Energy 296 (2024) 130888 Contents lists available at ScienceDirect Energy journal homepage: www.elsevier.com/locate/energy High-fidelity numerical modelling of a two-WEC array with accurate implementation of the PTO system and control strategy using DualSPHysics Nicolas Quartier a ,∗, Timothy Vervaet a , Gael Verao Fernandez a , José M. Domínguez b , Alejandro J.C. Crespo b , Vasiliki Stratigaki a , Peter Troch a a Ghent University, Technologiepark 60, 9052 Ghent, Belgium b Environmental Physics Laboratory (EPhysLab), CIM-UVIGO, Universidade de Vigo, Spain ARTICLE INFO Keywords: Smoothed particle hydrodynamics DualSPHysics Power take-off system Project chrono WECfarm project Heaving wave energy converter ABSTRACT In order for Wave Energy Converters (WECs) to become economically viable they will have to be installed closely together in so-called WEC arrays. When WECs are placed closely together they influence each other’s motion, which is challenging to implement in a numerical model. The accurate modelling of WECs in WEC arrays is crucial for their further development into cost-effective renewable energy converters. This paper presents the study of numerical modelling of an array of two heaving WECs (a two-WEC array) placed in close proximity of each other in order to increase their interaction effect. The numerical model uses DualSPHysics, a Smoothed Particle Hydrodynamics (SPH) based software to model wave-WEC interaction, coupled to Project Chrono for the modelling of the Power Take-Off (PTO) system and the friction. First, a series of numerical system identification tests are simulated: a free-decay test of an isolated WEC, radiation tests and an excitation test. During the radiation test one WEC is moved by an external force with a chirp-up noise signal, while the other WEC is kept still. This allows the computation of the impedance matrix, and thereby also the hydrodynamic coefficients of the two-WEC array, which has not been done before in SPH. Results were validated with experimental data from the WECfarm test campaign. Secondly, a series of numerical dynamic tests are simulated of the two-WEC array in regular and irregular waves. In the experimental test campaign a friction force caused by the motor and gearbox was counteracting the WEC’s motion, so a friction compensation (FC) force was added to cancel out this friction. This friction and FC force was numerically modelled by adding a novel implementation in Project Chrono. Furthermore, a novel global control strategy was implemented using Project Chrono, in which the PTO force on each WEC depends on the velocity of both WECs. Overall, good to very good agreement was found in terms of heave and free surface elevation when comparing numerical to experimental results. 1. Introduction Wave energy is a potential source of clean electricity that can make a significant contribution to the de-carbonization of the world’s energy supply. Wave energy is considered as the most concentrated and least variable form of renewable energy that can generate power up to 90% of the time, compared to 25% for solar and wind devices [1]. A wide variety of devices to harvest wave energy, known as wave energy converters (WECs), have been designed over the last decades [1]. However, none of them have yet achieved the level of technological development required to become economically viable and generate electricity commercially [2,3]. In the current research focus is laid on the heaving point absorber, one of the most investigated types of WECs [4]. Such a heaving point absorber typically consist of a floating buoy moved by the waves and connected to a Power TakeOff (PTO) system, which converts the WEC’s movement into electricity. In order to assess the performance and survivability of WECs, which are necessary for exploiting wave energy, the related wave-induced hydrodynamic forces and WEC motions have to be examined. Physical experiments are essential and very useful, nevertheless, they have very high costs and require a high level of expertise [5]. On the other hand, numerical methods have become very popular in last years [6], mainly due to the unprecedented growth of the computational resources available. A complete review on the numerical methods to simulate the hydrodynamic response of point absorbers can be found in [7]. ∗ Corresponding author. E-mail addresses: Nicolas.Quartier@UGent.be (N. Quartier), Timothy.Vervaet@UGent.be (T. Vervaet), Gael.VeraoFernandez@UGent.be (G.V. Fernandez), jmdominguez@uvigo.es (J.M. Domínguez), alexbexe@UVigo.es (A.J.C. Crespo), Vicky.Stratigaki@UGent.be (V. Stratigaki), Peter.Troch@UGent.be (P. Troch). https://doi.org/10.1016/j.energy.2024.130888 Received 25 April 2023; Received in revised form 22 January 2024; Accepted 1 March 2024 Available online 14 March 2024 0360-5442/© 2024 Elsevier Ltd. All rights reserved. Energy 296 (2024) 130888 N. Quartier et al. revealing significant hydrodynamic interaction between the WECs and the floating wind turbine. In the work of [46] the impact of long-term variations in wave resource around Iceland on the power generation of a WEC array of 45 WECs was studied. In all of these studies however, the WEC’s motion and power absorption is computed based on linear potential flow theory, thereby assuming low amplitude motions. In this work a two-WEC array is modelled in DualSPHysics, where the two WECs are placed in close proximity to each other. As validation, data from the experimental test campaign in Aalborg University is used, resulting from the work of [47]. Furthermore, an accurate implementation for the representation of the Coulomb and viscous based drivetrain (motor, gearbox and rack and pinion) friction model present in the experimental tests is added using the DualSPHysics– Project Chrono coupling [22]. The novelty of this work lies in the simulation of a WEC array in SPH, which to our knowledge has not been done before. Furthermore, an accurate model of the friction present in the experiments is implemented, and a new global control strategy to maximize the power absorption of the WEC array was numerically added. The benefit of using SPH, and specifically DualSPHysics, for WEC array modelling lies in the fact that this method does not assume small amplitudes, which leads to underestimations of the induced force and overestimations of the average absorbed power. Also, nonlinear friction forces, PTO system forces or global control strategies can be implemented in DualSPHysics in a straightforward way. The simulations of the WEC array allow the calculation of the impedance matrix, which has not yet been performed before in SPH. Moreover, this work shows that DualSPHysics has the ability to not only accurately compute the WEC’s motion, but also the free surface elevation near the WEC array. The paper is organized as follows: Section 2 describes the basic theoretical principles of SPH and its implementation in DualSPHysics; Section 3 gives an overview of the experimental campaign carried out at the University of Aalborg (AAU) and used for validation in the current research; Sections 4 and 5 show the main results obtained using DualSPHysics and compared to experimental data from [47], and finally Section 6 presents an overview of the conclusions. Most studies concerning WECs employ potential flow theory based on the linearized form of the Navier–Stokes equations. Applying linear potential flow theory allows numerical modelling of WECs in the time or frequency domain [8], enabling fast calculations of the WEC’s motion. This theory however needs to assume small amplitude oscillations of the WEC and the fluid to be incompressible, inviscid and with an irrotational motion. These assumptions lead to an underestimation of the wave-induced forces on WECs under highly-nonlinear sea states [9]. These restrictive assumptions are not applied in high-fidelity models such as the CFD (computational fluid dynamics) methods. The most commonly used CFD methods in hydrodynamics are meshbased. Different point absorbers have been studied numerically using these methods in [10,11]. Despite being very robust mathematically and computationally, mesh-based methods face important challenges when capturing the free surface and the rapidly-evolving nonlinearities. Due to their ability to overcome these drawbacks, meshless CFD methods have gained attention during the last years, the smoothed particle hydrodynamics (SPH) method being the most widely used [12– 14]. In contrast with the mesh-based methods, in SPH the fluid is discretized in a set of points, named particles, that move with the velocity calculated from the Navier–Stokes equations carrying all physical properties with them [15]. Its meshless lagrangian approach makes the SPH method a very