as
⎧ that is simply solved.
⎨ 0, i = 1, …3
⎪
⃒
φi = ⃒∇φ|2 ​ on ​ Γ 5 (17) 3. Numerical implementation
⎪
⎩ − gz, i = 4
2
3.1. Gridding and boundary integral discretization
The boundary condition on the inflow boundary is identified as
⎧
⎪ 0, i = 1, ..., 3 A proper distribution function of nodal points is needed to define the
∂φ ⎨ ( ) variation of the potential flux of an arbitrary wave theory along the
= ∂ ∂φwave ​ on ​ Γ 1 (18)
∂n i ⎪⎩ − , i=4 wavemaker. Therefore, a fifth-order polynomial distribution function
∂t ∂x (Abbasnia et al., 2021) is devoted to enriching the resolution of nodal
and points close to the free surface where the variation of the wave flux is
robust. To discretize Eq. (6), the NURBS basis function is inserted into
∂φ the boundary integral to define the boundary geometry and simulta­
= 0, i = 1, …4 ​ on ​ Γ 1 , Γ2 (19)
∂n i neously the distribution of the boundary value. The generic formulation

Fig. 7. Snapshots of the free surface evolution for the free decay test of the single cylinder.

5
A. Abbasnia et al. Ocean Engineering 257 (2022) 111320

Fig. 8. The comparison of the heave motion for a single cylinder.

of the basis function is given as points. wj is the weighted function corresponding to the th
j control point.
∑ m ς ̅→ The B-spline basis function Nςj (u) is the ς-order non-uniform interpola­
̅→ j=1 Nj (u) Z j wj
L 〈 = (x(u), z(u))〉 = ∑ (22) tion function (Piegl and Tiller, 1995). Hence Eq. (6) is written as
m ς
j=1 Nj (u)wj
∑m [ ∂
(
̅→ ̅→
) ] ∑ [ (
m ̅→ ̅→ ∂φ
) ]
c φ σ
i i i,l +φ G L i − L l J w
l l = G L i − L l J w
̅→
where u is a local axis that varies over the interval [0, 1], Z j represents
l=1 l
∂nl l=1 ∂nl l l
th
the coordinates of the j control point and m is the number of control (23)

6
A. Abbasnia et al. Ocean Engineering 257 (2022) 111320

Fig. 9. The comparison of the axial force for a single cylinder.

where i = 1, ..., m and m represents the number of distributed nodal and wl is the integrating weight corresponding to the nodal point.
The integrating weight is the function of the type of nodes’ distribution
points over the computational boundary (Γ). Jl is the Jacobian of the lth
along with the local axis u. σ i,l represents the Dirac delta function. Hence
nodal point which is given as
the Dirichlet boundary condition predominates on the free surface and
⃒ ⃒
⃒ ̅→ ⃒
⃒D L l ⃒ the Neumann boundary condition on the rest of the boundaries, a system
Jl = ⃒⃒ ⃒
⃒ (24) of equations are yielded by sorting the unknown boundary value on the
⃒ Du ⃒ left-hand side of Eq. (23) (Abbasnia and Ghiasi, 2014). By solving the

7
A. Abbasnia et al. Ocean Engineering 257 (2022) 111320

Fig. 10. Snapshot of free surface evolution in the vicinity of the platform for all sea states at (l/ = 12.5).
T

Fig. 11. Snapshots of the free surface elevation within the relaxation zone and the damping zone for the interaction of the platform with the sea state 2.

system of equations, the velocity potential and its normal derivative surface boundary in the Lagrangian manner and the motion of the body
over the boundaries are given to manipulate into the time marching instantaneously.
scheme for updating the control volume. Therefore, Eq. (8) is devoted to
obtaining the time derivative of the boundary conditions on the exact
free surface and simultaneously the acceleration potential field is solved 3.2. Treatment of corner point
to calculate the kinematics of the floating body. By taking advantage of
the NURBS derivative, Cartesian components of the fluid particle on the Discontinuity of the normal vector at the intersection of the free
instantaneous free surface, as well as the first order and the second order surface and the rigid walls produces a singularity for computing the fluid
of the tangential derivatives in Eq. (16), is computed accurately. The particle velocity which is called the corner point problem. Since the
fourth order Runge-Kutta time integration is adopted to update the free normal derivative of the velocity potential is achieved through the so­
lution of the velocity potential field, two different values of the potential

8
A. Abbasnia et al. Ocean Engineering 257 (2022) 111320

Fig. 12. The comparison of the heave motion for the FPV structure.

