Coastal Engineering 126 (2017) 11–26

Contents lists available at ScienceDirect

Coastal Engineering
journal homepage: www.elsevier.com/locate/coastaleng

Towards simulating floating offshore oscillating water column converters

with Smoothed Particle Hydrodynamics
A.J.C. Crespo a, *, C. Altomare b, c, J.M. Domínguez a, J. Gonz

mez-Gesteira a
alez-Cao a, M. Go
a
Environmental Physics Laboratory, Universidade de Vigo, Campus As Lagoas s/n, 32004 Ourense, Spain
b
Flanders Hydraulic Research, Berchemlei 115, 2140 Antwerp, Belgium
c
Department of Civil Engineering, Ghent University, Technologiepark 904, 9052 Ghent, Belgium

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

Keywords: The mesh-free code DualSPHysics is applied to simulate the interaction between sea waves and an Oscillating
Oscillating water column Water Column device (OWC). In this work, capabilities and limitations of DualSPHysics are shown in simulating
Numerical modelling OWCs. On the one hand, the new capabilities of DualSPHysics are shown by simulating the effect of mooring
Meshless methods
systems on a floating offshore OWC. On the other hand, simulations only consider a single-phase (water) so that
Smoothed particle hydrodynamics
the full OWC behaviour is partially reproduced, i.e. air pressure fluctuations are not modelled. The model was first
Mooring lines
validated with one laboratory test that consists of a fixed OWC with an open chamber. Next, water surface os-
cillations inside the chamber of a real OWC (located in Mutriku, Spain) have been predicted using the prevalent
wave conditions observed in the area. Finally, the capabilities of DualSPHysics were demonstrated by simulating
an offshore OWC moored to the seabed.

1. Introduction i) LIMPET plant near the Scottish island of Islay with a total installed
capacity of 500 kW (Heath et al., 2000);
Significant research is being conducted into renewable resources due ii) PICO plant in Azores (Portugal) for a power of 400 kW (Falc~ao,
to the increasing demand for energy and to the uncertainty to climate 2000);
change. Wave energy is in fact one of the most available and cleanest iii) MUTRIKU plant in Spain that consists of 16 chambers of 18.5 kW
renewable energy sources. Wave energy has the advantage of being each (296 kW in total) and covers the energy consumption of 1000
considered as the most concentrated and least variable form of renewable population in 1 year (Torre-Enciso et al., 2009).
energy. Previous research (Drew et al., 2009) has shown that wave power
devices can generate power up to 90 per cent of the time, compared to 25 However, sea waves propagating towards the coast suffer from
per cent for solar and wind devices. attenuation, refraction and shoaling as they approach the shoreline
There are several projects worldwide regarding wave energy, but the (Goda, 2010). So that, some of the wave power is lost and offshore
potential of this source is still not fully investigated. At present, different floating OWCs can be a better option. Some examples of floating OWC
wave energy concepts are being investigated by companies and academic devices at an advanced stage of development are the OE Buoy and the
research groups. Oscillating water column (OWC) devices consist of a Oceanlinx Mk3. The OE Buoy is developed by Ocean Energy company
partially submerged reservoir with water open to the sea and a chamber and has been deployed in Atlantic waters and demonstrated the ability to
of trapped air. The ocean waves change the water level inside the tank, generate power and survive in the most extreme conditions of the ocean
which compresses and decompresses the air inside the chamber. This (http://oceanenergy.ie/platform/). A 1/4th scale model of the OE Buoy
trapped air is allowed to flow to and from the atmosphere via a turbine device was tested in the Galway Bay (Ireland) (Thiebaut et al., 2011) and
whose rotation is used to generate electricity. The shoreline OWC's are the Oceanlinx Mk3 prototype was tested in the coast of Port Kembla
currently the most sensible designs since they do not have any moving (Australia), in 2010 (Falc~ao, 2010).
parts in the water, leading to easier maintenance works. There are several More complete reviews about OWC can be found in Heath (2012) and
full-scale prototypes of OWC around the world such as: Falc~ao and Henriques (2016) and the linear interactions between ocean
waves and oscillating systems are properly described in Falnes (2002).

* Corresponding author.
E-mail address: [email protected] (A.J.C. Crespo).

http://dx.doi.org/10.1016/j.coastaleng.2017.05.001
Received 30 March 2016; Received in revised form 3 January 2017; Accepted 12 May 2017
Available online 6 June 2017
0378-3839/© 2017 Elsevier B.V. All rights reserved.
A.J.C. Crespo et al. Coastal Engineering 126 (2017) 11–26

Three important steps can be considered in the design of a wave the motions of ships and assumes that all the hydrodynamic forces on a
energy converter; numerical modelling, model scale tests in wave tanks floating body (i.e. wave energy converters) can be modelled using a set of
and testing in situ. Many designs have been developed and tested through hydrodynamics coefficients. With information of incoming waves as
numerical and physical modelling (i.e. wave tanks) at a research stage input, BEM computes added masses, the radiation damping coefficients
and only a small number of devices has been tested in the ocean. In fact, and excitation force coefficients. The movements of the structure (heave,
these prototypes are generally built by companies or thanks to private pitch and roll) are obtained from the solution of the frequency-domain or
investors. Laboratory experiments on an axisymmetric floating OWC in a time-domain equations and require information on the sea state and on
wave channel were reported by Whittaker and McPeake (1985). Morris- the PTO (power take-off system). The main advantage of these models is
Thomas et al. (2007) carried out experiments to determine the influence that the codes are very fast and efficient. The problem is that assuming a
of wall thickness, shape of the front wall and draught of the front wall for linear behaviour, WECs cannot be modelled in energetic sea states or
various wave parameters on the hydrodynamic efficiency of an OWC close to resonance. The main limitations derive from the small wave
device. More recently, L
opez et al. (2015) presented physical model tests amplitude and small motion amplitude assumptions and the incapability
that allowed to quantify the effects of the damping coefficient, wave to account for real fluid (viscous) effects (boundary layers, turbulence,
conditions and tidal level on the performance of an OWC chamber. vortex shedding).
Nevertheless, there are insufficient experimental databases with OWC The approach of Evans (1982) was later applied to particular OWC
devices and, in most of the cases, the variables of interest (chamber free- geometries, using BEM (Brito-Melo et al., 2001; Delaur
e and Lewis, 2003;
surface oscillation, air pressure variation and air flux between the Josset and Cl
ement, 2007). Alves et al. (2010) performed a numerical
chamber and the atmosphere) have not been measured jointly. Thus, analysis of an axisymmetric floating OWC using a boundary element
numerical models are a suitable and necessary tool to contribute in OWCs method to account for the hydrodynamic interferences between the buoy
design once models are properly validated against experimental data. (a cylindrical floater with a tail tube) and the OWC. Iturrioz et al. (2014)
The main advantage of the numerical simulation is the capability to validated their own time domain model with experimental data of a fixed
simulate complex scenarios and provide physical data that can be diffi- detached OWC. Some software examples are the commercial WAMIT
cult, or even impossible, to measure in real or scale models. Despite of the (WAMIT, 2012), WADAM (DNV, 2008) or WaveDyn and the open-source
accuracy of the numerical models, these cannot replace the construction codes Nemoh developed by LHEEA (Babarit and Delhommeau, 2015) and
of scale models, but they can reduce significantly the number of physical WEC-SIM developed by NREL/Sandia (Lawson et al., 2015; Combourieu
tests. This leads to important savings since the construction of physical et al., 2015) at http://wec-sim.github.io/WEC-Sim/.
models is very expensive and time consuming. The second main group consists of those models based on the Rey-
The hydrodynamic interaction between WECs and ocean waves is a nolds Averaged Navier–Stokes (RANS) equations and presents several
complex non-linear process that has being numerically studied using advantages, not only for solving the velocity field in the whole domain
different approaches. McCormick (1974, 1976) developed the first nu- but also for overcoming the limitations of nonlinearity. Thereby, CFD
merical models on OWC devices, based on empirical values for the hy- models that approximate Navier–Stokes equations are considered one of
drodynamic coefficients. Evans (1978) developed a theoretical model for the best numerical tools to study the hydrodynamic interaction between
a fixed OWC device considering the internal free surface as a weightless waves and WECs. Particularly powerful within this group are those
piston. Later the models of Evans (1982) and Sarmento and Falc~ao (1985) models that include the Volume Of Fluid (VOF) method to capture the
considered the deformation of the free surface through the application of movements of the free surface. These models can simulate viscous losses
the oscillating surface-pressure distribution condition. The numerical and non-linearities that occur in the interaction between the device
models to simulate OWC that can be found in the literature are here (fixed or floating) and the wave train, so that, violent flows with large
summarised in two groups: i) time domain models based on linear water amplitudes can now be simulated. However, they need an extra algo-
wave theory, and ii) CFD codes based on the integration of the Navier- rithm to track free-surface and meshing complex geometries or floating
Stokes equations (meshbased and meshless). Table 1 summarises the bodies is a hard task. Actually, due to computational requirements, the
main features of the different approaches that will be presented in detail. numerical integration of the RANS equations to model OWC converters
In the first group, time-domain models based on frequency domain was applied, in many cases in the literature, to two-dimensional geom-
data are usually built upon the Cummins equation (Cummis, 1962). The etries. Paix~ao Conde and Gato (2008) carried out a numerical study of an
Boundary Element Method (BEM) is used to solve the Laplace equation OWC with the commercial CFD code Fluent (also based on finite vol-
for the velocity potential, which assumes the flow is inviscid, incom- umes), investigating the flow distribution in the chamber and the prop-
pressible, and irrotational. BEM was originally formulated for analysing erties of the air-jet impinging on the free-surface. Teixeira et al. (2013)
analysed the chamber geometry (front wall depth, chamber length and
chamber height) and turbine characteristic relation by means of Fluinco
Table 1 model (a semi-implicit Taylor-Galerkin method) where a comparison
Numerical models for solving the dynamics involved in WEC analysis.
between Fluinco and Fluent models was also carried out, obtaining a
Time domain models CFD models good agreement. There are many other works using two-phases RANS
Meshbased CFD Meshless CFD and VOF models for OWC modelling, but few of them show validation
Equations Cummis Navier Stokes Navier Stokes
with experimental data. Zhang et al. (2012) developed a 2D-RANS model
equation þ Linearised to study wave interaction with a semi-submerged OWC chamber and
hydrodynamic coeffs. analysed its impact on the energy efficiency. The validation of the model
Method Boundary Element Method Finite Volume Smoothed Particle was carried out using the experiments presented by Morris-Thomas et al.
Hydrodynamics
(2007). Some recent works also present the validation of the STAR-
Viscosity Inviscid Viscous Viscous
Linearity Linear Non-linear Non-linear CCM þ model in L
opez et al. (2014) where a RANS-VOF numerical model
Suited for Low amplitude motions Viscous losses Large is used to study the OWC performances for different wave conditions and
Small oscillations deformations þ Rapidly damping values. L
opez and Iglesias (2014) applied artificial neural net-
moving geometries works (ANNs) to predict the pneumatic efficiency of OWC converters.
Efficiency Fast and efficient Time Very time consuming
consuming þ Mesh
The validation of the open-source code REEF3D (https://reef3d.
generation wordpress.com/) is presented in Kamath et al. (2015a). REEF3D is also
Codes WAMIT, WADAM. VOF, OpenFoam, DualSPHysics used in Kamath et al. (2015b) where the PTO damping on the chamber is
AquaDyn, WaveDyn IH-Foam, Fluent, represented using a linear pressure drop law with the permeability co-
WEC-Sim, Nemoh Fluinco, REEF3D
efficient derived from Darcy's equation for flow through porous media.