interesting alternative when simulating freesurface flows with wave–structure interaction, such as the case at hand. A complete overview of the use of SPH in ocean and coastal engineering is given in [16]. In [17] the hydrodynamic response of a point absorber was computed using SPH and a finite volume solver. The works of [18,19] exploited the capabilities of the SPH to model nonlinearities to study the interaction between a point absorber WEC and extreme waves. In this research, the open-source software DualSPHysics [20] (available at www.dual.sphysics.org) is employed to model the motion of two heaving WECs placed in close proximity to each other. Furthermore, DualSPHysics has been coupled [21,22] with Project Chrono [23], enabling the effect of the PTO system to be included in the simulations. This software has proven to be a valuable tool in the modelling of wave–structure interaction in general and of floating WECs in particular — see [3,19,24–27]. The application of SPH to industrial problems is one of the SPHERIC Grand Challenges [28] and, in particular, the simulation of WECs is one of the most promising application fields [29]. The studies of [30– 32] were pioneers presenting SPH simulations of oscillating wave surge devices. The SPH method has also been applied for the modelling of Oscillating Water Column (OWC) WECs [33–37]. The works of [18,38] were the first ones to deal with extreme waves interacting with pointabsorbers using SPH. More recently, heaving WECs were simulated in DualSPHysics in [3,19,27]. In [39] a linear damping and Coulomb damping PTO system was applied to a heaving WEC, in [3] a linear spring–damper system was used as PTO system for a heaving WEC, in [36] a PTO system of an OWC WEC was modelled, and in [40] a moored heaving WEC with a linear spring–damper PTO system was modelled. In these cases however, only a single WEC was considered. In order for WECs to become economically viable they will have to be installed in large numbers grouped together in WEC arrays [2,41]. When WECs are positioned close to each other they can affect each other’s motion due to the modified wave field surrounding an oscillating WEC. It is known that the type of PTO system also has an influence on these surrounding modified wave fields, as studied in [42]. Furthermore, the optimal PTO system configuration can depend on the motion of a neighbouring WEC and therefore be different than the optimal PTO configuration for an isolated WEC [43]. WEC arrays and the control of WEC arrays have been numerically studied before: in the work of [44] the performance of an array of two and three WECs was studied, showing an increased efficiency when motion constraints are applied. Zhou et al. [45] studied multiple WECs placed closely together around a floating wind turbine are studied, 2. Smoothed particle hydrodynamics—DualSPHysics The DualSPHysics solver applies the SPH method, which is a meshless lagrangian method in which the fluid is discretized in a set of particles for which the position, velocity, density and pressure are computed by solving the Navier–Stokes equations and by interpolation of the values of neighbouring particles. The contribution depends on the distance between the particle of interest and the neighbouring particles, with ππ the initial inter-particle distance, and on the Kernel function π . This Kernel has an area of influence defined by the smoothing √ length β, which is a function of ππ (β = ππππβ 3ππ in 3D) — see [20]. The mathematical description of the SPH method and DualSPHysics specifically is based on the description given in [20]. 2.1. Governing equations The Navier–Stokes equations written in their SPH notation are solved each timestep for each of the particles. The SPH-based continuity and momentum equations are given in Eqs. (1) and (2), respectively. In the following equations the velocity and density of particle π is calculated, with π each of its neighbouring particles: ∑ π πππ ∑ = ππ (ππ − ππ )∇π πππ + πΏβππ πππ ⋅ ∇π πππ π (1) ππ‘ ππ π π ) ( ∑ πππ ππ + ππ =− ππ ∇π πππ + π + π (2) ππ‘ ππ ππ π with ππ the mass of particle π, ππ the velocity vector of particle π, ππ the pressure of particle π, ππ the density of particle π, π the dissipation term and π the gravity acceleration. πππ represents the Kernel function and depends on the distance between particles π and π. In this work a 2 Energy 296 (2024) 130888 N. Quartier et al. Quintic Kernel [48] is applied, since this type of Kernel is well suited for general free surface problems — see [49]. One of the main advantages of the Quintic Kernel is that the tensile instability that appears using other kernels such as the Cubic Spline, is avoided using the kernel adopted in this work. More information can be found in the work of [49]. The diffusion term introduced in the right hand side of the continuity equation (Eq. (1)) acts as a numerical noise filter thereby improving the numerical stability and smoothing the density and pressure field — see [50–52]. In this paper the recently proposed density diffusion term of [53] is applied, since it has proven to produce more accurate results for the pressure field near boundaries while keeping the computational cost limited. The magnitude of the diffusion term is controlled with the parameter πΏ, set to 0.1 in this study. The artificial dissipation term πππ takes the following form: π· πππ (3) πππ = 2(ππ· π − ππ )β β βπππ β in [58]. For each boundary particle of mDBC a πβππ π‘ ππππ is mirrored into the fluid. Boundary particles then receive the properties of the fluid at the position of the ghost node. The density is calculated using a first order consistent SPH interpolation, as proposed in [59]. The mDBC method has proven to show results with more realistic physical values for the pressure at the boundaries and a significant reduction in the size of the gap between fluid and boundary without a significant extra computational cost. Throughout this paper mDBC will be applied for the modelling of the WECs. 2.3. Floating bodies DualSPHyics has the capability to accurately simulate fluid-driven objects as described and validated in [26,60], which will be used extensively in this paper. The force per unit mass acting on one boundary particle π of the floating body is calculated by summing up the contributing forces exerted on this boundary particle π by fluid particles π within the compact support of the Kernel: ∑ π ππ (8) ππ = where πππ = ππ −ππ with ππ the position of particle π and ππ· the dynamic density, equal to the difference of the total (ππ ) and hydrostatic (ππ» ) density: ππ· = ππ − ππ» . π represents the dissipation term, which is either the artificial viscosity scheme proposed in [15] or the laminar viscosity scheme with a Sub-Particle Scale (SPS) large eddy simulation model as proposed in [54,55] (Eq. (5)). In this research the laminar viscosity scheme with SPS model is used, with the dissipation term π in Eq. (2) written as: π = π0 ∇2 ππ + 1 ∇ ⋅ ππππ π Eq. (4) contains a laminar viscosity term written as: ∑ 4π0 πππ ⋅ ∇π πππ ππ π0 ∇2 ππ = (ππ + ππ )(π2ππ + π 2 ) π π with π π the force per unit mass acting on boundary particle π of the floating body and π ππ the force per unit mass acting on boundary particle π exerted by fluid particle π, calculated with Eq. (2). Once the force acting on the floating body is computed, the body’s motion can be determined assuming it is rigid: ∑ ππ½ = ππ π π (9) π ππ‘ π ππ΄ ∑ πΌ ππ (ππ − πΉ) × π π (10) = ππ‘ π (4) (5) with π the body’s total mass, πΌ the moment of inertia, π½ the velocity, π΄ the rotational velocity, πΉ the centre of mass, ππ the mass of floating boundary particle π and ππ the position of floating particle π. Eqs. (9) and (10) are integrated over time in order to compute the velocity ππ of the floating particle: and a Sub-Particle Scale (SPS) model term written as: ∑ β ππππ − ππππ β 1 β ∇π π ∇ ⋅ ππππ = ππ β β ππ ππ β π ππ π π β β (6) with ππ the velocity vector of particle k, π 2 = 0.1 h2 , and π0 is the kinematic viscosity of the fluid (10−6 m2 βs for water). Since the fluid is weakly-compressible in DualSPHysics, an equation of state is used to calculate the fluid pressure as function of the density instead of solving a Poisson-like equation: [( ) ] π 2 π0 ππ πΎ −1 (7) ππ = π πΎ π0 ππ = π½ + π΄ × (ππ − πΉ) (11) 2.4. Coupling DualSPHysics–Project Chrono Project Chrono is an open-source software package that enables the numerical modelling of mechanical constraints and collision between objects [23]. It has recently been successfully coupled to DualSPHysics [22] and can be used in this case for the modelling of the PTO system of a WEC. Already implemented in the coupling was the linear damping, which is applied here for the simulation with a linear damping PTO system — see Eq. (12) with πΎ = 7 the polytropic constant, ππ the numerical speed of sound and π0 the reference fluid density. The speed of sound ππ is artificially lowered such that a reasonable time step can be employed while ensuring that density variations are kept lower than 1% during the simulation. πΉπ π π,π (π‘) = −πΆπ π π,π π£(π‘) (12) with πΆπ π π,π the linear PTO system damping coefficient and π£(π‘) the WEC’s heave velocity. The DualSPHysics–Project Chrono coupling has been extended with the Coulomb damping model in [39], applying the force described in Eq. (13), with πΆπ π π,π the PTO damping coefficient for a Coulomb damping PTO system. 