9
A. Abbasnia et al. Ocean Engineering 257 (2022) 111320

Fig. 13. The comparison of the pitch motion for the FPV structure.

10
A. Abbasnia et al. Ocean Engineering 257 (2022) 111320

Fig. 14. The comparison of the exciting axial force for the FPV structure.

11
A. Abbasnia et al. Ocean Engineering 257 (2022) 111320

normal derivative are concluded at the corner point as shown in Fig. 3.

Since the fluid particle velocity at the corner point has a unique
value, the component of the particle velocity is calculated as
⎧
⎨→ ν ⋅→
n Body = φn Body
→ (25)
⎩ ν ⋅→n Free ​ surface = φnFree ​ surface

4. Experimental setup

Experiments were conducted in a wave flume at the Hydraulics

Laboratory, Queen’s University Belfast which is 4.6 m wide, approxi­
mately 18 m long and has a maximum operating depth of 0.8 m. The
experimental models consisted of two horizontal semi-immersed cylin­
der type platforms which represent a simplified version of the gable
slender concept being developed by UpSolar and SolarMarine Energy.
The models tested were a representation of the leading floats in a very
large array which would extend several wavelengths in the direction of
wave travel. Consequently, the rear tube was restrained in surge and the
horizontal reaction force was measured with load cells. In Fig. 4 the
single cylinder and twin cylinder models are shown, and the models are
scaled using Froude scaling laws with a geometric scaling ratio of 1:4.5.
The models were instrumented with two in-line axial load cells and a
Qualisys motion capture system to measure the surge loading and dy­
namic motions of the structures. An Edinburgh Designs Ltd wavemaker
system was used to generate sea states which were characteristic of a
sheltered nearshore location. A range of regular monochromatic sea
states was calibrated using 7 Edinburgh Designs Ltd resistance-based
wave probes. The models were tested in a series of wave characteris­
tics and the data was collected. Furthermore, visual recordings were
documented alongside documentation of the observed fluid-structure
interactions.
The records of the physical model testing were used to compare with
the numerical results and to assess the present framework.

5. Applications and results

To approach a reliable framework for studying the floating solar

platform with tubular floaters, a numerical model was developed to
simulate the time domain dynamics of the floating solar platform. Hence
a numerical wave flume which was developed in Abbasnia and Guedes
Soares, (2018), was enhanced to model twin hull bodies. On the other
hand, a set of experimental data was carried out for single cylinder and
double cylinders under a broad range of wave frequencies. Therefore,
the objective was to obtain a comprehensive comparison to verify the
reliability of the hydrodynamic analysis framework developed. Pro­
gramming the numerical model is done in FORTRAN and the specifi­
cation of the computer machine includes a PC laptop with Intel(R) Core
i7-10850H CPU @ 2.70 GHz and 16.0 GB RAM. The water depth is d =
0.765 m and the incident wave characteristics are tabulated in Table 1.
The diameter of the cylinder is 0.11 m, the length of the cylinder is to
4.58 m and the draft is set to r in the calm water.

5.1. Convergence test and accuracy of the numerical model

Fig. 15. Spectral analysis of axial forces on the FPV structure.

The free decay test of a single cylinder was chosen to evaluate the
Table 3 convergence test of the numerical model developed for different mesh
RMS error of numerical results against the experiment. sizes on the free surface and time step sizes. Since the numerical model
Sea state Heave Pitch Axial force
takes advantage of the BEM and MEL to solve the fluid flow, the mesh
size affects the numerical stability of the solution due to the numerical
Single Double Single Double Single Double
integration of Green’s function in the boundary integral equations. On
1 1.8% 1.1% – 1.8% 2.4% 4.2% the other hand, updating the boundary conditions at each time step is
2 1.3% 5.5% 2.4% 2.2% 3.2%
–
substantially dependent on the time step size during the time marching
3 1.1% 4.3% – 1.6% 3.6% 4.3%
4 1.2% 1.4% – 6.2% 2.3% 4.1% process. Therefore, the numerical stability of the solution needs to be
investigated to find the valid range of the mesh size as well as the time
step size. Hence, four numbers of nodal points are considered to simulate

12
A. Abbasnia et al. Ocean Engineering 257 (2022) 111320

Fig. 16. Parametric study on the dynamics of the platform.