12
A.J.C. Crespo et al. Coastal Engineering 126 (2017) 11–26

The model IH-Foam is also employed in Iturrioz et al. (2015) where sphysics.org. The first validation of the GPU implementation of the
numerical results of chamber free-surface oscillation and air pressure code was presented in Crespo et al. (2001) and more details about
variation are compared with experimental data performed in IH- DualSPHysics can be found in Crespo et al. (2015). Recently, Dual-
Cantabria. Luo et al. (2014) employed Fluent to simulate a heave-only SPHysics code was applied to coastal engineering problems, e.g. to study
floating OWC including validation against experimental, analytical and the run-up on a real armour block coastal breakwater in Altomare et al.
other numerical results. (2014) and to estimate sea wave impact on coastal structures (Altomare
Nevertheless, models based on finite volumes present the limitations et al., 2015a) where the numerical results have been validated with the
coming from the meshbased approach to deal with extremely large de- experimental data for typical cases from the Belgian coast. In addition,
formations and rapidly moving geometries. for the purpose of providing a general model, relevant modes of inter-
Although there are many methods based on RANS for wave-structure action are not always evident, in which case high spatial and temporal
interaction, including the modelling of floating objects, traditional resolutions must be attainable. Furthermore, some simulations may
methods are either meshbased or meshless methods. The meshbased require remarkably large domains. This stresses the need for high per-
approach solves the equations at fixed nodes in the space. In the meshless formance codes and implementations and it means that DualSPHysics is a
description, the positions where equations are solved move with the fluid perfect candidate since it is the most efficient SPH code worldwide
and a fixed mesh is not used. The meshbased methods (finite elements, (Domínguez et al., 2013; 2013b).
finite differences and finite volumes) are currently very robust, well In summary, simulating OWCs is an arduous task, especially for
developed and have been applied to a wide range of applications floating devices, that involves multiphysics and multiscale. A full simu-
providing highly accurate results. These meshbased methods are ideal for lation implies: i) properly solving the wave field; ii) handling the inter-
systems where the domain is perfectly defined and for simulations where action between waves and the device; iii) reproducing the movement of
the boundaries remain fixed. However, the creation of the mesh can be the device under external forces; iv) solving the interaction between
very inefficient if the system is complex. In recent years, numerous water and air inside the chamber and, finally; v) introducing the turbine
meshless methods have appeared and grown in popularity as they can be that moves according to air flow due to the pressure changes inside the
applied to problems that are highly nonlinear in arbitrarily complex chamber. As far as we know, there is not a single model able to deal with
geometries and are difficult for mesh-based methods. Within the mesh- the five steps of the full simulation. Here we only focus on the three first
less methods now available, Smoothed Particle Hydrodynamics (Violeau, steps. As we mentioned above, one of the main advantages of the SPH
2012) is, possibly, the most popular and has attained the required level of technique is its ease to introduce solid objects inside the fluid that move
maturity to be used for engineering purposes. Hence, SPH is an ideal following the forces induced by waves. In spite of DualSPHysics already
technique to simulate free-surface flows and presents several advantages includes a multiphase solver (Mokos et al., 2015) it was not included
compared with meshbased methods to simulate violent wave-structure here. The reason for this decision is twofold, on the one hand the
interaction: i) there is no special treatment to detect the free surface so computational efficiency of the solver is still limited (it must be opti-
large deformation can be efficiently treated since there is no mesh mised) and, on the other hand, we are here interested on proving first the
distortion, ii) moving complex boundaries and interfaces are easily capability of the model to deal with floating moored objects. Thus, the
handled, iii) multiphase flows are simulated without the need of special aim of this paper is the modelling of the hydrodynamic interaction of
variables to detect the phases in the space since each individual particle waves with the OWC device, so that the air pressure is not simulated at
holds material properties of its phase and iv) natural incorporation of this stage and the pneumatic conversion of the OWC mechanism is not
coefficient discontinuities and singular forces into the numerical scheme. considered. The DualSPHysics model used to simulate OWC is presented
SPH has been used to describe a variety of free-surface flows (wave here as a first step towards the modelling of wave energy converters in
propagation over a beach, plunging breakers, impact on structures and extreme conditions. The code is first validated with experiments (labo-
dam breaks). Monaghan (1994) presented the first attempt to study free- ratory scale) of regular waves interacting with a fixed OWC chamber with
surface flows. Monaghan also studied the behaviour of gravity currents, such an orifice in the top that the air pressure in the chamber was
solitary waves and wave arrival at a beach. Later on, the model was negligible. The model is then applied to simulate an OWC chamber of
applied to the study of the wave-structure interaction such as in Cola- Mutriku (Spain) using its real dimensions; proper wave generation,
grossi and Landrini (2003) that considered the study of interfacial flows. propagation and active absorption are guaranteed so chamber free-
The classical dam-break problem is also studied in 3D by G
omez-Gesteira surface oscillations are provided for different wave conditions. Finally,
and Dalrymple (2004). SPH has been employed to the study of the wave- a floating OWC device is reproduced using DualSPHysics to mimic its
breakwater interaction in Gotoh et al. (2004) and to the prediction of behaviour in real sea conditions.
wave impact pressure due to sloshing waves in Khayyer and Gotoh
(2009). SPH has been successfully applied to real engineering applica- 2. Smoothed Particle Hydrodynamics model
tions such as to simulate free-surface flows encountered in Pelton tur-
bines (Marongiu et al., 2010) and to study a real spillway in France 2.1. SPH basis
connecting the reservoir of a river dam to a valley with complex bottom
shape (Lee et al., 2010). However, there are few works in literature Smoothed Particle Hydrodynamics is a meshless method that de-
where SPH is used to simulate wave energy converters. Rafiee et al. scribes the fluid as a set of discrete elements (named particles). The
(2013) and Edge et al. (2014) simulated oscillating wave surge devices physical quantities of a particle (a) can be computed as an interpolation
and Yeylaghi et al. (2015) studied point absorber devices but no other of the values of the nearest neighbouring particles. Mathematically, the
works deal with the simulation of oscillating water column using contribution of the neighbouring particles is weighted according to their
SPH models. distance from particle a using a kernel function (W) and a smoothing
The DualSPHysics code is developed by researchers from Uni- length (h). Herein, the distance from particle a at which the interaction
versidade de Vigo (Spain) and The University of Manchester (U.K.). with the rest of the particles can be neglected is 2h.
DualSPHysics has been developed to use SPH for real engineering The fundamental principle is to approximate any function F by the
problems with software that can be run on either CPUs or GPUs (graphics integral approximation:
cards with powerful parallel computing). GPUs offer now a higher
computing power than CPUs and they are an affordable option to FðrÞ ¼ ∫ Fðr'ÞWðr  r'; hÞdr' (1)
accelerate SPH with a low economic cost. Thereby, the simulations can
be performed using a GPU card installed on a personal computer. Dual- where W is the kernel function, h the smoothing length and r the position
SPHysics is open source and can be freely downloaded from www.dual. vector (r’ is the position where the function F is defined).