2.2. Boundary conditions In DualSPHysics the boundaries are described by a set of particles for which the same equations (Eqs. (1) and (2)) as for fluid particles are solved. However, the particles belonging to the boundary do not move according to the forces acting on them: boundary particles remain either fixed or move according to a predefined movement. These boundary conditions are called Dynamic Boundary Conditions (DBC) and have the advantage of being able to deal with complex geometries and being computationally efficient [56,57]. However, due to excessive repulsive forces near the boundary between a structure and the fluid, a gap appears of the order of magnitude of the smoothing length β — see [58]. Furthermore at this same boundary unphysical values of the pressure are observed. These issues are dealt with in the modified DBC (mDBC) implementation recently added to DualSPHysics, as described πΉπ π π,π (π‘) = −πΆπ π π,π π πππ(π£(π‘)) (13) In this work the DualSPHysics–Project Chrono coupling is further extended by (i) adding an accurate implementation of the drivetrain friction and (ii) adding a new control strategy scheme. 3. Experimental campaign During the WECfarm project two heaving WECs were built and tested in the wave basin of AAU, Denmark [47,61]. A realistic rendering 3 Energy 296 (2024) 130888 N. Quartier et al. Fig. 1. Rendering of the ‘‘WECfarm’’ WEC. Source: Figure adopted from [61]. of the WEC model and its components is shown in Fig. 1. Note however that in the numerical simulations and results the positive heave is defined opposite as the definition shown in Figs. 1 and 2, where positive heave is defined when the WEC is below the still water level (SWL). Among the tests are: system identification tests (radiation and excitation tests), drivetrain friction assessment tests, free decay tests and power absorption tests with PTO system and control strategies. The hydrodynamic part of the WEC is an Acrylonitrile-ButadieneStyrene (ABS) thermofolded truncated cylindrical buoy with a Polymethylmethacrylate PMMA plate on top — see Fig. 1. The WEC buoy is connected to guide shafts and a rack, the latter being connected to a pinion. Three OAV (OAV Air Bearings, Princeton, NJ, USA) 40 mm air bushings are installed to provide zero friction between the guide shafts and bushings. A PTO system force πΉπ π π is applied to the WEC buoy with a Beckhoff Permanent-Magnet Synchronous Motor drive (PMSM) connected with a rack and pinion via a gearbox with ratio 1:4. The gearbox break-away torque π01 is expected to be around 0.20–0.40 N m, with convergence towards lower values for a longer operational lifetime. The motor static friction ππ is reported to be equal to 0.02 Nm. As a result, the total Coulomb friction experienced by the WEC will be mainly determined by π01 . Moreover, additional viscous friction attributed to the drivetrain will occur. The presence of the gearbox and motor induced a friction force acting on the WEC, therefore in some tests a friction compensation force was added to compensate this friction force. This friction compensation is also explained in Section 5. The experimental test campaign was carried out in a water depth of 1.010 m. Tests were performed for an isolated WEC and for a twoWEC array where the two WECs were placed next to each other with an interdistance of two diameters from centre to centre, equivalent to 1.20 m. The motor drive receives real-time position and velocity values from a single-turn absolute encoder. A detailed explanation of the experimental test campaign is given in [47,61]. Eight wave gauges (WGs) were placed in the wave basin to measure the free surface elevation (FSE) — see Fig. 3. Table 1 gives an overview of the experimental test cases that were numerically simulated and validated in the current work. The test case id is kept the same as in the experimental test campaign [47]. Note that laboratory scale is used in the present simulations, however SPH can be also applied to full-scale modelling. The considerations about Fig. 2. Dimensions of the WECfarm heaving WEC [m]. Source: Figure adopted from [61]. scaling and potential scale effects (such as surface tension effects, or viscous scale effects) will be the same as the ones considered in physical modelling or any other CFD tool. 4. System identification tests A series of system identification (SID) tests are carried out in DualSPHysics for which the results are compared to experimental data. Agreement between numerical and experimental tests is expressed in terms of an index of agreement ππ defined in Eq. (14): β§ ππ΄πΈ βͺ1 − πππ΄π· ππ = β¨ πππ΄π· βͺ ππ΄πΈ − 1 β© if ππ΄πΈ ≤ πππ΄π· (14) if ππ΄πΈ > πππ΄π· with scaling factor c = 2. MAE is the mean absolute error defined by: 1 ∑| | ππ΄πΈ = (15) |ππ − ππ | π π and MAD the mean absolute deviation: 1 ∑| | ππ΄π· = |ππ − π| | | π π 4 (16) Energy 296 (2024) 130888 N. Quartier et al. Fig. 3. Wave gauges layout during the experimental test campaign — dimensions in cm. Source: Figure adopted from [47]. Table 1 Description of experimental test cases, numerically simulated in the current work. Test case id Test case description Detailed description Test 256 Test 254 Test 247 Test 282 Test 224 Test 225 Test 226 Test 238 Test 90 Test 186 Radiation test Radiation test Excitation test Regular wave test Regular wave test Regular wave test Regular wave test Focussed wave test Irregular wave test Irregular wave test WEC A imposed motion, WEC B fixed Both WECs imposed motion π» = 0.16 m, π = 1.5 s, WEC A fixed π» = 0.16 m, π = 1.5 s, WEC A without motor π» = 0.16 m, π = 1.5 s, both WECs with friction π» = 0.16 m, π = 1.5 s, both WECs with FC π» = 0.16 m, π = 1.5 s, both WECs with PTO system π»π = 0.20 m, ππ = 1.1 s, both WECs without FC and PTO system π»π = 0.16 m, ππ = 1.5 s, WEC A with FC and PTO system π»π = 0.16 m, ππ = 1.5 s, both WECs with FC and PTO system with global control Table 2 Rating of index of agreement ππ [63]. ππ Rating 0.90–1.00 0.80–0.90 0.70–0.80 0.50–0.70 0.30–0.50 (−1.00)–0.30 Excellent Very good Good Reasonable/Fair Poor Bad Table 3 Convergence study for a simulation of the free-decay test, for 16 s physical time with the initial inter-particle distance ππ, number of particles πππππ‘πππππ , runtime and index of agreement ππ , run on a NVIDIA A100-SXM4. with ππ the predicted SPH output at time π, and ππ te experimental output at time π [62]. A classification of index of agreement is given in Table 2, as defined in [62,63]. Besides the index of agreement, two other non-dimensional error estimators are given: the linear correlation coefficient π and the error in amplitude π΄. The linear correlation coefficient quantifies the error in phase [26]: 1 ∑ π (ππ − π )(ππ − π) π π = (17) ππ ππ √∑ √ √ π π2 π΄ = √∑ π (18) 2 π ππ ππ [m] πππππ‘πππππ Runtime [h] ππ 0.05 0.025 0.02 0.38E6 2.64E6 5.02E6 0.12 1.60 3.67 0.83 0.91 0.95 of the water, which means that bottom slamming will occur. This freedecay test with an isolated WEC is performed without attaching the WEC to the motor. The only possible friction in the experiment could come from the airbushings, but this friction was minimal. The initial inter-particle distance ππ for the free-decay test was 0.025 m, decided after a convergence study where the agreement between the heave from DualSPHysics and experiment was compared for three different interparticle distances — see Fig. 4 and Table 3. The index of agreement ππ increases from 0.91 to 0.95 when reducing ππ from 0.025 m to 0.02 m, but to keep the computational runtime within reasonable limits, a ππ of 0.025 m is chosen. 4.2. Radiation test The closer R and A are to 1, the better the agreement between DualSPHysics and experiment. Finally, the relative root square error (RRSE) is calculated as well as: √∑ √ √ (ππ − ππ )2 π π ππΈ = √ ∑π (19) 2 π (ππ − π ) During a first radiation test, Test 256 as defined in [47], πΉπ π π is imposed on one of the WECs, in this case on WEC A, in a still basin without waves while WEC B is kept fixed. The numerical basin used during the radiation tests has a width of 6 metres and has numerical damping layers on all sides, as is shown in Fig. 5. A ππ of 0.025 m was used, based on the convergence test analysis (Fig. 4). The velocity of WEC A and the force on WEC B caused by the radiated waves surrounding WEC A are measured, allowing the calculation of the impedance matrix Z describing the relationship between the force FΜ and velocity vΜ in the frequency domain: ππ , π , π΄ and π π ππΈ are summarized for all studied testcases in Table 5. 