hand, the convergence of the model is not maintained with the lowest
nodal resolution for the entire range of time step size. The time history of
the heave motion of the cylinder for the free decay test is compared with
the experimental data in Fig. 6. The time step size Tn 40 and the nodal
/

resolution 60 are used in the simulation and the origin of the hydro­
dynamic coordinate system is positioned at the centre of cylinder. The
comparison shows a good consistency between the numerical and
experimental data. The decay ratio (ε) is given as
Zt
ε= (26)
Z0

where Zt represent the heave amplitude and Z0 is the maximum heave

amplitude to demonstrate the viscous effect which mainly accounts for
the minor inconsistency between the numerical and experimental data.
As expected, the asymptotic envelope of the diminishing heave
amplitude in the experimental result is more convex than the numerical
result. The free decay period looks quite close which approaches to 0.54
s in both results. Therefore, it is concluded that the pressure damping
came from the wave radiation driven by heave motion, which pre­
dominates in the free decay test rather than the viscous damping.
Fig. 17. Parametric study on the pitch motion of the platform.
To demonstrate the efficiency of the computer code’s algorithm on
the computer machine, the runtime of each iteration for different mesh
sizes and time step sizes are sorted in Table 2. The threshold of code
the free decay test as 40, 60, 80, and 120 that represent a broad range of
break is set ε = 10E − 2. There is a significant difference in the runtime
the nodal resolution. Also, three different time step sizes
( ) between the finer mesh size and the coarser mesh size which is raised by
Δt = Tn 30, Tn 40, Tn 60 are considered through the simulation where
increasing the size of the coefficient matrix of the boundary value
/ / /
√̅̅̅̅̅̅̅̅̅̅
Tn is the natural period Tn = π πr/g. The third order B-spline basis problems. Correspondingly, the solver of the system of equations takes a
function is chosen to define the free surface and the length of the tank is longer time to converge. On the other hand, decreasing the time step size
taken 10 m. The cylinder is placed in the middle of the tank and the does not reduce the efficiency of the code performance substantially,
nodal points are allocated uniform radially (π /36) over the wet surface which shows the time integration algorithm is performing properly
of the cylinder (Γ4 ). Linear elements are sufficiently accurate to define along the simulation.
the wavemaker, the end wall and the tank bottom. The mesh sizes d/ 30 Moreover, the performance of damping zones at the ends of the
and L/120 are used over the vertical walls (Γ 1 and Γ3 ) and the bottom numerical flume are demonstrated in Fig. 7 in which the progressing of
(Γ 2 ) respectively. Two damping zones are adopted at the ends of the the radiated waves through the damping zones are tracked along the
tank for diminishing the radiated waves. time. Diminishing the radiated waves along the damping zones testifies
The characteristic of the damping zone are α = 1, β = 2, ω = 2π/ Tn the good performance of the damping zones and the accuracy of the
and k = 4/π r. Hence the period of free decay test is subjected to chase model.
the convergence of the model for different mesh sizes and time step size In the following, the two-degree of freedom of the single cylinder’s
Fig. 5. The ratio of the free decay period T0 to the natural period is given motion which is confined in the surge direction (x direction), is simu­
( )
as the dimensionless free decay period T0/T . Accordingly, the model lated for the incident waves in Table 1 and the results are compared with
n
the experimental data in Fig. 8 and Fig. 9 that represent the time history
convergence is highest for time steps Δt = Tn 30 and Tn 40 when the
/ /
of the heave motions of the cylinder and the exciting axial force on the
nodal point resolution is above 40. For Δt = Tn 60 the convergence of
/
cylinder respectively.
the model is retained for the nodal resolution 60 and 80 whereas the
The time step size is a function of the incident wave period which is
highest nodal resolution does not maintain the model convergence
set to T/40. Moreover, the mesh size on the free surface is a function of
preserved. Indeed, the higher density of nodal points does not guarantee
the incident wavelength taken as λ/60 and the length of the wave tank is
the numerical stability of the boundary integral equation. On the other
taken as 8λ. The properties of the relaxation zone over the downstream

13
A. Abbasnia et al. Ocean Engineering 257 (2022) 111320

Fig. 18. The FPV structure motions in two-degree of freedom for a high steep wave.