13
A.J.C. Crespo et al. Coastal Engineering 126 (2017) 11–26

This equation can be expressed in a discrete form, where the mass terms (Monaghan, 1992), since the weighted summation results in a
approximation of the function is interpolated at a given location (or decrease of the density in the interface between fluids, near the free
particle a) and a summation is performed over all the particles within the surface, and in the proximity of the boundaries.
region of compact support of the kernel: The fluid in the standard SPH formalism is treated as compressible,
X which allows the use of an equation of state to determinate fluid pressure.
Fðra Þ≈ Fðrb ÞWðra  rb ; hÞΔvb (2) However, the compressibility is adjusted to slow the speed of sound so
b that the time step in the model (dependent of the speed of sound) has a
reasonable value. This facilitates the use of an equation of state to
where the subscript a denotes an individual particles, Δvb is the volume of
determinate the pressure of the fluid since this is faster than solving the
a neighbouring particle (b) and the kernel function follows that
Poisson's equation that appears in the incompressible approach.
X Following Monaghan (1994) and Batchelor (1974), the relationship be-
Wðra  rb ; hÞ≈1 (3)
b
tween pressure and density follows the Tait's equation of state. It can be
noted that a small oscillation in density will generate a large variation
If vb ¼ mb/ρb, being m and ρ, the mass and the density of particle b. in pressure.
respectively, then Eq. (2) becomes:
 γ 
X ρ
mb P¼B 1 (9)
Fðra Þ≈ Fðrb Þ Wðra  rb ; hÞ (4) ρ0
b
ρb
The parameter B is related to the compressibility of the fluid;
Performance of an SPH model depends heavily on the choice of the ρ0 ¼ 1027 kg/m3 is the reference density, taken as the density of the fluid
smoothing kernel. Kernels are expressed as a function of the non- at the surface and γ is the polytrophic constant that usually ranges from 1
dimensional distance between particles (q), given by q ¼ r/h, where r to 7. The value of B ¼ c2 ρ0 =γ provides a maximal limit for the density.
is the distance between any two given particles a and b and the parameter The speed of sound, c, is defined as the square root of the derivative of
h (the smoothing length) controls the size of the area around particle a in Tait's equation with respect to density. The choice of B plays a key role
which neighbouring particles are considered. The Quintic (Wendland, since it determines the speed of sound. Using a value corresponding to
1995) is here employed with an influence domain of 2h ðq  2Þ the real value of the speed of sound in water, a very small time step must
 q4 be chosen for numerical modelling, based on the Courant–Frie-
W ðr; hÞ ¼ αD 1  ð2q þ 1Þ 0 q 2 (5) drichs–Lewy condition. Monaghan showed that the speed of sound could
2
be slowed significantly in an artificial way for fluids without affecting the
where αD is equal to 7/4πh2 in 2-D and 21/16πh3 in 3-D. fluid motion, however Monaghan (1994) suggests that the minimum
speed of sound should be about ten times greater than the maximum
expected flow velocity.
2.2. Governing equations
Equations are integrated in time using a two-stage Symplectic
method. This integration algorithm is time reversible in the absence of
The artificial viscosity scheme, proposed by Monaghan (1992), is a
friction or viscous effects (Leimkuhler et al., 1996). It can also preserve
common method within fluid simulation using SPH due primarily to its
geometric features, such as the energy time-reversal symmetry present in
simplicity. In SPH notation, momentum equation can be written as
the equations of motion, leading to improved resolution of long-term
d va X Pb þ Pa 
solution behaviour. The scheme used here is an explicit second-order
¼ mb þ Π ab ∇a Wab þ g (6) Symplectic scheme with an accuracy in time of O (Δt2) and involves a
dt b
ρb ρa
predictor and corrector stage. With explicit time integration schemes the
where Pk and ρk are the pressure and density that correspond to particle k time step is dependent on the Courant-Friedrich-Levy (CFL) condition,
(as evaluated at a or b) and g is gravitational acceleration. The viscosity the force terms and the viscous diffusion term, so that, a variable time
term Π ab is given by step Δt is calculated according to Monaghan and Kos (1999).

8
< α cab μab
>
vab ⋅ r ab < 0 2.3. Boundary conditions
Πab ¼ ρab (7)
>
: 0 vab ⋅r ab > 0 In DualSPHysics, the boundary is described by a set of particles that
are considered as a separate set to the fluid particles. The Dynamic
where ρab ¼ 0:5ðρa þ ρb Þ, rab ¼ ra  rb and vab ¼ va  vb vab ¼ va  vb with Boundary Condition (DBC) is the default method provided by Dual-
rk and vk being the particle position and velocity respectively. μab ¼ hvab ⋅ SPHysics (Crespo et al., 2007). This method uses boundary particles that
r ab =ðrab
2
þ η2 Þ, cab ¼ 0:5ðca þ cb Þ is the mean speed of sound, α is a coef- satisfy the same equations as fluid particles, however they do not move
ficient that needs to be tuned in order to introduce the proper dissipation according to the forces exerted on them. Instead, they remain either fixed
in position or move according to an imposed/assigned motion function
and η2 ¼ 0:01h2 avoids numerical divergence when the distance between
(i.e. moving objects such as gates or wave-makers). When a fluid particle
particles tends to zero. The value of α ¼ 0.01 has proven to give the best
approaches a boundary and the distance between the boundary particles
results in the validation of wave flumes to study wave propagation and
and the fluid particles becomes smaller than twice the smoothing length
wave loadings exerted onto coastal structures (Altomare et al.,
(h), the density of the affected boundary particles increases, resulting in a
2015a, 2015b).
pressure increase. In turn, this results in a repulsive force being exerted
The mass of each particle remains constant throughout the duration
on the fluid particle due to the pressure term in the momentum equation.
of a weakly-compressible SPH simulation (as presented herein) and only
Stability of this method relies on the length of time step taken being
their associated density fluctuates. These density changes are computed
suitably short in order to handle the highest present velocity of any fluid
by solving the conservation of mass, or continuity equation, in SPH form:
particles currently interacting with boundary particles and it is therefore
dρa X an important issue when considering how the variable time step is
¼ mb vab ⋅∇a Wab (8) calculated. Validations with dam-break flows and sloshing tanks have
dt b
been published with good results and also comparing these BCs with
A time differential is used instead of a weighted summation of the other approaches (Domínguez et al., 2015). In addition, DBC have also

14
A.J.C. Crespo et al. Coastal Engineering 126 (2017) 11–26

been shown to be suitable to reproduce complex geometries (Altomare for an entire body. By assuming that the body is rigid, the net force on
et al., 2014). each boundary particle is computed according to the sum of the contri-
butions of all surrounding fluid particles according to the designated
2.4. Extra functionalities kernel function and smoothing length. Each boundary particle k experi-
ences a force per unit mass given by
2.4.1. Wave generation X
The waves are generated in DualSPHysics by means of moving fk ¼ f ka (10)
a2WPs
boundaries that aim to mimic the movement of a wavemaker as in
physical facilities. The wave generation using moving boundary in
where fka is the force per unit mass exerted by the fluid particle a on the
DualSPHysics consists of piston- and also flap-type wavemakers.
boundary particle k, which is given by
First- and second-order wave generation theories are implemented in
DualSPHysics. The first-order generation is fully described in Altomare mk f ka ¼ ma f ak (11)
et al. (2015b), where generation methods for both monochromatic
(regular) and random (irregular) waves are reported. Altomare et al. For the motion of the moving body, the basic equations of rigid body
(2015b) analysed the performance of the wave generation against dynamics can then be used
theoretical results for long-crested waves. For second-order wave gen- X
dV
eration of monochromatic waves, the reader is referred to the solution M ¼ mk f k (12a)
proposed by Madsen (1971) where the author advises to use it only for dt k2BPs
HL2/d3 < 8π/3, where H and L are the wave height and wavelength
respectively (referred to a case of monochromatic waves) and d is the dΩ X
I ¼ mk ðr k  R0 Þ  f k (12b)
water depth at the wavemaker location in still water conditions. The first- dt k2BPs
order wave generation for monochromatic waves is extended to random
waves in DualSPHysics based on the method described in Liu and Frig- where M is the mass of the object, I the moment of inertia, V the velocity,
aard (2001). Then, the solution for random waves is extended to second- Ω the rotational velocity and R0 the centre of mass. Eqs. (12a) and (12b)
order wave generation to suppress spurious long waves. The method are integrated in time in order to predict the values of V and Ω at the
implemented in DualSPHysics is based on the correction for the signal of beginning of the next time step. Each boundary particle within the body
the wavemaker described in Barthel et al. (1983). Two standard wave has a velocity given by
spectra are implemented and used to generate random waves: JONSWAP
and Pierson-Moskowitz spectra. The generation system allows having uk ¼ V þ Ω  ðr k  R0 Þ (13)
different random time series with the same significant wave height (Hm0)
Finally, the boundary particles within the rigid body are moved by
and the same peak period (Tp), just defining different phase seeds.
integrating Eq. (13) in time. The works of Monaghan et al. (2003) and
In this way, wave height, wave period and depth are the key input
Monaghan (2005) show that this technique conserves both linear and
parameters in DualSPHysics, therefore the time series of wavemaker
angular momentum. Bouscasse et al. (2013) presented success valida-
displacement is computed using the aforementioned wave theory. Only
tions of nonlinear water wave interaction with floating bodies in SPH
the piston-type is used in this work.
comparing with experimental data that includes deformations in the free-
surface due to the presence of floating boxes and the movement of those
2.4.2. Active wave absorption
objects during the experiment (heave, surge and roll displacements).
Active wave absorption is used in physical facilities to absorb the
Several validations using DualSPHysics are performed in Canelas et al.
reflected waves at the wave-maker in order to avoid that they will be
(2015) and Ren et al. (2015). Canelas et al. (2015) analyses the
reflected back into the domain. In this way, the active absorption pre-
buoyancy-driven motion with solid objects larger than the smallest flow
vents the introduction into the system of extra spurious energy that will
scales and with various densities. They compared SPH numerical results
bias the results. With active absorption, the position of the wave maker is
with analytical solutions, with other numerical methods (Fekken, 2004)
corrected every time step.
and with experimental measurements. Ren et al. (2015) showed the ca-
The active wave absorption system (AWAS) implemented in Dual-
pabilities of DualSPHysics to simulate the nonlinear interactions between
SPHysics is based on the approach that appears in Shaffer and Klopman
waves and floating bodies. The evolution in time of the dynamic response
(2000). The water surface elevation η at the wave-maker position is used
of a freely floating body under nonlinear waves was successfully vali-
and transformed by an appropriate time-domain filter to obtain a control
dated with experimental results.
signal that corrects the wave paddle displacement in order to absorb the
reflected waves every time step.
2.4.4. Quasi-static mooring implementation
Hence, the target wave-maker position is corrected to avoid reflection
A new implementation to simulate the behaviour of mooring lines
at the wave-maker. The position in real time of the wave-maker is ob-
with DualSPHysics is presented in Barreiro (2015). This new approach
tained through the velocity correction of its motion. For a piston-type
allows reproducing the forces on floating bodies, such as vessels, boats,
wave-maker the velocity correction is calculated using linear long
wave energy devices and other off-shore structures moored to the seabed.
wave theory in shallow water (Didier and Neves, 2001; Shaffer and
A mooring line is a rope, a cable or a chain that holds in position any
Klopman, 2000). For that it is necessary to estimate the free-surface
kind of floating object, more precisely the implementation will be
elevation of the reflected waves, ηR, to be absorbed comparing the
focusing on continuous ropes and wires that can be described by the
target incident water surface elevation, ηI, with the measured one in front
catenary function (swing moorings). The catenary function refers to the
of the wave-maker, ηSPH. The corrected wave-maker velocity is then the
mathematical description of a line with mass hanging between two
summation of velocity correction and the theoretical incident wave-
points. These two points can be at any position relative to each other. The
maker velocity. For further details, the reader is referred to Altomare 
et al. (2015b). equation that must be satisfied in Cartesian coordinates is y ¼ ξ cosh xξ ,
where y and x are vertical and horizontal coordinates respectively, and ξ
2.4.3. Fluid-driven objects corresponds to the form parameter and this parameter shapes the cate-
The movement of an object by considering its interaction with fluid nary function width.
particles and using these forces to drive its motion can be also derived in The maths from the work of Faltinsen (1993) is behind this
DualSPHysics. This can be achieved by summing the force contributions