4.1. Free-decay test During the free-decay test a single WEC (WEC A) was lifted 0.214 m above the water and then released, while its heave motion π§ was measured. Note that during this test the WEC was lifted completely out FΜ = ZvΜ 5 (20) Energy 296 (2024) 130888 N. Quartier et al. Fig. 4. Convergence analysis for a free-decay test of 1 WEC. Fig. 5. Top view of the numerical wave basin used in DualSPHysics for the radiation tests of the WEC array. or: [ ] [ πΉΜπ πππ = πΉΜπ πππ ][ ] πππ π£Μ π πππ π£Μ π frequencies from 0.4 to 1.75 Hz are covered in the input πΉπ π π . Note that from Eq. (21) only the coefficient πππ and πππ can be determined, since π£Μ π equals zero. To determine the remaining coefficients of the impedance matrix a second test should be carried out during which WEC A is kept still and WEC B is heaving. However, it is considered that the impedance matrix is symmetrical, with πππ = πππ and πππ = πππ . The index of agreement for the velocity of WEC A and the heave force on WEC B is 0.87 and 0.82 respectively. All complex quantities are obtained by calculating the discrete Fourier transform from their real time series signal. Fig. 8 shows good agreement between the experimentally and numerically determined magnitude and phase of the impedance. Furthermore, the magnitude and phase of the impedance determined with the BEM solver Capytaine [64] (based on NEMOH, (21) with πΉΜπ the complex PTO force imposed on WEC A, πΉΜπ the complex force acting on WEC B measured with loadcells, π£Μ π the complex velocity of WEC A and π£Μ π the complex velocity of WEC B, which is in this specific case equal to zero since WEC B is kept fixed. The input PTO force πΉπ π π imposed on WEC A in both the experiment and DualSPHysics is a chirp-up noise signal, which is a polychromatic sinusoidal signal with a constant amplitude, covering a range of frequencies — see Figs. 6 and 7. This input πΉπ π π signal shown in Fig. 6 is the input force before application of the friction compensation. From Fig. 7 it can be seen that 6 Energy 296 (2024) 130888 N. Quartier et al. Fig. 6. Radiation test 256 with a noise (chirp-up) signal. Fig. 7. Frequency of the chirp-up input force signal πΉπ π π over time in radiation test 256. [65]) are shown. It proves that DualSPHysics is able to accurately capture the interaction between both WECs. From the impedance matrix Z the added mass and hydrodynamic damping matrices can be deduced by taking the real and imaginary part of the impedance: Z(π) = B(π) + π(π(M + A(π)) − Kβ(π)) [ ] [ ] πππ πππ π΅ππ π΅ππ = πππ πππ π΅ππ π΅ππ β‘ β’ [ β’ β ππ +π β’π β β’ β 0 β’ β β’ β£ ] [ 0 π΄ππ + ππ π΄ππ ] β π΄ππ β − π΄ππ β β two reasons: (1) an additional inertia induced by the motor, gearbox and rack and pinion, and (2) a significant drag force effect due to large relative velocity differences between the WEC and the fluid at resonance, similarly as was found in [66]. Fig. 9 shows that the real part of πππ – which corresponds to the hydrodynamic damping π΅ππ in Eq. (22) – is underestimated by Capytaine. This can be related to the absence of viscosity in the BEM solver, resulting in a lower hydrodynamic damping. A similar conclusion was made in [3]. The underestimation of the hydrodynamic damping with the BEM solver cannot be dedicated to the absence of friction, since in the experiment and DualSPHysics friction compensation was modelled, cancelling the friction coming from the motor, gearbox and rack and pinion. The difference in hydrodynamic damping π΅ππ (= π πππ΄π΄ ) between Capytaine and DualSPHysics is around 30 Ns/m at resonance, which means the drag coefficient πΆπ of the WEC can be estimated at 0.46 using the method mentioned in [39]. Similarly as was found in [3], there is an underestimation of the real part of πππ in DualSPHysics at low frequencies, however, this underestimation was not present for πππ . A second radiation test, Test 254, was carried out in which both WECs were excited by a PTO force. During this test, the velocities of both WECs were measured. In Fig. 10 very good agreement between the velocity from DualSPHysics and experiment is shown, with an index of agreement ππ of 0.86 and 0.88 for WEC A and WEC B respectively. Also the free surface elevation (FSE) measured in DualSPHysics at WG7 agrees well with the FSE from the experiment. The non-linear effects resulting in steeper crests and flatter troughs are well captured in DualSPHysics. (22) [ πΎπ 0 ]β€ 0 β₯ πΎπ β₯ β₯ β₯ π β₯ β₯ β¦ (23) with π΅ππ the hydrodynamic damping, π΄ππ the added mass, π the WEC’s mass, πΎ the hydrostatic stiffness and π and π either π or π. In Fig. 9 the real and imaginary parts of the matrix components of Z are shown, computed with DualSPHysics and with the BEM solver Capytaine, and compared to the experimental results. The resonance frequency is the frequency at which the heave velocity π£ is in phase with the excitation force πΉπ§ , so at resonance the phase and the imaginary part of πππ should be zero. In the experiment and DualSPHysics this happens at f = 0.93 Hz, corresponding to a resonance period of 1.08 s, whereas in Capytaine this happens at f = 1.02 Hz, corresponding to a resonance period of 0.98 s. This shift in resonance frequency can be explained by 7 Energy 296 (2024) 130888 N. Quartier et al. Fig. 8. Magnitude and phase of the components of the impedance matrix determined in radiation test 256. Fig. 9. Real and imaginary part of the components of the impedance matrix determined in radiation test 256. In Fig. 11 a visual comparison between experiment and DualSPHysics is displayed, showing a similar pattern of radiated waves around the WECs. Furthermore, the vertical velocity of the fluid particles is shown, showing significant interaction between the radiated waves of both WECs. good agreement (ππ = 0.89) with the experimental FSE. The WEC is placed at 4.6 m from the piston, which is the same distance as in the experiments. This distance is sufficient since it is recommended to place the WEC at least one wavelength away from the piston. For the excitation tests again the laminar viscosity model is applied in DualSPHysics. Excellent agreement (ππ = 0.93) is found when comparing the experimental with the numerical πΉπ — see Fig. 13, where the FSE at the wave gauge between the two WECs (WG7 in experiment, see Fig. 3) as well as the πΉπ on WEC A is shown. The force on only one of the WECs is shown since the case is symmetrical and therefore πΉπ is identical for both WECs. 4.3. Excitation test Excitation tests are carried out by keeping the WEC(s) fixed while a surface elevation is imposed. This allows measurement of the heave force πΉπ acting on the WEC(s). In the experimental campaign this πΉπ was measured with loadcells. Similarly as in the experimental campaign, this excitation test is carried out for the two WECs together. The numerical basin has a width of 3 metre, with periodic boundaries and numerical damping on the sides reducing the lateral reflection — see Fig. 12. The numerical basin has a beach with numerical damping to prevent reflection from the back. Regular waves with a wave height (π») of 0.16 m and a wave period (π ) of 1.5 s, resulting in a wavelength (πΏ) of 3.36 m, are numerically generated with a piston, moving with the same motion as in the experimental campaign. The wave generation and propagation was tested in an empty numerical basin, showing very 5. Dynamic tests of the WEC array During the dynamic tests the WEC array is simulated in regular, irregular and focussed waves. The same numerical wave basin as shown in Fig. 12 is applied, while the inter-particle distance ππ is set to 0.02 m after a convergence analysis. An overview of the order of simulations of the two WECs in regular waves is represented in Fig. 14. First, the WECs are simulated without any friction or PTO system (Project Chrono is not used), which corresponds to the experimental 8 Energy 296 (2024) 130888 N. Quartier et al. Fig. 10. Radiation test 254 with a noise (chirp-up) signal. Table 4 Convergence study for a simulation of Test 282, for 50 s physical time with the interparticle distance ππ , number of particles πππππ‘πππππ , runtime and index of agreement ππ , run on a NVIDIA A100-SXM4. ππ [m] πππππ‘πππππ Runtime [h] ππ 0.025 0.020 0.017 1.72E6 3.24E6 5.23E6 5.06 11.88 19.08 0.89 0.92 0.93 which