Fig. 19. The axial force on the FPV structure for a high steep wave.

side of the wavemaker is taken α = 1, β = 1 and for the damping zone axial force is increasing for higher wave frequency which shows that the
adjacent to the end wall is α = 1, β = 2. The mesh size on the vertical axial force is driven mostly by the wave inertia and viscous effects do not
walls and the bottom is the same as before. Since the length of the nu­ make a significant difference in the axial force for lower wave fre­
merical tank and the experimental one is not the same, approaching the quencies.
periodic steady-state occurs at different ratios of t/T for each sea state. In addition, the two-degree of freedom motion of the FPV structure
Indeed, while the wave celerity is decreasing, the wave takes a longer with double cylinder floaters and Lcy = 0.965 is used to evaluate the
time to propagate over the longer tank. This issue is observed present framework for the simulation of the structural response under
throughout the comparisons between the numerical results and experi­ the incident waves shown in Table 1 and snapshots of the free surface in
mental data. The comparison of the heave motion over the range of the vicinity of the platform at the time-periodic steady state are depicted
incident wave frequency shows substantial compliance between the in Fig. 10. The snapshots demonstrate the accomplishment of the nu­
numerical calculations and experimental results. The highest difference merical model for simulating the fully nonlinear free surface. Moreover,
in the heave motion is for sea state 1 where the viscous effect on the fluid the good performances of the relaxation zone at the upstream and the
flow is higher than the other sea states. Indeed, the radiated wave damping zone at the downstream are shown in Fig. 11. Hence snapshots
induced by the cylinder movement is diminished while the viscous effect of the free surface at three sequence time steps are compared over the
is rising. On the other hand, the difference between the axial forces is relaxation zone and the damping zone for the sea state 2. It shows that
more noticeable than the heave motion results. The amplitude of the the open sea condition is preserved through the relaxation zone whereas

14
A. Abbasnia et al. Ocean Engineering 257 (2022) 111320

quite promising whereas the pitch motion error for sea state 4 is the
highest error which is triggered by the strong nonlinear flow of the wave
breaking between the floats. Accordingly, the nonlinear components of
the fluid flow are amplified which raise the RMS errors of the axial force
for the sea states.
To study the influence of the twin hull floater on the dynamics of the
platform, a parametric analysis is performed on the heave motion and
the axial force over the range of incident waves. Indeed, the ratio of
wavelength to Lcy is chosen as a variable that represents the nonlinearity
of wave interaction with the twin hull platform. Although it is expected
that nonlinearities are magnifying when the ratio L/Lcy becomes smaller,
the oscillation of the free surface between the cylinders may influence
the resultant dynamics of the platform. Oscillation of the free surface
between the cylinder is produced by superposing the transmitted inci­
dent wave and the radiation waves of floaters, which are reflected by
them during the simulation. Since the dynamics of a single cylinder was
obtained, the comparison between the results of the single cylinder and
the platform reveals the impact of the gap between the cylinders on the
Fig. 20. Response of the platform under an irregular wave. dynamics of the platform as shown in Fig. 16. Dimensionless heave
motion is defined as 2|Z|/H where |Z| represents the magnitude of the
the attenuation of wave proceeds well along the damping zone. heave motion whereas the magnitude of the axial force is non-
Accordingly, the time history of the heave motion, pitch motion (ψ ) and dimensioned by 0.5ρgHπr2 . Both comparisons show that the effect of
exerted axial force is compared with the experimental record in the gap between the cylinder is significant on the dynamics of the
Figs. 12–14. platform over the range 1.6 < L/Lcy < 2.5. As expected, the nonlinear
The comparison of the heave motion in Fig. 12 demonstrates that the dynamics of the platform are diminishing for L/Lcy > 4 and the effect of
difference at the wave troughs between the experimental data and nu­ the gap between the cylinders is decreasing on the heave motion as well
merical results is amplified at the periodic steady state when the heave as the axial force. On the other side for L/Lcy < 1.6, the gap effect on the
emotion amplitude is decreased. Indeed, the amplitude of the heave axial force is more considerable than on the heave motion. The magni­