15
A.J.C. Crespo et al. Coastal Engineering 126 (2017) 11–26

implementation and the solution for the inelastic catenary equations is maximum top slot used in (Iturrioz et al., 2014), and it was chosen here
obtained. In general, the force of the chain corresponds to the proportion since the goal of the manuscript is not to simulate pressure changes inside
of the chain that is lifted. In this way, different conditions for moorings the chamber as mentioned above. It is a well known fact that the spring-
are considered such as the resting chain, partially resting chain, partially like effect of air compressibility increases with the chamber height as
extended chain and the fully extended chain. studied in Sarmento and Falc~ao (1985). However, at the scale 1/30 and
This mooring implementation has been already validated in Barreiro with a top opening of 5 cm, the pneumatic part of the problem was
(2015) with experimental data from Johanning et al. (2006), 2007) negligible for this test, and the pressure in the OWC air chamber was
where static experiments were performed for single and multiple lines. equal to the atmospheric pressure.
Some instants of the simulation with DualSPHysics can be observed in
3. Validation Figs. 2 and 3. Colour of the particles corresponds to the velocity value.
The effect of the dissipative beach can be observed in Fig. 2 where waves
The experiments carried out in the IH-Cantabria wave flume and break and no reflection is created. The zoom into the area of the OWC
employed in Iturrioz et al. (2014) are used here for the validation of chamber is shown in Fig. 3. The instants chosen in Figs. 2 and 3 represent
DualSPHysics. The experiments were carried out using Froude's scale the interaction of one individual wave with the OWC chamber: at
laws with a geometrical scale of 1/30. The tank was 20.60 m long and the time ¼ 68.8s the wave is approaching, at time ¼ 69.2s the wave is already
water depth was of 0.60 m. A cavity open in the bottom to the channel close to the left wall, at time ¼ 70s the wave has impacted with the OWC
was placed at 5.4 m of the piston to mimic an OWC chamber. A dissi- lateral wall and at time ¼ 70.8s the wave is propagating through the end
pative beach was located at the end of the flume to avoid reflection. Free- of the domain.
surface elevation was measured in the middle of OWC during the phys- Fig. 4 represents the free-surface elevation measured in the middle of
ical tests. This free-surface elevation was measured in three parallel OWC chamber using the time-domain model presented in Iturrioz et al.
sections in order to prove the two-dimensionality of the experiments. The (2014), using IH-Foam (same meshbased CFD model used in Iturrioz
numerical simulations are therefore executed only in 2D. Fig. 1 shows the et al., 2015) and using DualSPHysics. The water surface elevation
setup of the numerical tank with the same dimensions as the experi- measured during the physical model tests is also included for compari-
mental one. son. As it should be expected, SPH numerical results were found to be in
A set of regular and irregular wave series with wave heights (H) better agreement with experimental results than with results from the
varying between 0.03 and 0.08 m and periods (T) between 1.1 and 3.2 s time-domain analysis based on linear water wave theory. It can be
were tested with the different top slots. In this work, the experiment with observed how DualSPHysics and IH-Foam predict the crests and troughs
regular waves of T ¼ 3.2 s, H ¼ 0.08 m, initial water depth d ¼ 0.6 m and of free-surface oscillations inside the chamber better than the time-
with a top opening of 0.05 m is numerically simulated. This is the domain model. DualSPHysics and IH-Foam show a similar accuracy for

Fig. 1. Numerical domain with dimensions of the experiment from Iturrioz et al. (2014).

Fig. 2. Different instants of the simulation with regular waves interacting with the fixed OWC in the wave flume of IH-Cantabria.

16
A.J.C. Crespo et al. Coastal Engineering 126 (2017) 11–26

Fig. 3. Different instants of the simulation with zoom into the area of the OWC chamber.

Fig. 4. Comparison of experimental data (Iturrioz et al., 2014) with chamber free-surface oscillations obtained using the time-domain model, IH-Foam and DualSPHysics code.

this validation case. It is worth noting that the time-domain model is physical time took 31 min on a Tesla K40), dp ¼ 0.01 m led to 79,000
applied under several simplifications (assuming an inviscid, incom- particles (it took 2 h on the same GPU) and dp ¼ 0.005 m led to 314,000
pressible, and irrotational flow) and, unlike the SPH model here pre- particles (it took 10.4 h).
sented, it is able of modelling the damping due to air pressure in the The statistical parameters from Eqs. (14) and (15) are computed to
chamber. On the other hand, the objective of using a time-domain model measure the agreement of the different numerical results with the
is different to the CFD models; a linear behaviour is assumed and only experimental data as shown in Table 2. Note that values of AF higher than
motions of small wave amplitude can be studied. This leads to the small 1 means that the numerical solution have overstimated the experimental
discrepancies observed with those models when compared with IH-Foam values in amplitude, however all of them are close to 1, which means a
and DualSPHysics. good agreement. It can be observed that a better agreement is obtained
The agreement between experimental free-surface oscillations and with SPH simulations than with time-domain model. On the other hand,
numerical results can be quantified considering two statistical parame- lower phase values are achieved when increasing the resolution. The
ters; amplitude AF and phase PF: resolution with dp ¼ 0.01 m was shown to be accurate enough to obtain a
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi good comparison with experimental data without the need of increasing
u N ,
uX  X N the number of particles, so that, the computational runtime.
AF ¼ t Finum 2
ðFexp
2
i Þ (14) In order to compare the computational runtimes of IH-Foam and
i¼1 i¼1 DualSPHysics, the same case was simulated with both models using the
same spatial resolution (cell size of 0.01 m in IH-Foam and the initial
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
, ffi
u N particle spacing of 0.01 m in DualSPHysics). IH-Foam was executed on
uX  X N
PF ¼ t num
Fi  Fi exp 2
ðFi Þexp 2
(15) CPU Intel X5550 at 2.66 GHz using 8 threads and DualSPHysics on Tesla
i¼1 i¼1 K40 (DualSPHysics runs on GPU cards). The simulation of 100 s of

where Fi is the magnitude to be analysed, num refers to numerical values,

exp to experimental values and N is the number of samples. A perfect Table 2
agreement between signals would result in AF→1 and PF→0. Statistical comparison between the positions of the free-surface measured in the experi-
ment and calculated by the time-domain, IH-Foam and SPH models.
A convergence study of DualSPHysics has been carried out where
different resolutions are used to simulate the same test and the numerical AF PF
free-surface elevations inside the chamber are compared with the Time-domain 1.08 0.27
experimental ones. Different initial dp (initial interparticle distance) IH-Foam 1.08 0.19
values were simulated to obtain numerical results with different resolu- SPH dp ¼ 0.020 m 1.04 0.18
SPH dp ¼ 0.010 m 1.05 0.17
tions: dp ¼ 0.02 m led to 20,000 particles (the simulation of 100s of
SPH dp ¼ 0.005 m 1.06 0.15

17
A.J.C. Crespo et al. Coastal Engineering 126 (2017) 11–26

Fig. 5. Numerical domain of Mutriku coast and OWC using real dimensions.