the motor was attached, no friction compensation was applied in the experiment. This means that the WEC will feel the full friction force. Initially the friction force added in DualSPHysics was as described in Eqs. (24) and (25), however it was found that this lead to a heave motion in DualSPHysics that was significantly lower than in the experiment. Since there was very good agreement between DualSPHysics and experiment in Test 282, it was concluded that the disagreement in Test 224 is caused by an incorrect friction force implemented in DualSPHysics. Therefore it was decided to tune the Coulomb damping coefficient in the friction force in DualSPHysics until the heave motion in DualSPHysics matched the experimental heave motion of both WECs in Test 224. Finally, the following expression was found for the friction force for WEC A and WEC B: Test 282, where no motor is present. The only possible friction in the experiment could come from the airbushings, but this friction was minimal. After Test 282 the WEC array is simulated with friction in order to determine the correct Coulomb and linear damping terms of the friction force caused by the gearbox and motor. This corresponds to Test 224 in the experiments. Once the correct friction force is determined, Test 225 is simulated during which friction compensation (FC) was added. Finally, FC and a PTO system is added on both WECs, as was done in Test 226. Test 282 was simulated with three different ππ values, showing an increasing index of agreement with decreasing ππ — see Table 4. Fig. 15 shows the heave from DualSPHysics with three different ππ values and the heave from the experiment. Finally, a ππ of 0.02 m was chosen, since this resulted in excellent agreement between numerical and experimental heave for WEC A while still keeping the computational runtime limited. An isolated WEC was tested in regular waves of H = 0.16 m and π = 1.5 s without PTO system. It was found from the experimental test campaign that the total friction acting on the WEC can be described as a combination of Coulomb and linear damping. Note that this description of the friction force is a simplification since the non-linear friction shows complex behaviour. From the experimental test campaign however it followed that simplifying the friction force as a combination of Coulomb and linear damping is a justified simplification. From the initial experimental tests the following expression was found for the friction force πΉπ ,π΄ and πΉπ ,π΅ acting on WEC A and WEC B respectively: (24) πΉπ ,π΅ (π‘) = −49.6 ⋅ π πππ(π£(π‘)) − 180 ⋅ π£(π‘) (25) (26) πΉπ ,π΅ (π‘) = −30 ⋅ π πππ(π£π΅ (π‘)) − 180 ⋅ π£π΅ (π‘) (27) Note that this friction force πΉπ is different for WEC A and WEC B due to small differences in the motors and gearboxes of these WECs, as was also observed in the experiments. As is clear from Fig. 16, excellent agreement is found between SPH and experiment with the updated Coulomb damping coefficient in πΉπ — the index of agreement ππ for the heave is 0.95 and 0.91 for WEC A and WEC B respectively, while ππ for the FSE at WG7 equals 0.93. The πΉπ described in Eqs. (26) & (27) is kept throughout the remainder of the simulations. In a next test, Test 225, the WEC array was tested with friction compensation. This means that in the experiments an external force was applied via the motor in order to compensate for the friction. In the experimental campaign the friction compensation force (πΉπΉ πΆ ) was as described in Eqs. (28) & (29): 5.1. WECs without PTO system πΉπ ,π΄ (π‘) = −44.8 ⋅ π πππ(π£(π‘)) − 180 ⋅ π£(π‘) πΉπ ,π΄ (π‘) = −25 ⋅ π πππ(π£π΄ (π‘)) − 180 ⋅ π£π΄ (π‘) πΉπΉ πΆ,π΄ (π‘) = 44.8 ⋅ π πππ(π£π΄ (π‘)) + 144 ⋅ π£π΄ (π‘) (28) πΉπΉ πΆ,π΅ (π‘) = 49.6 ⋅ π πππ(π£π΅ (π‘)) + 144 ⋅ π£π΅ (π‘) (29) Note that only 80% of the linear friction term was compensated in the experimental tests, since compensating for more than 80% resulted in unstable behaviour of the WEC. Test 225 was carried out in DualSPHysics by adding the πΉπΉ πΆ from Eqs. (28) and (29) to the friction force πΉπ from Eqs. (26) and (27), using the DualSPHysics–Project Chrono coupling. In the experimental test campaign the friction compensation was only applied when the absolute value of the WEC’s velocity was higher than 0.0035 m/s, in order to avoid rapid sign changes in the πΉπΉ πΆ when the WEC’s velocity was around zero. This velocity threshold The Coulomb and linear damping can be added in DualSPHysics by using the coupling with Project Chrono. In the simulation of Test 224 in 9 Energy 296 (2024) 130888 N. Quartier et al. Fig. 11. Pictures from experimental Test 254 (left column), snapshots from numerical Test 254 at the same time instants (middle column) and snapshots of the numerical vertical fluid velocity (right column). was also implemented in DualSPHysics by changing the definition of the Coulomb damping force in Project Chrono. Fig. 17 shows very good agreement between the FSE (ππ = 0.85), and the WEC’s heave motion (ππ = 0.89 for WEC A and ππ = 0.90 for WEC B) from DualSPHysics and experiment. The absorbed power π is defined in Eq. (32): π (π‘) = −πΉπ π π (π‘) ⋅ π£(π‘) When the absorbed power is calculated, an overestimation is found in DualSPHysics when the WECs have a negative velocity. A possible reason for this overestimation could be that the Coulomb damping value of the friction force of the motor in the experiment has a different value for negative and positive velocities. 5.2. WECs with PTO system Next, a linear damping PTO system force πΉπ π π is added to both WECs, with πΉπ π π depending on the WEC’s heave velocity: πΉπ π π,π΄ (π‘) = −πΆπ π π,π΄π΄ ⋅ π£π΄ (π‘) (30) πΉπ π π,π΅ (π‘) = −πΆπ π π,π΅π΅ ⋅ π£π΅ (π‘) (31) (32) 5.3. Focussed wave test Before simulating the WEC array in irregular waves, the WEC array is simulated in a focussed wave. A focussed wave test contains information in a range of frequencies without requiring a large computational time. Focussed waves have been modelled in DualSPHysics before [40]. The focussed wave has a significant wave height of 0.2 m and peak period of 1.1 s, and the WECs are modelled without FC or PTO system damping so the full friction force is added. The ππ used in the simulation was again equal to 0.02 m. Comparison of the heave In this case, πΆπ π π,π΄π΄ and πΆπ π π,π΅π΅ both equal 300 Ns/m, as in the experimental Test 226. Fig. 18 shows very good to excellent agreement between experimental and DualSPHysics results for the FSE at WG7 (ππ = 0.92) and the heave of WEC A (ππ = 0.93) and WEC B (ππ = 0.89). 10 Energy 296 (2024) 130888 N. Quartier et al. Fig. 12. Lateral and top view of the numerical wave basin used in DualSPHysics for the excitation and dynamic tests of the WEC array. Fig. 13. Excitation test (Test 247): FSE at WG7 from experiment and SPH (above), and vertical wave excitation force πΉπ on WEC A from experiment and SPH (below). with experimental results is shown in Fig. 19. An index of agreement ππ of 0.87 and 0.88 was found for WEC A and B respectively. is larger. This hints at an error in the estimation of the Coulomb damping coefficient in the friction force. At these low amplitudes the WEC is often not moving in the experiment, while in DualSPHysics the WEC keeps moving, so possibly the friction compensation is overestimated in DualSPHysics for this test. A possible explanation could be that the friction force present in the experiments was not constant through time, meaning that the Coulomb friction term could be varying during the same test or between different tests. The friction force could have been higher during Test 90 in the experimental test campaign. The absorbed power shown in Fig. 20 is even more overestimated in DualSPHysics, since errors in velocity are increased by a power 2 in the absorbed power. The average absorbed power in the experiment was 3.58 W, while in DualSPHysics this was 4.46 W. Next, the two-WEC array is simulated in the same irregular wave conditions. Contrary to the conventional way of describing the PTO damping force as proportional to the WEC’s own velocity, the PTO damping force now depends on both its own velocity and the velocity of 5.4. Irregular wave test Finally, the WECs are simulated in irregular waves while PTO damping is applied. A ππ of 0.02 m was used, as in the regular waves and focussed wave test. First, an isolated WEC (WEC A) is simulated in irregular waves with a Jonswap spectrum with π»π = 0.16 m and ππ = 1.5 s, with πΆπ π π,π΄π΄ = 280 Ns/m, since this is the damping coefficient that theoretically leads to maximum average absorbed power. This is the set-up applied in Test 90. The PTO force πΉπ π π,π΄ is as described in Eq. (30). An index of agreement ππ of 0.79 is found when comparing the heave of the WEC with the experimental