motion is roughly halved in sea state 2 and sea state 3 compared to sea tude of the axial force is rising by almost 40% due to the gap effect.
state 1 and sea state 4. Accordingly, the discrepancy in the trough of the During the physical model testing, the strong nonlinear free surface
heave motion for sea state 2 and sea state 3 is substantially higher than flow was observed in the gap region and turbulent flow was obvious due
the heave motion for sea state 1 and sea state 4. On the other hand, the to wave breaking as well. Therefore, the dimensionless magnitude of
difference in the wave crests in the heave motion is lower than the pitch motion (2|ψ | /kH) is drawn against L/Lcy in Fig. 17. For L/Lcy ≥ 2.5
trough for all sea states. the variation of the pitch motion is linear whereas there is a substantial
Though the numerical and experimental results of the pitch motion decrement for L/Lcy ≤ 1.6 that may be triggered partly by the nonlinear
comply properly in Fig. 13 rather than the comparison of the heave
flow with the gap.
motion, the discrepancy of the pitch motion for the sea state 4 is more
remarkable than the other sea states. Indeed, as the same as the heave
motion, the highest deviation occurs for the small amplitude of the pitch 5.2. Application of the numerical model for robust nonlinear incident
motion. Comparing the amplitude of the axial wave force on the struc­ wave
tures, shown in Fig. 14 manifests a significant predominance of the
pressure drag force in the axial exciting force than the friction drag The developed numerical model in the present study takes advantage
force. Due to a greater mismatch between the axial forces for the sea of the high order interpolation function and the MEL approach to
states 2 and 3 than the other sea states, the spectral analysis is carried simulate the fully nonlinear wave body interaction in the time domain.
out on the time series of the axial forces in Fig. 15 for evaluating the Therefore, a strong nonlinear incident wave is chosen to interact with
capability of the present numerical model for the nonlinear simulation the FPV platform and to assess the stability of the numerical model.
of the wave-body interaction. For both sea states, 170 components are There is a limitation on the wave height of the incident wave that can be
taken into account and a comparison of the experimental data and the used as the numerical model becomes unstable and breaks down if the
numerical results shows that the numerical model approximates the float detaches from the free surface. Hence the wave properties are set to
nonlinear components of the axial forces successfully. The amplitude of Hk/2π = 3.8E − 2 which represents a high steep incident wave to
the nonlinear components in sea state 2 is stronger than the axial force interact with the FPV structures. The stream function theory is applied to
components for sea state 3. The main source of the nonlinearity ampli­ the wavemaker to emulate the wave steepness of the incident wave. The
fication is originated from sloshing over the confined free surface be­ response of the structure for two-degree freedom is compared with the
tween the floatation tubes. The wave transmission through the structure experimental data in Fig. 18. Moreover, the axial force on the structure is
is affected due to the gap length (Lcy ) between the floaters and their computed and compared with the experimental data in Fig. 19. The
geometrical characteristics. Furthermore, to quantify the deviation be­ mesh size and the time step size are set as before, and the properties of
tween the numerical solution and the experimental results, the Root the damping zone and the relaxation zone are specified as the same as
Mean Square (RMS) function given as before. The time series of the heave motion in Fig. 18 shows the sinkage
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
√[ / ]̅ of the structure in the experimental results whereas the amplitude of the
√ ∑No ( Numeric ​ i − Experiment ​ )2 heave motion in both the numerical and experimental time traces are
RMS = √ i
No (27)
i=1 Experiment ​ i close. On the other hand, there is good agreement between the time
series of the pitch motion for both experimental and numerical results.
is computed for heave, pitch and axial force of the single cylinder and The amplitude of the pitch motion is significantly larger than the results
the double cylinder platform over the sea states as shown in Table 3, for the low steepness waves of Table 1.
where No represents the number of data in a single period. There is a significant deviation between the axial force results in
For the single cylinder, the performance of the numerical model is Fig. 19 due to a strong nonlinear oscillation of the confined free surface