physical time of the OWC case took 1 h 40 min with IH-Foam and about performed to study the wave dynamics and the movement of the water
2 h in DualSPHysics. Therefore, IH-Foam is slightly more efficient than inside the OWC chamber. The numerical tank consists of a piston to
DualSPHysics, while DualSPHysics accuracy slightly outperforms IH- generate the typical waves of that coast, an area of wave propagation
Foam when using this moderate resolution. following the realistic depth of the area and the OWC device at the end.
Several considerations were taken into account in the initial design of
4. Cases of study the numerical tank:

The SPH model is now applied to study free-surface oscillations inside - 2-D behaviour is assumed when simulating only one of the OWC
the chamber of a real onshore OWC located at the Northern Spanish coast chambers.
and later inside a floating offshore OWC located at open sea. Note that - The length of the tank is large enough to guarantee the proper wave
these cases will be simulated using real dimensions, large domains and generation and propagation (2–3 times L, being L the wavelength).
long-time events, but the GPU computing of DualSPHysics makes - The dimensions of the structure and the chamber are the real ones.
possible the execution of millions of particles at a reasonable time. - The initial water height inside the chamber is the mean sea level.
- The numerical piston-type wave-maker generates regular waves ac-
4.1. Onshore OWC in Mutriku (Spain) cording to wave theory presented in section 2.4.1 using the wave
heights and periods measured in that area.
The wave power plant in Mutriku was launched in July 2011 by the - The active wave absorption system (AWAS) described in 2.4.2 is used
EVE (Ente Vasco de Energía) and it is located at the northern Spanish to absorb reflected waves at the wave-maker.
coast (43.3073 N 357.6144 E). The plant consists of 16 chambers that
constitute the body of the vertical breakwater. The potential is 18.5 kW Fig. 5 shows the dimensions of the numerical tank. The channel has a
installed by camera, giving a total of 296 kW. The turbines are Wells length of 215 m, being 200 m the distance between the piston and the
type and their main feature is that, due to their geometry, always rotate OWC. The still water level at the piston location correspond to a water
in the same direction regardless of the direction of airflow in depth of d ¼ 10 m. The bottom part of Fig. 6 also shows the dimensions of
the chamber. the chamber.
In this section, DualSPHysics is used to simulate the OWC of Mutriku To generate the numerical system with DualSPHysics, an initial par-
where wave propagation and the interaction with the chambers are key ticle spacing of dp ¼ 0.05 m has been used, leading to a total number of
to study performance and energy harnessing. The air pressure inside the particles in the system of 737,000. Following Altomare et al. (2015a), dp
chamber is essential to reproduce the full behaviour of the device. should be chosen to guarantee more than 8–10 particles are allocated in
However, the pneumatic part of the OWC is not modelled here, but only the wave height. The physical time simulated in the numerical cases
the hydrodynamic interaction. As we mentioned above, only a phase (Table 3) is 150 s, which means that around 12–17 waves (depending on
(water) was considered in our simulations. Several simulations are the wave period and wave celerity) are propagated to interact with the

18
A.J.C. Crespo et al. Coastal Engineering 126 (2017) 11–26

Fig. 6. Comparison of numerical (blue line) and theoretical (red line) wave surface elevation and orbital velocities with test M5 in a large tank of constant depth.

OWC chamber. Computational runtime of each execution using the GPU

Table 3
Different wave conditions in Mutriku coast.
Tesla K40 is over 14 h.
As mentioned before, the piston moves to generate waves with typical
Height Period Depth Wavelength Relative Wave theory
values of height and period in that coast. Consulting data from Puertos del
(H) (T) (d) (L) d (d/L)
Estado (www.puertos.es), the mean wave height is lower than 2 m, with
Test M1 0.8 m 10 s 10 m 92.4 m 0.108 Stokes, 2nd
wave periods that vary from 8 to 12 s. The reference case will be the one
order
Test M2 1.2 m 10 s 10 m 92.4 m 0.108 Stokes, 2nd
with the values of high occurrence, so that, wave height H ¼ 0.8 m and
order period T ¼ 10 s. However, Table 3 presents the thirteen different cases of
Test M3 1.6 m 10 s 10 m 92.4 m 0.108 Stokes, 2nd regular waves that will be performed; same wave height (0.8 m) but
order different periods (8, 8.5, 9, 9.5, 10 s), same wave period (10 s) but
Test M4 2.0 m 10 s 10 m 92.4 m 0.108 Stokes, 2nd
different wave heights (0.8, 1.2, 1.6, 2.0, 2.4 m) and different sea levels
order
Test M5 2.4 m 10 s 10 m 92.4 m 0.108 Stokes, 2nd considering the mean sea level (MSL), low tide (2.25 m, d ¼ 7.75 m)
order and high tide (þ2.25 m, d ¼ 12.25 m). Note that the area is characterised
Test M6 0.8 m 8s 10 m 70.9 m 0.141 Stokes, 2nd
by a mesotidal regime and the maximum and minimum sea levels
order correspond to spring tides. On the other hand, all values of relative depth
Test M7 0.8 m 8.5 s 10 m 76.4 m 0.131 Stokes, 2nd (d/L) indicate that waves belong to intermediate depths and the Stokes
order 2nd order wave theory is used to determine the wave kinematics for
Test M8 0.8 m 9s 10 m 81.7 m 0.122 Stokes, 2nd
all tests.
order
Test M9 0.8 m 9.5 s 10 m 87.1 m 0.115 Stokes, 2nd A possible procedure to determine the wave energy resource at a
order given location is provided by L
opez et al. (2016). The omnidirectional
Test M1 0.8 m 10 s 10 m 92.4 m 0.108 Stokes, 2nd composition of the wave energy resource in terms of wave heights and
order
periods, as well as the occurrence and energy content of each interval are
Test M10 0.8 m 10 s 7.75 m 82.7 m 0.094 Stokes, 2nd presented in the form of combined scatter and energy diagrams. After-
order wards, in order to reduce the computational cost, L
opez et al. (2016)
Test M11 0.8 m 10 s 8.88 m 87.8 m 0.101 Stokes, 2nd
defined a double criterion to screen possible wave cases. The authors
order
Test M1 0.8 m 10 s 10 m 92.4 m 0.108 Stokes, 2nd selected the cases considering the wave cases with the highest percentage
order of the total energy, and those with the highest percentage of time.
Test M12 0.8 m 10 s 11.12 m 96.6 m 0.115 Stokes, 2nd Before analysing the interaction between the incident waves and the
order
OWC, wave propagation is validated by comparing numerical values with
Test M13 0.8 m 10 s 12.25 m 100.6 m 0.122 Stokes, 2nd
order
theoretical ones. A modified wave tank is used with constant depth
(d ¼ 10 m, the maximum depth in the Mutriku configuration) and this
Bold values refer to the same test conditions.

19
A.J.C. Crespo et al. Coastal Engineering 126 (2017) 11–26

Fig. 7. Comparison of numerical wave surface elevation inside the OWC chamber with (blue line) and without (red line) AWAS system for test M5 (H ¼ 2.4 m, T ¼ 10 s) in Mutriku.

Table 4 extra energy into to system that will bias the results. This will affect
Statistical comparison between free-surface and orbital velocities computed finally the oscillation of the free-surface inside the OWC chamber. Thus,
theoretically and calculated by DualSPHysics for test M5.
an active wave absorption system (AWAS) is implemented to absorb the
AF PF reflected waves at the piston to mimic the behaviour of an open sea
η 0.92 0.13 where reflected waves propagate outside the simulated domain. The
VX 0.91 0.14 wave conditions of test M5 (H ¼ 2.4 m, T ¼ 10 s) are simulated using the
VZ 0.99 0.13 numerical setup of Mutriku (Fig. 5). In this case, reflected waves are
created once the wave train hits the external wall of the OWC. Fig. 7
tank is large enough to avoid reflection during the 100 s of physical time. shows the numerical wave surface elevation in the middle of the OWC
Hence, test M5 (Table 3) is chosen to prove the wave propagation, but chamber with and without AWAS with test M5. It can be observed that,
similar results are obtained for the rest of the cases. Fig. 6 shows the when AWAS is activated, the free-surface oscillations are regular cycles,
comparison of free-surface elevation values measured at 115 m of the so that a steady regular pattern is obtained inside the OWC chamber.
piston (WG in Fig. 5) during the SPH simulation with theoretical values at Wave height and wave period are computed for the time series with their
the same location following Stokes 2nd order theory. In the same way, standard deviation. Since only regular waves have been modelled, the
numerical and theoretical values of orbital (horizontal and vertical) ve- oscillations inside the chamber are expected to have a regular pattern as
locities at the same position (x ¼ 115 m, z ¼ 3 m) are also compared in well. Values of H ¼ 6.38 ± 0.09 m and T ¼ 10.01 ± 0.01 s computed
Fig. 7. Additionally, the same statistical analysis shown in Table 2 is here inside the chamber are obtained using AWAS, while without the ab-
reproduced to compare numerical results with theoretical ones. Ampli- sorption system, H ¼ 6.25 ± 0.56 m and T ¼ 9.92 ± 0.23 s show higher
tudes (close to 1) and phases (close to 0) are presented in Table 4 for free- deviations, so that, waves are less harmonic. Thus, starting from now, all
surface elevations and orbital velocities. These results show the good results are obtained using AWAS system.
agreement between SPH and theory, so waves are being properly Different instants of the simulation using the numerical channel
generated and propagated. (Fig. 5) and waves of test M1 (H ¼ 0.8 m, T ¼ 10 s) are depicted in Fig. 8.
Once wave generation and propagation have been proven to be ac- Waves propagate and interact with the OWC of Mutriku and oscillations
curate, the thirteen cases presented in Table 3 were executed with in the free-surface elevation outside the chamber are also propagated
DualSPHysics. The OWC will reflect part of the wave energy, creating inside due to the movement of the fluid through the entrance of the
reflected waves that will travel back towards the wave-maker. This will device. The oscillations of the surface in the chamber are affected by the
create re-reflection that should be prevented in order to get a regular compression and decompression of the trapped air. Therefore, these
wave train with only incident (and reflected) component as it happens in simulations are not completely realistic because air is not modelled. As a
the real case. The re-reflection in fact can introduce spurious waves and first step, only the fluid phase is being solved, since the multi-phase

Fig. 8. Different instants of the simulation with test M1 (H ¼ 0.8 m, T ¼ 10 s) in Mutriku.