heave, which is a lower ππ than what was obtained in previous simulations. From Fig. 20 it can be seen that at low amplitudes the overestimation of the heave in DualSPHysics 11 Energy 296 (2024) 130888 N. Quartier et al. Fig. 14. Schematic representation of order of simulations. Fig. 15. Test 282: Heave of WEC A for three different ππs in DualSPHysics, and from experiment. the neighbouring WEC (see Eqs. (33) & (34)), since it is known that the optimal PTO system force leading to maximum average absorbed power depends on the velocity of both WECs [67–69]. This coupled resistive control strategy, called global control, was also applied in Test 186 of the experiments. πΉπ π π,π΄ (π‘) = −πΆπ π π,π΄π΄ ⋅ π£π΄ (π‘) − πΆπ π π,π΄π΅ ⋅ π£π΅ (π‘) (33) πΉπ π π,π΅ (π‘) = −πΆπ π π,π΅π΅ ⋅ π£π΅ (π‘) − πΆπ π π,π΅π΄ ⋅ π£π΄ (π‘) (34) WEC A was 0.75. The average absorbed power in the experiment was 3.90 W, while in DualSPHysics it was 4.81 W (for WEC A). Using the result of Test 90, a π-value (defined in Eq. (35)) of 1.09 is obtained in the experiment and 1.08 in DualSPHysics, meaning that in both the experiment as well as in DualSPHysics an increase in average absorbed power of around 8%–9% in the two-WEC array was observed, caused by constructive interference of the radiated waves of both WECs. The WEC’s large diameter to draft ratio causes high radiated waves, which can enlarge the heave amplitude of the neighbouring WEC due to their close proximity. πΆπ π π,π΄π΄ and πΆπ π π,π΅π΅ were set to 300 N s/m and πΆπ π π,π΄π΅ and πΆπ π π,π΅π΄ were set to 35 N s/m. A schematic representation of the coupling between DualSPHysics and Project Chrono for this control strategy is shown in Fig. 21, mainly based on the scheme of [20,21] (see Fig. 22). Again, the absorbed power seems to be overestimated in DualSPHysics. The index of agreement for the heave of both WEC A and B was 0.82, while the index of agreement for the absorbed power of π= ππ πΈπΆπππππ¦ πππ ππππππ πΈπΆ (35) In Table 5 a summary is given of the agreement between DualSPHysics and experiment for all simulated testcases. This agreement is expressed with the index of agreement ππ , the error in amplitude π΄, the correlation coefficient π and the relative root square error π π ππΈ. In 12 Energy 296 (2024) 130888 N. Quartier et al. Fig. 16. Test 224: FSE at WG7, Heave of WEC A and Heave of WEC B from SPH and experiment. Fig. 17. Test 225: FSE at WG7, Heave of WEC A and Heave of WEC B from SPH and experiment. Fig. 18. Test 226: FSE at WG7, Heave of WEC A and Heave of WEC B from SPH and experiment. 13 Energy 296 (2024) 130888 N. Quartier et al. Fig. 19. Test 238: Heave of WEC A and Heave of WEC B from SPH and experiment. Fig. 20. Test 90: Heave of WEC A from DualSPHysics and experiment, and absorbed power π of WEC A from SPH and experiment. Fig. 21. Schematic representation of the coupling procedure between DualSPHysics and Project Chrono for the applied control strategy. 14 Energy 296 (2024) 130888 N. Quartier et al. Fig. 22. Test 186: FSE at WG7, Heave of WEC A and Heave of WEC B from SPH and experiment, and absorbed power π of WEC A and B from SPH and experiment. most cases, very good–excellent agreement is found, with the exception of the absorbed power in irregular waves. Table 5 ππ , π , π΄ and π π ππΈ for all testcases, simulated with laminar viscosity, mDBC and density diffusion. 6. Conclusions ππ A R RRSE Rating 0.91 0.95 0.99 0.13 Excellent 0.87 0.82 0.91 0.99 0.96 0.93 0.28 0.37 Very good Very good 0.79 0.85 0.87 1.03 1.15 1.05 0.91 0.97 0.97 0.44 0.29 0.26 Good Very good Very good 0.90 0.94 0.84 0.95 0.99 0.99 0.23 0.13 Excellent Excellent 0.93 0.92 0.99 0.16 Excellent 0.93 0.95 0.91 0.98 0.99 1.03 0.99 1.00 0.98 0.16 0.11 0.19 Excellent Excellent Excellent 0.85 0.89 0.90 1.01 0.85 0.90 0.96 0.99 0.98 0.34 0.22 0.21 Very good Very good Excellent 0.92 0.93 0.89 1.10 1.10 1.09 0.99 1.00 0.98 0.18 0.16 0.23 Excellent Excellent Very good 0.87 0.88 1.08 1.06 0.97 0.98 0.28 0.23 Very good Very good 0.79 0.72 1.20 1.23 0.94 0.91 0.44 0.54 Good Good 0.82 0.82 0.75 1.18 1.14 1.20 0.95 0.96 0.91 0.38 0.36 0.52 Very good Very good Good Free-decay test In this work a WEC array of two heaving cylindrical truncated WECs is modelled in DualSPHysics and validated with experimental data provided from the WECfarm test campaign held in the wave basin of AAU in February 2022 [47]. A series of system identification tests are performed, leading to the following observations and conclusions: z WEC A Radiation test 256 v WEC A πΉπ§ WEC B Radiation test 254 • A free-decay test with an isolated WEC was carried out in DualSPHysics, showing excellent agreement with the experimental heave of the WEC when mDBC and laminar viscosity were applied in DualSPHysics; • An excitation test with the two-WEC array was carried out in DualSPHysics, showing very good agreement with the experimental vertical excitation force; • Two radiation tests with the two-WEC array were carried out in DualSPHysics, allowing the calculation of the impedance matrix. In this work the impedance matrix for a two-WEC array is for the first time determined with SPH. Very good agreement with the experimental heave velocity was found, and with the free surface elevation at the location between the two WECs. DualSPHysics is able to accurately capture the non-linear wave field modifications surrounding the WEC array. WG v WEC A v WEC B Excitation test 247 WG πΉπ§ WEC A Regular wave test 282 without motor z WEC A Regular wave test 224 with friction WG z WEC A z WEC B Regular wave test 225 with FC WG z WEC A z WEC B In the experiments a motor is present to add a PTO force to the WEC, which also introduces friction. This friction was cancelled in experimental tests by adding a friction compensation force. It was found that this friction compensation in the experiments overcompensated the real friction. In order to add the friction force in DualSPHysics, a combination of Coulomb damping and linear damping was applied, using the coupling between DualSPHysics and Project Chrono. Furthermore, a novel implementation was made in Project Chrono that only applies the friction compensation force when the absolute value of the WEC’s heave velocity exceeded a certain threshold, exactly as was done in the experiments. Once the correct friction and friction (over)compensation was determined, a series of tests in regular and irregular waves were performed. These showed very good agreement between the experimental heave and the heave from DualSPHysics. In irregular waves however the agreement between the absorbed power from experiment and DualSPHysics was only reasonable: it was found that the absorbed power in DualSPHysics was overestimated, which could be due to a mismatch in the estimation of the friction force at low heave velocities. A novel implementation was added in the Regular wave test 226 with FC and PTO WG z WEC A z WEC B Focussed wave test 238 z WEC A z WEC B Irregular wave Test 90 z WEC A P WEC A Irregular wave Test 186 z WEC A z WEC B P WEC A DualSPHysics–Project Chrono coupling to model the global control strategy. This allowed the definition of a PTO force that depends 15 Energy 296 (2024) 130888 N. Quartier et al. on the velocity of both WECs corresponding to a causal impedance matching formulation for a two-WEC array. Both in the experiments and in DualSPHysics it was found that an increase in absorbed power of up to 9% is observed per WEC when the WECs are installed in the studied array configuration. The good agreement between numerical and experimental results proves the applicability of DualSPHysics for the modelling of a WEC array. Future research following this work will look into larger WEC arrays, e.g. a 2 × 2 or a 5-WEC array in star configuration. In order to simulate WEC arrays with more than two WECs a larger numerical basin and thus longer computational runtimes will be required. Therefore, further research looking into a multi-GPU version of DualSPHysics should be performed. Furthermore, a variable resolution scheme might be necessary to reduce the required number of particles of the numerical basin. [3] Ropero-Giralda P, Crespo AJC, Coe RG, Tagliafierro B, Domínguez JM, Bacelli G, Gómez-Gesteira M. Modelling a heaving point-absorber with a closed-loop control system using the DualSPHysics code. Energies 2021;14(3):760. http://dx.doi.org/ 10.3390/en14030760. [4] Magagna D, Monfardini R, Uihlein A. JRC ocean energy status report 2016 edition - technology, market and eco-nomic aspects of ocean energy in Europe. Publ Off Eur Union 2016;59. [5] Vervaet T, Stratigaki V, Backer BD, Stockman K, Vantorre M, Troch P. Experimental modelling of point-absorber wave energy converter arrays : A comprehensive review , identification of research gaps and design of the wecfarm setup. 