15
A. Abbasnia et al. Ocean Engineering 257 (2022) 111320

between the floats along with the occurrence of wave breaking. Subse­ motion, pitch motion, and axial forces corresponding to the single cyl­
quently, it amplifies the nonlinearities in the experimental results inder and the floating solar platform were tabulated for the sea states,
whereas the amplitudes of the axial force in both numerical and which is quite promising in overall. Although the present numerical
experimental results compare. model is capable of estimating the fully nonlinear free surface, the
To illustrate the feasibility of the present model for an irregular complex free surface flow within the cylinders is close to wave breaking
incident wave as a realistic sea state, the JONSWAP spectrum (Hassel­ which results in the turbulence flow effect magnifying in the physical
mann et al., 1973) is chosen with the significant wave height of 0.05m model test. To determine the impact of strong nonlinear flow over the
and the peak period 2s to interact with the platform. 100 mono­ gap between the cylinders of the FPV platform, a parametric study was
chromatic wave components are extracted from the spectrum and a conducted. Hence, comparisons of the heave motion and the axial force
random function is used to obtain the initial phases of the components. due to the range of incident waves were done between the results of the
Superposing the velocity potential functions of the components yields single cylinder and the platform. Results of the single cylinder represent
the resultant velocity potential that is specified on the wavemaker to the dynamics of two cylinders which the gap effect is omitted. For the
consider the effect of implicitly coupling wave components on the range of the ratio of wavelength to the gap distance between 1.6 and
response of the platform. The flux of the resultant velocity potential is 2.51, the gap effect was substantial which showed a significant decre­
designated as the boundary condition on the wavemaker and the spec­ ment of magnitudes in the heave motion and the surge force. Moreover,
tral analysis is conducted on the response of the platform as shown in the pitch motion showed a sharp decrement for the dimensionless
Fig. 20. wavelength less than 1.6, whereas the strong nonlinear flow in the
There is a comparison between the wave spectrum and the spectrum confined free surface between the cylinders was observed in the physical
of the platform responses. The nonlinearities of the axial force are modelling over this range as well. It is concluded that the gap effect
stronger than the heave and pitch motion whereas there is no amplifi­ might affect the pitch motion partially in this range of the ratio of
cation in high order nonlinearity of the heave and pitch motion. The dimensionless wavelength. Since the numerical model takes advantage
amplification of high order nonlinearity of axial force might be triggered of the high-order solver to model fully nonlinear free surface boundary
by the gap between the floaters which affects the free surface flow. conditions, a high steep wave was chosen to programme the wavemaker
Moreover, implicit coupling of wave components might make the using the stream function theory for the simulating of the wave inter­
amplification of nonlinearities in the axial force. action with the FPV structure. Once the floats did not detach from the
free surface, the time history of the structure dynamics was compared
6. Conclusions with the experimental results which revealed the ability and accuracy of
the present model. To demonstrate the range of validity of the proposed
A hydrodynamic framework was outlined in the present study to numerical model, an irregular wave was taken to interact with the
calculate the nonlinear dynamics of the solar platform under the ocean platform and the spectral analysis was performed on the time history of
waves. The buoyancy of the platform is reserved by double cylinders the platform responses to study the nonlinearities of the solutions. In
which turn the platform into a twin hull body for hydrodynamic anal­ summary, the achievements of this study are concluded as.
ysis. Therefore, the gap between the cylinders might affect the dynamics
of the platform which is exposed to the waves. To investigate the plat­ • The fully nonlinear numerical potential model is a proper method to
form response in waves, a set of experiments and a fully nonlinear nu­ compute the nonlinearities of the interaction of nonlinear waves and
merical simulation were performed in the present study for a broad the floating solar platform.
range of incident waves. The nonlinearity of the wave was widened from • A fully numerical model was developed to simulate the twin hull
the second order Stokes wave as a weak nonlinear incident wave to the platform.
near breaking wave, that was calibrated in the physical wave tank. The • the stream function theory was implemented into the numerical
physical model testing was performed on a single cylinder and a floating model to emulate the near breaking incident waves.
platform with double cylinders. On the other hand, a fully nonlinear • The stability of the numerical model was lost once the platform de­
potential flow model was developed based on the NURBS boundary taches from the free surface.
element method and MEL approach to computing the fluid flow. The • The turbulent flow due to wave breaking phenomena that occurs on
model was enhanced by the acceleration potential function to determine the free surface confined amongst the twin hull structure is not taken
the dynamics of the floating structures implicitly using the decomposi­ into account in the numerical simulation.
tion method. The convergence and accuracy of the numerical algorithm • The present framework is proper for the parametric study to find out
were evaluated for different mesh sizes and the time step sizes. A free the amplification of nonlinearities over a range of irregular waves.
decay test for a single cylinder was used to examine the performance of • The maximum relative difference between the numerical results and
the present computer model for a range of computations. Then the time the physical tests is ultimately less than 6% which occurs for the
history of the free decay test simulation was compared with the highest steep wave interacting with the floating solar platform.
experiment. • The uncertainty in both the experimental and numerical results
Dynamics of the single cylinder and the FPV platform, which were might be considered as a source of the incompatibility throughout
estimated experimentally and numerically were compared in the time comparisons.
domain to evaluate the reliability of the study. Hence motions of the
structures in two-degree freedom and the surge forces were used for the CRediT authorship contribution statement
comparison. Though there was good compliance over the majority of the
comparisons, there were considerable deviations in a few cases which Arash Abbasnia: Conceptualization, Methodology, /, Study design,
may have been triggered by viscous effects as well as the significant Software, Validation, Formal analysis, Investigation, Resources, Data
turbulent flow over the confined free surface between the floats. To curation, Writing – original draft, Writing – review & editing, Visuali­
investigate the capability of the present numerical model in the simu­ zation. Madjid Karimirad: Conceptualization, Methodology, /, Study
lation of nonlinearities, spectral analysis was conducted on the axial design, Validation, Formal analysis, Investigation, Resources, Writing –
forces on the FPV structure for the sea states which led to the strongest review & editing, Supervision, Project administration, Funding acqui­
nonlinearity in the time history of the records. The comparison of the sition. Dallán Friel: Conceptualization, Methodology, /, Study design,
spectrums revealed the ability of the numerical model in the simulation Validation, Formal analysis, Investigation, Resources, Data curation,
of the nonlinearities. To quantify the deviation between the present Writing – original draft, Writing – review & editing, Visualization.
numerical model and the experimental data, the RMS error of the heave Trevor Whittaker: Conceptualization, Methodology, /, Study design,