20
A.J.C. Crespo et al. Coastal Engineering 126 (2017) 11–26

Fig. 9. Free-surface oscillations inside the OWC chamber of Mutriku for different wave conditions (different wave heights, wave periods and sea levels).

version of the code needs to be adapted to simulate the air phase to obtain operator) of the oscillating water mass is maximised, however the
more reliable results. Nonetheless, the hydrodynamic part of the OWC dependence is clearly not linear. L
opez et al. (2014) also
problem can be numerically studied and the free-surface oscillations in- concluded that there is not a linear influence of wave period, but
side the OWC chamber are computed for different incident waves note that they considered the air phase in the RANS-VOF model so
(Table 3), which is also useful to identify the most favourable cases in damping values coming from pressure and flow rate were also
terms of the wave energy that can be converted by the device. included in their analysis.
Wave conditions in Table 3 have been simulated to compute the free- iii) Same wave period and height but different sea levels (lower
surface elevation inside the OWC chamber and Fig. 9 shows the results panel); it can be observed how the high tide leads to low free-
for the following waves: surface oscillations inside the OWC in this case. In any case,
different water levels means slightly different wavelengths for the
i) Same wave period but different heights (upper panel of the same period and, as shown in previous case (ii), the movement of
figure); as expected higher wave heights lead to higher oscilla- the fluid through the chamber depends on the ratio between
tions inside the chamber for waves of the same wavelength. wavelength and chamber dimensions (such as the depth the col-
ii) Same wave height but different periods (middle panel); the umn is open to the sea below the water line). The decrease of free-
maximum free-surface elevation inside the chamber is obtained surface oscillations inside the chamber are also influenced by the
with the intermediate value of wave period, so there is no a direct fact that when still water level is raised, waves are travelling
relation between the incident wave periods and the resulting os- farther in vertical from the submerged entrance of the chamber.
cillations in the chamber. In fact, the movement of the fluid
through the chamber depends on the inner size of the chamber in Fig. 9 showed the surface elevation inside the chamber for different
comparison with the wavelength. The size of the chamber de- waves and results depicted in Fig. 10 correspond to the maximum am-
termines the natural frequency of the system. On the other hand, plitudes (A) for the thirteen different incident waves reported in Table 3.
the dimensions of the opening certainly affect the oscillation The black bar indicates the same reference case (H ¼ 0.8 m, T ¼ 10 s). In
because of diffraction that is strictly related with the wave period. the first plot, the amplitude is always increasing because there is not any
The influence of the natural frequency on the response inside an PTO damping modelled. In the second plot, the shape of the curve is
oscillating water column device was also demonstrated in Wang clearly related to the RAO of the water mass inside the chamber. In the
and Falzarano (2013); in detail, for natural frequencies close to third plot, the effect of the submergence of the chamber entrance is
the incident wave frequencies the RAO (response amplitude observed. It can be now concluded that waves of the same wave height

21
A.J.C. Crespo et al. Coastal Engineering 126 (2017) 11–26

Fig. 10. Maximum amplitudes of the free-surface elevations inside the OWC chamber of Mutriku for different wave conditions.

(H ¼ 0.8 m) and different wave periods will lead to different efficiency opening of 1 m tall. A square floating structure is connected to the device
and the mentioned no linear behaviour is also observed for different to improve the stability in floatation (section 2.4.3). This floating object
depths or sea levels (SL) relative to the MSL. Waves with different is secured to the bottom of the sea by four moored lines and it could be
wavelength (different wave period or different depth) can determine exposed to extreme wave conditions forcing the moorings to work in
more or less favourable interactions. extreme regimes. Forces of catenary moorings are added to the total force
of the floating structure as described in section 2.4.4. The mooring lines
4.2. Offshore floating OWC in the open sea have a total length of 14 m and a wet volumetric density of 7850 kg/m3,
mimicking the properties of steel wire.
As mentioned before, sea waves propagating towards the coast suffer The initial particle spacing dp is equal to 0.15 m so that a total number
from attenuation, refraction, and shoaling as they approach the shore. In of 3,700,000 particles have been created to represent the case under
this way, some of the wave power is lost when onshore devices capture study (being 3,600,000 fluid particles, 2000 floating particles repre-
wave energy at the shoreline. Floating devices located in the open sea can senting the OWC and arms and the rest the boundaries of the system). As
be a good option but they still have to demonstrate their ability to mentioned before Altomare et al. (2015a) concluded that at least 8–10
generate power and survive in the most extreme conditions of the ocean. particles should represent the wave height H. A physical time of 60 s was
DualSPHysics is here applied to simulate these floating devices in loca- simulated during 51 h using the Tesla K40 GPU card.
tions with highly non-linear wave movement, severe waves, wave Regular no-breaking waves were simulated in order to obtain pre-
breaking, and overtopping. Once again, only the hydrodynamic inter- liminary results of free-surface oscillations inside an offshore floating
action between the floating body and waves is analysed while the OWC in the open ocean. Therefore, the wave conditions are H ¼ 1.8 m,
pneumatic part of the OWC is not simulated in this first stage. T ¼ 9.0 s, d ¼ 10 m with L ¼ 81. In addition, the 3-D numerical tank was
The setup of the numerical domain can be observed in Fig. 11 where designed to mimic the real situation in the open sea. Therefore, any type
the dimensions of the tank and the structure can be observed. The ge- of wave reflection should be avoided:
ometry and size of the floating OWC have been chosen for computational
demonstration purposes rather than as the geometry and size of a realistic - Passive wave absorption in form of dissipative beach is introduced at
device. The channel has a width of 12.8 m and a length of 190 m, being the end of the tank.
57 m the distance between the piston and the OWC. The maximum depth - Active wave absorption system (AWAS) is used to absorb any re-
is 10 m with a flat region of 80 m and then the beach (8 degrees of flected wave energy coming from the impact with the external front
slope). The floating OWC device is 3 m tall and 2.2 m wide with an wall of the floating OWC.

Fig. 11. Floating OWC moored to the sea bottom; numerical domain of wave tank and dimensions of the device.

22
A.J.C. Crespo et al. Coastal Engineering 126 (2017) 11–26

- Open periodic boundaries (Crespo et al., 2015) are used instead of Fig. 12 plots different instants (a-f) of the simulation. Colour of the
lateral solid walls to avoid friction that could difficult the wave surface indicates the velocity field. By modelling only regular no-
propagation and to avoid any small reflection in the direction breaking waves with the use of an active wave absorption system, a
perpendicular to the incoming wave direction. regular pattern of surface oscillations inside the chamber is achieved.

Fig. 12. Different instants of the simulation with floating OWC in the open sea.

Fig. 13. Free-surface oscillations inside the floating OWC chamber.

23
A.J.C. Crespo et al. Coastal Engineering 126 (2017) 11–26

Fig. 13 represents the wave surface elevation measured in the middle results are realistic. As it happens in nature, no reflection should come
point of the chamber, where a certain repeatability is observed. Starting from the numerical frontiers or from the numerical piston. The inexis-
from 20 s, regular cycles are measured in the chamber. Note that the tence of re-reflected wave is easier to check when using regular waves
shaded area of the figure tries to identify one of these cycles, which also instead of irregular wave trains. This is the reason way a regular train was
corresponds to the instants (a-f) shown in Fig. 12. The achieved regular considered although DualSPHysics allows different wave generations as
pattern of free-surface oscillations inside the OWC chamber proves that mentioned above.

Fig. 14. Time histories of the motions of the floating OWC (surge, sway and heave motion, and roll, pitch and yaw rotation).

Fig. 15. Time histories of the numerical tensions in the four mooring lines.