2022, p. 1–31. [6] Josh D, Ronan C. Efficient nonlinear hydrodynamic models for wave energy converter design-a scoping study. J Mar Sci Eng 2020;8(1):1–65. http://dx.doi. org/10.3390/jmse8010035. [7] Li Y, Yu YH. A synthesis of numerical methods for modeling wave energy converter-point absorbers. Renew Sustain Energy Rev 2012;16(6):4352–64. http: //dx.doi.org/10.1016/j.rser.2011.11.008. [8] Folley M. Numerical modelling of wave energy converters. 2016. [9] Windt C, Davidson J, Ringwood JV. High-fidelity numerical modelling of ocean wave energy systems: A review of computational fluid dynamics-based numerical wave tanks. Renew Sustain Energy Rev 2018;93(May):610–30. http://dx.doi.org/ 10.1016/j.rser.2018.05.020. [10] Yu YH, Li Y. Reynolds-averaged Navier-Stokes simulation of the heave performance of a two-body floating-point absorber wave energy system. Comput & Fluids 2013;73:104–14. http://dx.doi.org/10.1016/j.compfluid.2012.10.007. [11] Reabroy R, Zheng X, Zhang L, Zang J, Yuan Z, Liu M, Sun K, Tiaple Y. Hydrodynamic response and power efficiency analysis of heaving wave energy converter integrated with breakwater. Energy Convers Manage 2019;195(February):1174–86. http://dx.doi.org/10.1016/j.enconman.2019. 05.088. [12] Gotoh H, Khayyer A. On the state-of-the-art of particle methods for coastal and ocean engineering. Coast Eng J 2018;60(1):79–103. http://dx.doi.org/10.1080/ 21664250.2018.1436243. [13] Manenti S, Wang D, Domínguez JM, Li S, Amicarelli A, Albano R. SPH modeling of water-related natural hazards. Water (Switzerland) 2019;11(9):1–26. http: //dx.doi.org/10.3390/w11091875. [14] Violeau D, Rogers BD. Smoothed particle hydrodynamics (SPH) for free-surface flows: Past, present and future. J Hydraul Res 2016;54(1):1–26. http://dx.doi. org/10.1080/00221686.2015.1119209. [15] Monaghan JJ. Smoothed particle hydrodynamics. Annu Rev Astron Astrophys 1992;30(1):543–74. http://dx.doi.org/10.1146/annurev.aa.30.090192.002551. [16] Luo M, Khayyer A, Lin P. Particle methods in ocean and coastal engineering. Appl Ocean Res 2021;114:102734. http://dx.doi.org/10.1016/j.apor.2021.102734. [17] Westphalen J, Greaves D, Raby A, Hu Z, Causon D, Mingham C, Omidvar P, Stansby P, Rogers B. Investigation of wave-structure interaction using state of the art cfd techniques. Open J Fluid Dyn 2013;4. http://dx.doi.org/10.4236/ojfd. 2014.41003. [18] Omidvar P, Stansby P, Rogers B. SPH for 3D floating bodies using variable mass particle distribution. Internat J Numer Methods Fluids 2013;72:427–52. http://dx.doi.org/10.1002/fld.3749. [19] Ropero-Giralda P, Crespo AJ, Tagliafierro B, Altomare C, Domínguez JM, GómezGesteira M, Viccione G. Efficiency and survivability analysis of a point-absorber wave energy converter using DualSPHysics. Renew Energy 2020;162:1763–76. http://dx.doi.org/10.1016/j.renene.2020.10.012. [20] Domínguez JM, Fourtakas G, Altomare C, Canelas RB, Tafuni A, García-Feal O, Martínez-Estévez I, Mokos A, Vacondio R, Crespo AJC, Rogers BD, Stansby PK, Gómez-Gesteira M. DualSPHysics: from fluid dynamics to multiphysics problems. Comput Part Mech 2022;9(5):867–95. http://dx.doi.org/10.1007/s40571-02100404-2. [21] Canelas RB, Brito M, Feal OG, Domínguez JM, Crespo AJ. Extending DualSPHysics with a differential variational inequality: modeling fluid-mechanism interaction. Appl Ocean Res 2018;76(April):88–97. http://dx.doi.org/10.1016/ j.apor.2018.04.015. [22] Martínez-Estévez I, Domínguez J, Tagliafierro B, Canelas R, García-Feal O, Crespo A, Gómez-Gesteira M. Coupling of an SPH-based solver with a multiphysics library. Comput Phys Comm 2023;283:108581. http://dx.doi.org/10. 1016/j.cpc.2022.108581, URL https://www.sciencedirect.com/science/article/ pii/S0010465522003009. [23] Tasora A, Anitescu M. A matrix-free cone complementarity approach for solving large-scale, nonsmooth, rigid body dynamics. Comput Methods Appl Mech Engrg 2011;200(5–8):439–53. http://dx.doi.org/10.1016/j.cma.2010.06.030. [24] Altomare C, Domínguez JM, Crespo AJ, González-Cao J, Suzuki T, GómezGesteira M, Troch P. Long-crested wave generation and absorption for SPH-based DualSPHysics model. Coast Eng 2017;127(August 2016):37–54. http://dx.doi. org/10.1016/j.coastaleng.2017.06.004. [25] Verbrugghe T, Domínguez JM, Crespo AJ, Altomare C, Stratigaki V, Troch P, Kortenhaus A. Coupling methodology for smoothed particle hydrodynamics modelling of non-linear wave-structure interactions. Coast Eng 2018;138(April):184–98. http://dx.doi.org/10.1016/j.coastaleng.2018.04.021. CRediT authorship contribution statement Nicolas Quartier: Conceptualization, Formal analysis, Funding acquisition, Investigation, Methodology, Validation, Visualization, Writing – original draft, Writing – review & editing. Timothy Vervaet: Data curation, Funding acquisition, Investigation, Methodology, Validation. Gael Verao Fernandez: Data curation, Methodology, Writing – review & editing. José M. Domínguez: Resources, Software, Writing – original draft, Writing – review & editing. Alejandro J.C. Crespo: Conceptualization, Methodology, Resources, Software, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing. Vasiliki Stratigaki: Funding acquisition, Project administration, Supervision, Writing – original draft, Writing – review & editing. Peter Troch: Funding acquisition, Project administration, Resources, Supervision, Writing – review & editing. Declaration of competing interest The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Nicolas Quartier reports financial support was provided by Research Foundation Flanders. Data availability The data that has been used is confidential. Acknowledgements This research is being supported by the Research Foundation Flanders (FWO), Belgium - FWO fellowship No. 1SC5421N of Nicolas Quartier, and FWO fellowship No. 11A6919N of Timothy Vervaet. In addition, Vasiliki Stratigaki is a postdoctoral researcher (fellowship 1267321N) of the FWO, Belgium. The experimental test campaign was supported by WECANet (COST Action CA17105) funding for the short term scientific mission (STSM) at Aalborg University. The collaboration between UGent and UVigo was supported by WECANet funding, enabling STSMs of the first author of this work at EPhysLab, Ourense, Spain. This work was also supported by the project SURVIWEC PID2020-113245RB-I00 financed by MCIN/ AEI /10.13039/ 501100011033. References [1] Drew B, Plummer AR, Sahinkaya MN. A review of wave energy converter technology. Proc Inst Mech Eng A 2009;223(8):887–902. http://dx.doi.org/10. 1243/09576509JPE782. [2] Fernandez GV, Balitsky P, Stratigaki V, Troch P. Coupling methodology for studying the far field effects of wave energy converter arrays over a varying bathymetry. Energies 2018;11(11). http://dx.doi.org/10.3390/en11112899. 16 Energy 296 (2024) 130888 N. Quartier et al. [46] Penalba M, Ulazia A, Saénz J, Ringwood JV. Impact of long-term resource variations on wave energy farms: The icelandic case. Energy 2020;192:116609. http:// dx.doi.org/10.1016/j.energy.2019.116609, URL https://www.sciencedirect.com/ science/article/pii/S0360544219323047. [47] Vervaet T, Quartier N, Moreno EC, Fernandez GV, Stratigaki V, Ferri F, Troch P. Experimental modelling of power take-off control of an array of two wecfarm wave energy converters. Internal report Ghent Unversity; 2022. [48] Wendland H. Piecewise polynomial, positive definite and compactly supported radial functions of minimal degree. Adv Comput Math 1995;4(1):389–96. http: //dx.doi.org/10.1007/BF02123482. [49] Dehnen W, Aly H. Improving convergence in smoothed particle hydrodynamics simulations without pairing instability. Mon Not R Astron Soc 2012;425(2):1068–82. http://dx.doi.org/10.1111/j.1365-2966.2012.21439. x, arXiv:1204.2471. [50] Molteni D, Colagrossi A. A simple procedure to improve the pressure evaluation in hydrodynamic context using the SPH. Comput Phys Comm 2009;180(6):861–72. http://dx.doi.org/10.1016/j.cpc.2008.12.004. [51] Antuono M, Colagrossi A, Marrone S. Numerical diffusive terms in weaklycompressible SPH schemes. Comput Phys Comm 2012;183(12):2570–80. http: //dx.doi.org/10.1016/j.cpc.2012.07.006. [52] Antuono M, Colagrossi A, Marrone S, Molteni D. Free-surface flows solved by means of SPH schemes with numerical diffusive terms. Comput Phys Comm 2010;181(3):532–49. http://dx.doi.org/10.1016/j.cpc.2009.11.002. [53] Fourtakas G, Dominguez JM, Vacondio R, Rogers BD. Local uniform stencil (LUST) boundary condition for arbitrary 3-D boundaries in parallel smoothed particle hydrodynamics (SPH) models. Comput & Fluids 2019;190:346–61. http: //dx.doi.org/10.1016/j.compfluid.2019.06.009. [54] Lo EY, Shao S. Simulation of near-shore solitary wave mechanics by an incompressible SPH method. Appl Ocean Res 2002;24(5):275–86. http://dx.doi. org/10.1016/S0141-1187(03)00002-6. [55] Dalrymple RA, Rogers BD. Numerical modeling of water waves with the SPH method. Coast Eng 2006;53(2–3):141–7. http://dx.doi.org/10.1016/j.coastaleng. 