16
A. Abbasnia et al. Ocean Engineering 257 (2022) 111320

Formal analysis, Investigation, Resources, Writing – review & editing, Cointe, R., 1990. Numerical simulation of a wave channel. Eng. Anal. Bound. Elem. 7,
167–177. https://doi.org/10.1016/0955-7997(90)90003-R.
Supervision, Project administration, Funding acquisition.
Coslovich, F., Kjellberg, M., Östberg, M., Janson, C.E., 2021. Added resistance, heave and
pitch for the KVLCC2 tanker using a fully nonlinear unsteady potential flow
Declaration of competing interest boundary element method. Ocean. Eng. 229 https://doi.org/10.1016/j.
oceaneng.2021.108935.
Datta, R., Sen, D., 2006. A B-spline solver for the forward-speed diffraction problem of a
The authors declare that they have no known competing financial floating body in the time domain. Appl. Ocean Res. 28, 147–160. https://doi.org/
interests or personal relationships that could have appeared to influence 10.1016/j.apor.2006.06.001.
the work reported in this paper. Friel, D., Karimirad, M., Whittaker, T., Doran, J., 2021. Hydrodynamic investigation of
design parameters for a cylindrical type floating solar system. In: Guedes Soares, C.
(Ed.), Developments in Renewable Energies Offshore. Taylor & Francis Group,
Acknowledgements London, pp. 763–770. https://doi.org/10.1142/9789814317139_0008.
Friel, D., Karimirad, M., Whittaker, T., Doran, J., Howlin, E., 2019. A review of floating
photovoltaic design concepts and installed variations. In: 4th International
This work is supported by the Centre for Advanced Sustainable En­ Conference on Offshore Renewable Energy. CORE2019 Proceedings. ASRANet Ltd,
ergy (CASE). CASE is funded through Invest NI’s Competence Centre Glasgow.
Programme and aims to transform the sustainable energy sector through Ganesan, T.S., Sen, D., 2019. Prediction of non-linear wave loads on large floating
structures using a 3D numerical wave tank approach. Mar. Syst. Ocean Technol. 14,
business research. The funded project is entitled “R3245NBE: Floating 129–152. https://doi.org/10.1007/s40868-019-00062-3.
Solar Supporting De-Carbonisation of N. Ireland” which is run along the Gao, Z., Zou, Z., 2008. A NURBS-based high-order panel method for three-dimensional
collaboration with the industrial partners: B9 Energy Ltd, Brook Hall radiation and diffraction problems with forward speed. Ocean. Eng. 35, 1271–1282.
https://doi.org/10.1016/j.oceaneng.2008.02.007.
Estate, Senergy Innovations Ltd, and SolarMarine Energy Ltd. Special
Golroodbari, S.Z., van Sark, W., 2020. Simulation of performance differences between
thanks to Eamon Howlin and Jon Hancock from SolarMarine Energy Ltd offshore and land-based photovoltaic systems. Prog. Photovoltaics Res. Appl. 28,
for assisting in the technical review of the paper and in progression on 873–886. https://doi.org/10.1002/pip.3276.
the hydrodynamics of floating solar. Hasselmann, K., Barnett, T.P., Bouws, E., Carlson, H., Cartwright, D.E., Enke, K.,
Ewing, J.A., Gienapp, H., Hasselmann, D.E., Kruseman, P., Meerburg, A., M, P.,
The Bryden Centre project is supported by the European Union’s Olbers, D.J., Richter, K., Sell, W., W, H., 1973. Measurements of wind-wave growth
INTERREG VA Programme, managed by the Special EU Programmes and swell decay during the joint north sea wave project (JONSWAP). Ergänzungsh.
Body (SEUPB). The views and opinions expressed in this paper do not zur Dtsch. Hydrogr. Zeitschrift 1–95.
Hooper, T., Armstrong, A., Vlaswinkel, B., 2021. Environmental impacts and benefits of
necessarily reflect those of the European Commission or the Special EU marine floating solar. Sol. Energy 219, 11–14. https://doi.org/10.1016/j.