24
A.J.C. Crespo et al. Coastal Engineering 126 (2017) 11–26

The numerical time histories of the motions of the floating structure “Programa de Consolidacio
n e Estructuracio

n de Unidades de Inves-
are shown in Fig. 14. The surge and heave components present simple tigaci

on Competitivas (Grupos de Referencia Competitiva) GRC2013-
harmonic oscillations, while small sway motion in the lateral direction 001” and under project “NUMANTIA ED431F 2016/004”. The work is
can be seen. This is thanks to the role of the catenary moorings that limit also funded by the Ministry of Economy and Competitiveness of the
the displacement of the floating structure. In this way, the OWC is always Government of Spain under project “WELCOME ENE2016-75074-C2-1-
aligned with the direction of the incident waves (roll and yaw angles are R”. The authors gratefully acknowledge the support of NVIDIA Corpo-
negligible). Moreover, the motions are reduced in order to get a correct ration with the donation of the GTX Tesla K40 GPU used for this research.
vertical position of the water column inside the chamber which will The authors are especially thankful to IH Cantabria for the experi-
facilitate the operation of the OWC to maximise the energy. In addition, mental data.
the numerical tensions at mooring lines can be also computed with
DualSPHysics. Fig. 15 shows the values of tensions at the four lines (two References
at the front and two at the back of the structure). It can be observed how
the tensions experienced at both front lines are the same (and at both Altomare, C., Crespo, A.J.C., Rogers, B.D., Domínguez, J.M., Gironella, X., G
omez-
back lines as well) and the maximum tension registered during the Gesteira, M., 2014. Numerical modelling of armour block sea breakwater with
Smoothed Particle Hydrodynamics. Comput. Struct. 130, 34–45.
simulation are in the order of 75 tons for this device and wave conditions. Altomare, C., Crespo, A.J.C., Domínguez, J.M., G
omez-Gesteira, M., Suzuki, T.,
Verwaest, T., 2015a. Applicability of Smoothed Particle Hydrodynamics for
5. Conclusions and future work estimation of sea wave impact on coastal structures. Coast. Eng. 96, 1–12.
Altomare, C., Suzuki, T., Domínguez, J.M., Barreiro, A., Crespo, A.J.C., G
omez-Gesteira,
M., 2015b. Numerical wave dynamics using Lagrangian approach: wave generation
The SPH model is presented in this work as a numerical tool capable and passive & active wave absorption. Proceedings of the 10th SPHERIC
of simulating the hydrodynamic interaction between waves and offshore International Workshop, Parma, Italy.
Alves, M.A., Costa, I.R., Sarmento, A.J.N., Chozas, J.F., 2010. Performance evaluation of
floating OWC devices. SPH is a meshless numerical model that deals an axysimmetric floating OWC. Proceedings of the 20th International Offshore and
easily with high non-linearities and large deformation of the free surface. Polar Engineering Conference, Beijing, China.
DualSPHysics is implemented to exploit the parallel computing power of Babarit, A., Delhommeau, G., 2015. Theoretical and numerical aspects of the open source
BEM solver NEMOH. Proceedings of the 11th European Wave and Tidal Energy
GPUs so the millions of particles required to simulate real complex ge-
Conference (EWTEC2015), Nantes, France.
ometries and domains can be simulated at a reasonable computational Barreiro, A., 2015. Smoothed Particle Hydrodynamics Model for Civil and Coastal
time. These features allow SPH models to simulate floating WEC devices Engineering Applications [PhD Thesis]. Universidade de Vigo, Spain.
under the action of extreme waves, including wave breaking and over- Barthel, V., Mansard, E.P., Sand, S.E., Vis, F.C., 1983. Group bounded long waves in
physical models. Ocean. Eng. 10 (4), 261–294.
topping, in order to study their survival capacity and efficiency to absorb Batchelor, G.K., 1974. Introduction to Fluid Dynamics. Cambridge University Press.
the available wave energy. Bouscasse, B., Colagrossi, A., Marrone, S., Antuono, M., 2013. Nonlinear water wave
Three numerical experiments were designed to show the capabilities interaction with floating bodies in SPH. J. Fluids Struct. 42, 112–129.
Brito-Melo, A., Hofmann, T., Sarmento, A., Cl
ement, A.H., Delhommeau, G., 2001.
of the model. In the first experiment, which is 2D, the movement of water Numerical modelling of OWC-shoreline devices including the effect of surrounding
inside a chamber is analysed. The model has shown to reproduce prop- coastline and non-flat bottom. Int. J. Offshore Polar Eng. 11, 147–154.
erly the water elevation observed in experiments at least for those cases Combourieu, A., Lawson, M., Babarit, A., Ruehl, K., Roy, A., Costello, R., Weywada, P.L.,
Bailey, H., 2015. WEC3: Wave Energy Converters modeling Code Comparison
where changes in pressure inside the chamber are negligible. The top project. Proceedings of the 11th European Wave and Tidal Energy Conference
opening of the chamber made the pneumatic part of the problem negli- (EWTEC2015), Nantes, France.
gible for this test, and the pressure in the OWC air chamber was equal to Canelas, R.B., Domínguez, J.M., Crespo, A.J.C., G
omez-Gesteira, M., Ferreira, R.M.L.,
2015. A Smooth Particle Hydrodynamics discretization for the modelling of free
the atmospheric pressure. In addition, both accuracy and numerical ef-
surface flows and rigid body dynamics. Int. J. Numer. Methods Fluids 78, 581–593.
ficiency are shown to be similar to those obtained with state-of-the-art Colagrossi, A., Landrini, M., 2003. Numerical simulation of interfacial flows by smoothed
Eulerian models (IH-Foam). The second experiment, which is also 2D, particle hydrodynamics. J. Comput. Phys. 191, 448–475.
Crespo, A.J.C., G
omez-Gesteira, M., Dalrymple, R.A., 2007. Boundary conditions
analyses the free-surface oscillations inside a real OWC chamber located
generated by dynamic particles in SPH methods. CMC Comput. Mater. Continua 5
in Mutriku (Spain). The model is shown to be able to deal with problems (3), 173–184.
at real scale, not only at laboratory scale like the previous one. In addi- Crespo, A.J.C., Domínguez, J.M., Barreiro, A., G
omez-Gesteira, M., Rogers, B.D., 2001.
tion, the AWAS module included in DualSPHysics package is also tested. GPUs, a new tool of acceleration in CFD: efficiency and reliability on Smoothed
Particle Hydrodynamics methods. PLoS ONE 6 (6), e20685.
The module is shown to satisfactorily remove the effect of re-reflected Crespo, A.J.C., Domínguez, J.M., Rogers, B.D., G
omez-Gesteira, M., Longshaw, S.,
waves in the numerical domain. This fact is extremely important when Canelas, R., Vacondio, R., Barreiro, A., García-Feal, O., 2015. DualSPHysics: open-
simulating long lasting wave trains. Finally, the third experiment shows source parallel CFD solver on smoothed particle hydrodynamics (SPH). Comput.
Phys. Commun. 187, 204–216.
that the model works properly for 3D cases. A floating OWC device is Cummis, W.E., 1962. The impulse response function and ship motions. Schiffstechnik 9,
designed including a structure with arms that helps in the floatation and 101–109.
keeps the vertical position of the water column inside the chamber. In Delaur
e, Y.M.C., Lewis, A., 2003. 3D hydrodynamics modelling of a fixed oscillating
water column wave power plant by a boundary element method. Ocean. Eng. 30,
addition, the behaviour of mooring lines is also simulated. Catenary 309–330.
moorings help to reduce the heave, surge and roll motion of the floating Didier, E., Neves, M.G., 2001. A semi-infinite numerical wave flume using smoothed
device and play a key role to ensure the durability and economic viability particle hydrodynamics. Int. J. Offshore Polar Eng. 22 (3), 193–199.
Domínguez, J.M., Crespo, A.J.C., G
omez-Gesteira, M., 2013. Optimization strategies for
of offshore energy converters. As we mentioned above, the ease to deal CPU and GPU implementations of a smoothed particle hydrodynamics method.
with moving objects (moored or not) located inside the fluid is one of the Comput. Phys. Commun. 184 (3), 617–627.
main advantages of Lagrangian models such as SPH. Domínguez, J.M., Crespo, A.J.C., Valdez-Balderas, D., Rogers, B.D., G
omez-Gesteira, M.,
2013. New multi-GPU implementation for Smoothed Particle Hydrodynamics on
The present version of DualSPHysics still presents some limitations.
heterogeneous clusters. Comput. Phys. Commun. 184, 1848–1860.
On the one hand, the used version of the model only has a phase (water) Domínguez, J.M., Crespo, A.J.C., Cerc
os-Pita, J.L., Fourtakas, G., Rogers, B.D., Vacondio,
while air pressure is not simulated. Consequently, pneumatic conversion R., 2015. Evaluation of reliability and efficiency of different boundary conditions in
of the OWC mechanism is not considered. Future work will be focused on an SPH code. In: Proceedings of the 10th SPHERIC International Workshop, Parma,
Italy.
the simulation of the air phase inside the OWC chamber using a new Drew, B., Plummer, A.R., Sahinkaya, M.N., 2009. A review of wave energy converter
version of DualSPHysics that considers water-air phases (Mokos technology. J. Power Energy 223, 887–902.
et al., 2015). DNV, 2008. SESAM Users Manual, WADAM. http://www.sciencedirect.com/science/
article/pii/S0306261914003638-b0200.
Edge, B., Gamiel, K., Dalrymple, R.A., Herault, A., Bilotta, G., 2014. Application of gpusph
Acknowledgment to design of wave energy. Proceedings of the 9th SPHERIC International Workshop,
Paris, France.
This work was partially financed by Xunta de Galicia under project