2005.10.004. [56] Crespo AJ, Gómez-Gesteira M, Dalrymple RA. Boundary conditions generated by dynamic particles in SPH methods. Comput Mater Contin 2007;5(3):173–84. http://dx.doi.org/10.3970/cmc.2007.005.173. [57] Zhang F, Crespo A, Altomare C, Domínguez J, Marzeddu A, Shang SP, GómezGesteira M. Dualsphysics: A numerical tool to simulate real breakwaters. J Hydrodyn 2018;30(1):95–105. http://dx.doi.org/10.1007/s42241-018-0010-0. [58] English A, Dominguez JM, Vacondio R, Crespo AJ, Stansby PK, Lind SJ, Chiapponi L, Gomez-Gesteira M. Modified dynamic boundary conditions ( mDBC ) for general purpose smoothed particle hydrodynamics ( SPH ): application to tank sloshing , dam break and fish pass problems. Comput Part Mech 2022;9:1–15. http://dx.doi.org/10.1007/s40571-021-00403-3. [59] Liu MB, Liu GR. Restoring particle consistency in smoothed particle hydrodynamics. Appl Numer Math 2006;56(1):19–36. http://dx.doi.org/10.1016/j.apnum. 2005.02.012. [60] Canelas RB, Domínguez JM, Crespo AJ, Gómez-Gesteira M, Ferreira RM. A smooth particle hydrodynamics discretization for the modelling of free surface flows and rigid body dynamics. Internat J Numer Methods Fluids 2015;78(9):581–93. http://dx.doi.org/10.1002/fld.4031. [61] Vervaet T, Stratigaki V, Ferri F, De Beule L, Claerbout H, De Witte B, Vantorre M, Troch P. Experimental modelling of an isolated wecfarm real-time controllable heaving point absorber wave energy converter. J Mar Sci Eng 2022;10(10). http: //dx.doi.org/10.3390/jmse10101480, URL https://www.mdpi.com/2077-1312/ 10/10/1480. [62] Altomare C, Gironella X, Crespo AJ. Simulation of random wave overtopping by a WCSPH model. Appl Ocean Res 2021;116:102888. http://dx.doi.org/10.1016/ j.apor.2021.102888. [63] Gruwez V, Altomare C, Suzuki T, Streicher M, Cappietti L, Kortenhaus A, Troch P. An inter-model comparison for wave interactions with sea dikes on shallow foreshores. J Mar Sci Eng 2020;8(12):1–37. http://dx.doi.org/10.3390/ jmse8120985. [64] Ancellin M, Dias F. Capytaine: a python-based linear potential flow solver. J Open Source Softw 2019;4(36):1341. http://dx.doi.org/10.21105/joss.01341. [65] Babarit A, Delhommeau G. Theoretical and numerical aspects of the open source BEM solver NEMOH. In: Proceedings of the 11th European wave and tidal energy conference. 2015, p. 1–12, (September 2015). doi:hal-01198800. [66] Jin S, Patton RJ, Guo B. Viscosity effect on a point absorber wave energy converter hydrodynamics validated by simulation and experiment. Renew Energy 2018;129:500–12. http://dx.doi.org/10.1016/j.renene.2018.06.006. [67] Falnes J. Ocean waves and oscillating systems. Cambridge University Press; 2002. [68] Faedo N, Carapellese F, Pasta E, Mattiazzo G. On the principle of impedancematching for underactuated wave energy harvesting systems. Appl Ocean Res 2022;118(December). http://dx.doi.org/10.1016/j.apor.2021.102958. [69] Coe RG, Bacelli G, Forbush D. A practical approach to wave energy modeling and control. Renew Sustain Energy Rev 2021;142(December 2020):110791. http: //dx.doi.org/10.1016/j.rser.2021.110791. [26] Domínguez JM, Crespo AJ, Hall M, Altomare C, Wu M, Stratigaki V, Troch P, Cappietti L, Gómez-Gesteira M. SPH simulation of floating structures with moorings. Coast Eng 2019;153(November 2018). http://dx.doi.org/10.1016/j. coastaleng.2019.103560. [27] Tagliafierro B, Crespo AJC, Domínguez JM, Feal OG, Gesteira G, Canelas RB, Coe RG, Bacelli G, Cho H, Spencer SJ. Numerical modelling of a point - absorbing WEC model using DualSPHysics coupled with a multiphysics library. In: EWTEC. 2019, p. 1–8. [28] Vacondio R, Altomare C, De Leffe M, Hu X, Le Touzé D, Lind S, Marongiu JC, Marrone S, Rogers BD, Souto-Iglesias A. Grand challenges for smoothed particle hydrodynamics numerical schemes. Comput Part Mech 2021;8(3):575–88. http: //dx.doi.org/10.1007/s40571-020-00354-1. [29] Stansby P. Smoothed particle hydrodynamics (Sph) for renewable energies offshore. In: Proceedings of the 3rd international conference on renewable energies offshore. RENEW 2018, 2018. [30] Rafiee A, Elsaesser B, Dias F. Numerical simulation of wave interaction with an oscillating wave surge converter. In: Proceedings of the international conference on offshore mechanics and arctic engineering - OMAE. Vol. 5, 2013, http: //dx.doi.org/10.1115/OMAE2013-10195, (June). [31] Edge BL, Gamiel K, Dalrymple RA, Hérault A, Bilotta G. Application of GPUSPH to design of wave energy. In: Proceedings of the 9th international spheric workshop, Paris, France (June 2016). 2014, p. 342–7. [32] Brito M, Canelas RB, García-Feal O, Domínguez JM, Crespo AJ, Ferreira RM, Neves MG, Teixeira L. A numerical tool for modelling oscillating wave surge converter with nonlinear mechanical constraints. Renew Energy 2020;146:2024–43. http://dx.doi.org/10.1016/j.renene.2019.08.034. [33] Crespo AJ, Altomare C, Domínguez JM, González-Cao J, Gómez-Gesteira M. Towards simulating floating offshore oscillating water column converters with smoothed particle hydrodynamics. Coast Eng 2017;126(January):11–26. http: //dx.doi.org/10.1016/j.coastaleng.2017.05.001. [34] Wen H, Ren B, Yu X. An improved SPH model for turbulent hydrodynamics of a 2D oscillating water chamber. Ocean Eng 2018;150(February):152–66. http://dx.doi.org/10.1016/j.oceaneng.2017.12.047. [35] Zhu G, Graham D, Zheng S, Hughes J, Greaves D. Hydrodynamics of onshore oscillating water column devices: A numerical study using smoothed particle hydrodynamics. Ocean Eng 2020;218(May):108226. http://dx.doi.org/10.1016/ j.oceaneng.2020.108226. [36] Quartier N, Crespo AJ, Domínguez JM, Stratigaki V, Troch P. Efficient response of an onshore oscillating water column wave energy converter using a one-phase SPH model coupled with a multiphysics library. Appl Ocean Res 2021;115(September):102856. http://dx.doi.org/10.1016/j.apor.2021.102856. [37] Zhu G, Samuel J, Zheng S, Hughes J, Simmonds D, Greaves D. Numerical investigation on the hydrodynamic performance of a 2D U-shaped oscillating water column wave energy converter. Energy 2023;274:127357. http:// dx.doi.org/10.1016/j.energy.2023.127357, URL https://www.sciencedirect.com/ science/article/pii/S036054422300751X. [38] Yeylaghi S, Beatty S, Crawford C, Oshkai P, Buckham B, Moa B. SPH modeling of hydrodynamic loads on a point absorber wave energy converter hull. In: Proceedings of the 11th European wave and tidal energy conference (October). 2015, p. 1–7. [39] Quartier N, Ropero-Giralda P, M. Domínguez J, Stratigaki V, Troch P. Influence of the drag force on the average absorbed power of heaving wave energy converters using smoothed particle hydrodynamics. Water 2021;13(3):384. http: //dx.doi.org/10.3390/w13030384. [40] Tagliafierro B, Martínez-Estévez I, Domínguez JM, Crespo AJ, Göteman M, Engström J, Gómez-Gesteira M. A numerical study of a taut-moored pointabsorber wave energy converter with a linear power take-off system under extreme wave conditions. Appl Energy 2022;311:118629. http://dx.doi.org/10. 1016/j.apenergy.2022.118629. [41] Balitsky P, Verao Fernandez G, Stratigaki V, Troch P. Assessing the impact on power production of WEC array separation distance in a wave farm using one-way coupling of a BEM solver and a wave propagation model. In: Proceedings of the Twelfth European wave and tidal energy conference. 2017, p. 1176—-1176–10. [42] Balitsky P, Quartier N, Fernandez GV, Stratigaki V, Troch P. Analyzing the near-field effects and the power production of an array of heaving cylindrical WECS and OSWECs using a coupled hydrodynamic-PTO model. Vol. 11, 2018, http://dx.doi.org/10.3390/en11123489. [43] Falnes J. Wave-energy conversion through relative motion between two singlemode oscillating bodies. J Offshore Mech Arct Eng 1999;121(1):32–8. http: //dx.doi.org/10.1115/1.2829552. [44] Wu J, Yao Y, Zhou L, Chen N, Yu H, Li W, Göteman M. Performance analysis of solo duck wave energy converter arrays under motion constraints. Energy 2017;139:155–69. http://dx.doi.org/10.1016/j.energy.2017.07.152, URL https: //www.sciencedirect.com/science/article/pii/S0360544217313361. [45] Zhou B, Hu J, Jin P, Sun K, Li Y, Ning D. Power performance and motion response of a floating wind platform and multiple heaving wave energy converters hybrid system. Energy 2023;265:126314. http://dx.doi.org/10.1016/ j.energy.2022.126314, URL https://www.sciencedirect.com/science/article/pii/ S0360544222032005. 17
0
You can add this document to your study collection(s)
Sign in Available only to authorized usersYou can add this document to your saved list
Sign in Available only to authorized users(For complaints, use another form )