Programmes Body (SEUPB). solener.2020.10.010.
Kim, S.J., Kim, M.H., Koo, W., 2021. Nonlinear hydrodynamics of freely floating
symmetric bodies in waves by three-dimensional fully nonlinear potential-flow
References numerical wave tank. Appl. Ocean Res. 113, 102727 https://doi.org/10.1016/j.
apor.2021.102727.
Abbasnia, A., Ghiasi, M., 2014. A fully nonlinear wave interaction with an array of Kim, S.J., Koo, W., 2019. Development of a three-dimensional fully nonlinear potential
submerged cylinders by NURBS numerical wave tank and acceleration potential. numerical wave tank for a heaving buoy wave energy converter. Math. Probl Eng.
Ships Offshore Struct. 9, 404–417. https://doi.org/10.1080/ 17. https://doi.org/10.1155/2019/5163597, 2019.
17445302.2013.819688. Mignone, A., Inghirami, G., Rubini, F., Cazzaniga, R., Cicu, M., Rosa-Clot, M., 2021.
Abbasnia, A., Ghiasi, M., Abbasnia, A.H., 2017. Irregular wave transmission on bottom Numerical simulations of wind-loaded floating solar panels. Sol. Energy 219, 42–49.
bumps using fully nonlinear NURBS numerical wave tank. Eng. Anal. Bound. Elem. https://doi.org/10.1016/j.solener.2020.11.079.
82, 130–140. https://doi.org/10.1016/j.enganabound.2017.06.008. Ning, D., Wang, R., Zhang, C., 2017. Numerical simulation of a dual-chamber oscillating
Abbasnia, A., Guedes Soares, C., 2020. OpenMP parallelism in computations of three- water column wave energy converter. Sustain. Times 9. https://doi.org/10.3390/
dimensional potential numerical wave tank for fully nonlinear simulation of wave- su9091599.
body interaction using NURBS. Eng. Anal. Bound. Elem. 117, 321–331. https://doi. Ning, D.Z., Wang, R.Q., Zou, Q.P., Teng, B., 2016. An experimental investigation of
org/10.1016/j.enganabound.2020.05.008. hydrodynamics of a fixed OWC Wave Energy Converter. Appl. Energy 168, 636–648.
Abbasnia, A., Guedes Soares, C., 2019. Hydrodynamic analysis of a land-based oscillating https://doi.org/10.1016/j.apenergy.2016.01.107.
water column device using fully nonlinear numerical wave flume. In: Advances in Piegl, L., Tiller, W., 1995. The NURBS Book. Comput. Des.. https://doi.org/10.1016/
Renewable Energies Offshore - Proceedings of the 3rd International Conference on 0010-4485(96)86819-9.
Renewable Energies Offshore. RENEW, 2018. Rosa-Clot, M., Giuseppe, M.T. (Eds.), 2020. Floating PV Plants. Brian Romer. https://doi.
Abbasnia, A., Guedes Soares, C., 2018. Fully nonlinear simulation of wave interaction org/10.1016/C2018-0-01890-3.
with a cylindrical wave energy converter in a numerical wave tank. Ocean. Eng. 152, Tang, Y., Sun, S.L., Ren, H.L., 2021. Numerical investigation on a container ship
210–222. https://doi.org/10.1016/j.oceaneng.2018.01.009. navigating in irregular waves by a fully nonlinear time domain method. Ocean. Eng.
Abbasnia, A., Guedes Soares, C., 2017. Exact evaluation of hydrodynamic loads on ships 223, 108705 https://doi.org/10.1016/j.oceaneng.2021.108705.
using NURBS surfaces and acceleration potential. Eng. Anal. Bound. Elem. 85, 1–12. Tanizawa, K., 2000. The state of the art on numerical wave tank. Proc. 4th Osaka Colloq.
https://doi.org/10.1016/j.enganabound.2017.09.005. Seakeeping Perform. Ships 95–114.
Abbasnia, A., Rezanejad, K., Guedes Soares, C., 2021. Adaptive fully nonlinear potential Vinje, T., Brevig, P., 1981. Nonlinear ship motions. In: 3rd International Conference on
model for the free surface under compressible air pressure of oscillating water Numerical Ship Hydrodynamics. Paris, p. 257, 10 pp., 10 diag., 2 graphs.
column devices. Eng. Anal. Bound. Elem. 133, 153–164. https://doi.org/10.1016/j.
enganabound.2021.08.018.

17