25
A.J.C. Crespo et al. Coastal Engineering 126 (2017) 11–26

Evans, D.V., 1978. The oscillating water column wave energy convertors. J. Inst. Math. its L

opez, I., Pereiras, B., Castro, F., Iglesias, G., 2016. Holistic performance analysis and
Appl. 22, 423–433. turbine induced damping for an OWC wave energy converter. Renew. Energy 85,
Evans, D.V., 1982. Wave power absorption by systems of oscillating surface pressure 1155–1163.
distributions. J. Fluid Mech. 114, 481–499. Luo, Y., Wang, Z., Peng, G., Xiao, Y., Zhai, L., Liu, X., Zhang, Q., 2014. Numerical simulation
Falc~
ao, A.F.O., 2000. The shoreline OWC wave power plant at the Azores. Proceedings of of a heave-only floating OWC (oscillating water column) device. Energy 76, 799–806.
the 4th European Wave Energy Conf, Aalborg, Denmark. Madsen, O.S., 1971. On the generation of long waves. J. Geophys. Res. 76 (36),
Falc~
ao, A.F.O., 2010. Wave energy utilization: a review of the technologies. Renew. 8672–8683.
Sustain. Energy Rev. 14, 899–918. Marongiu, J.C., Leboeuf, F., Caro, J., Parkinson, E., 2010. Free surface flows simulations
Falc~
ao, A.F.O., Henriques, J.C.C., 2016. Oscillating-water-column wave energy converters in pelton turbines using an hybrid SPH-ALE method. J. Hydraulic Res. 48, 40–49.
and air turbines: a review. Renew. Energy 85, 1391–1424. McCormick, M.E., 1974. Analysis of a wave energy conversion buoy. J. Hydronautics 8
Falnes, J., 2002. Ocean Waves and Oscillating Systems: Linear Interactions Including (3), 77–82.
Wave-energy Extraction. Cambridge University Press. McCormick, M.E., 1976. Modified linear-analysis of a wave-energy conversion buoy.
Faltinsen, O.M., 1993. Sea Loads on Ships and Offshore Structures. Cambridge University Ocean. Eng. 3 (3), 133–144.
Press. Mokos, A., Rogers, B.D., Stansby, P.K., Domínguez, J.M., 2015. Multi-phase SPH
Fekken, G., 2004. Numerical Simulation of Free Surface Flow with Moving Rigid Bodies modelling of violent hydrodynamics on GPUs. Comput. Phys. Commun. 196,
[PhD Thesis]. University of Groningen, Netherlands. 304–316.
Goda, Y., 2010. Random seas and design of maritime structures. Adv. Ser. Ocean Eng. – Monaghan, J.J., 1992. Smoothed particle hydrodynamics. Annu. Rev. Astronomy
Volume 33 third ed. Astrophysics 30, 543–574.
G
omez-Gesteira, M., Dalrymple, R.A., 2004. Using a 3D SPH method for wave impact on a Monaghan, J.J., 1994. Simulating free surface flows with SPH. J. Comput. Phys. 110,
tall structure. Journal of Waterway Port. Coast. Ocean Eng. 130 (2), 63–69. 399–406.
Gotoh, H., Shao, S., Memita, T., 2004. SPH-LES model for numerical investigation of wave Monaghan, J.J., 2005. Smoothed particle hydrodynamics. Rep. Prog. Phys. 68,
interaction with partially immersed breakwater. Coast. Eng. J. 46 (1), 39–63. 1703–1759.
Heath, T.V., Whittaker, T.J.T., Boake, C.B., 2000. The design, construction and operation Monaghan, J.J., Kos, A., 1999. Solitary waves on a cretan beach. J. Waterway Port, Coast.
of the LIMPET wave energy converter (Islay, Scotland). Proceedings of the 4th Ocean Eng. 125 (3), 145–154.
European Wave Energy Conf, Aalborg, Denmark, pp. 42–47. Monaghan, J.J., Kos, A., Issa, A., 2003. Fluid motion generated by impact. Journal of
Heath, T.V., 2012. A review of oscillating water column. Philos. Trans. R. Soc. A 370, Waterway Port. Coast. Ocean Eng. 129, 250–259.
235–245. Morris-Thomas, M.T., Irvin, R.J., Thiagarajan, K.P., 2007. An investigation into the
Iturrioz, A., Guanche, R., Armesto, J.A., Alves, M.A., Vidal, C., Losada, I.J., 2014. Time- Hydrodynamic efficiency of an oscillating water column. J. Offshore Mech. Arct. Eng.
domain modeling of a fixed detached oscillating water column towards a floating 129, 273–278.
multi-chamber device. Ocean. Eng. 76, 65–74. Paix~ao Conde, J.M., Gato, L.M.C., 2008. Numerical study of the air-flow in an oscillating
Iturrioz, A., Guanche, R., Lara, J.L., Vidal, C., Losada, I.J., 2015. Validation of OpenFOAM water column wave energy converter. Renew. Energy 33, 2637–2644.
for oscillating water column three-dimensional modelling. Ocean. Eng. 107, Rafiee, A., Elsaesser, B., Dias, F., 2013. Numerical simulation of wave interaction with an
222–236. oscillating wave surge converter. Proceedings ASME 32nd International Conference
Johanning, L., Smith, G.H., Wolfram, J., 2006. Mooring design approach for wave energy on Ocean, Offshore and Arctic Engineering.
converters. Proc. Institution Mech. Eng. Part M J. Eng. Marit. Environ. 220 (4), Ren, B., He, M., Dong, P., Wen, H., 2015. Nonlinear simulations of wave-induced motions
159–174. of a freely floating body using WCSPH method. Appl. Ocean Res. 50, 1–12.
Johanning, L., Smith, G.H., Wolfram, J., 2007. Measurements of static and dynamic Sarmento, A., Falc~ao, A.F.O., 1985. Wave generation by an oscillating surface-pressure
mooring line damping and their importance for floating WEC devices. Ocean. Eng. and its application in wave-energy extraction. J. Fluid Mech. 150, 467–485.
34, 1918–1934. Shaffer, H.A., Klopman, G., 2000. Review of multidirectional active wave absorption
Josset, C., Cl
ement, A.H., 2007. A time-domain numerical simulator for oscillating water methods. J. Waterw. Port, Coast. Ocean Eng. 126, 88–97.
column wave power plants. Renew. Energy 32 (8), 1379–1402. Teixeira, P.R.F., Davyt, D.P., Didier, E., Ramalhais, R., 2013. Numerical simulation of an
Kamath, A., Bihs, H., Arntsen, Ø.A., 2015a. Numerical investigations of the oscillating water column device using a code based on Navier–Stokes equations.
hydrodynamics of an oscillating water column. Ocean. Eng. 102, 40–50. Energy 61, 513–530.
Kamath, A., Bihs, H., Øivind, Arntsen, Ø.A., 2015b. Numerical modeling of power takeoff Thiebaut, F., O'Sullivan, D., Kracht, P., Ceballos, S., Lopez, J., Boake, C., Bard, J.,
damping in an Oscillating Water Column device. Int. J. Mar. Energy 10, 1–16. Brinquete, N., Varandas, J., Gato, L.M.C., Alcorn, R., Lewis, A.W., 2011. Testing of a
Khayyer, A., Gotoh, H., 2009. Wave impact pressure calculations by improved SPH floating OWC device with movable guide vane impulse turbine power take-off.
methods. Int. J. Offshore Polar Eng. 19, 300–307. Proceedings of the 9th European Wave Tidal Energy Conference, Southampton, U.K.
Lawson, M., Garzon, B., Wendt, F., Yu, Y., Michelen, C., 2015. COER hydrodynamics Torre-Enciso, Y., Ortubia, I., L
opez de Aguileta, L.I., Marqu
es, J., 2009. Mutriku Wave
modeling competition: modeling the dynamic response of a floating body using the Power Plant: from the thinking out to the reality. Proceedings of the 8th European
WEC-SIM and FAST simulation tools. In: Proceedings of OMAE 2015. St. John’s, Wave and Tidal Energy Conference, Uppsala, Sweden, pp. 319–329.
Newfoundland, Canada. Violeau, D., 2012. Fluid Mechanics and the SPH Method: Theory and Applications. Oxford
Lee, E.-S., Violeau, D., Issa, R., Ploix, S., 2010. Application of weakly compressible and University Press.
truly incompressible SPH to 3-D water collapse in waterworks. J. Hydraulic Res. 48, WAMIT, 2012. WAMIT Users Manual.
50–60. Wang, H., Falzarano, J.M., 2013. Energy extraction from the motion of an oscillating
Leimkuhler, B.J., Reich, S., Skeel, R.D., 1996. Integration Methods for Molecular Dynamic water column. Ocean. Syst. Eng. 3 (4), 327–348.
IMA Volume in Mathematics and its Application. Springer. Wendland, H., 1995. Piecewiese polynomial, positive definite and compactly supported
Liu, Z., Frigaard, P., 2001. Generation and Analysis of Random Waves. Laboratoriet for radial functions of minimal degree. Adv. Comput. Math. 4 (1), 389–396.
Hydraulik og Havnebygning, Instituttet for Vand, Jord og Miljoeteknik, Aalborg Whittaker, T.J.T., McPeake, F.A., 1985. Design optimization of axi-symmetric tail tube
Universitet. buoys. In: Evan, D.V., Falc~ao, A.F. de O. (Eds.). Hydrodynamics of Ocean Wave
L
opez, I., Pereiras, B., Castro, F., Iglesias, G., 2014. Optimisation of turbine-induced Energy Utilization Symposium, Lisbon, Porgutal.
damping for an OWC wave energy converter using a RANS-VOF numerical model. Yeylaghi, S., Moa, B., Beatty, S., Buckham, B., Oshkai, P., Crawfoed, C., 2015. SPH
Appl. Energy 127, 105–114. Modeling of Hydrodynamic Loads on a Point Absorber Wave Energy Converter Hull.
L
opez, I., Iglesias, G., 2014. Efficiency of OWC wave energy converters: a virtual Proceedings of the 11th European Wave and Tidal Energy Conference (EWTEC2015),
laboratory. Appl. Ocean Res. 44, 63–70. Nantes, France.
L
opez, I., Pereiras, B., Castro, F., Iglesias, G., 2015. Performance of OWC wave energy Zhang, Y., Zou, Q.P., Greaves, D., 2012. Air–water two phase flow modelling of
converters: influence of turbine damping and tidal variability. Int. J. Energy Res. 39, Hydrodynamic performance of an oscillating water column device. Renew. Energy
472–483. 41, 159–170.

26