Open-access simulation dataset of the floating wind turbine DeepCwind OC4 in wind and waves

Journal of Physics: Conference Series PAPER • OPEN ACCESS Open-access simulation dataset of the floating wind turbine DeepCwind OC4 in wind and waves To cite this article: Ajie Brama Krishna Pribadi et al 2025 J. Phys.: Conf. Ser. 3131 012024 View the article online for updates and enhancements. You may also like Label Free Detection of Programmed Death Ligand 1 Protein Biomarker by Quartz Tuning Fork-Based Biosensor Mahmoud Al-Gawati, Qura Tul Ain, Khalid E Alzahrani et al. - Quartz-enhanced photo-acoustic spectroscopy of gas media using multifrequency THz quantum-cascade laser with spectra reconstruction by machine learning Artem Sazhin, Ilya Ozheredov, Maxim Bannikov et al. - Determination of the static spring constant of electrically-driven quartz tuning forks with two freely oscillating prongs Laura González, Roger Oria, Luis Botaya et al. - This content was downloaded from IP address 157.193.240.165 on 12/11/2025 at 12:44 Content from this work may be used under the terms of the Creative Commons Attribution 4.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. Published under licence by IOP Publishing Ltd EERA DeepWind Conference 2025 Journal of Physics: Conference Series 3131 (2025) 012024 IOP Publishing doi:10.1088/1742-6596/3131/1/012024 1 Open-access simulation dataset of the floating wind turbine DeepCwind OC4 in wind and waves Ajie Brama Krishna Pribadi1, Abdulelah Al-Ghuwaidi2 and Evert Lataire1 1 Ships and Marine Technology Division, Department of Civil Engineering (FEA15), Ghent University, Technologiepark 60, 9052 Ghe nt, Belgium 2 OWI-lab, Department of Applied Mechanics (MECH) , Vrije Universiteit Brussel, Pleinlaan 2, 1050 Brussels, Belgium, Institution, City, Country *E-mail: [email protected] e Abstract. This paper focuses on the impact of vari ous wind and wave conditions on the motion, tower base forces/moments and mooring line tension of a reference Floating Offshore Wi nd Turbine (FOWT) platform. The simulations are performed using OpenFAST to assess a total of 576 operational and 576 damaged scenarios. Selected results are presented to assess the impact of three different hydrodynamic modelling: i) Linear Potential Flow (LPF) ii) LPF combined with Morison drag (hybrid) and iii) hybrid approach with the inclusion of Quadratic Transfer Function (QTF). The importance of including the Morison drag term is demonstrated by assessing the platform’s transient motion and the contribution of the heave disk to the heave motion. The impact of mean-drift from the difference-frequency second-order wave forces is evaluated by analysing the surge motion of the platform, comparing it with the LPF-only model. Power Spectral Density (PSD) of the tower base moment reveals several peaks in the frequency range corresponding to the contribution of the sum-frequency QTF. The impact of including QTF is more apparent for shorter wave periods. The loss of one mooring line has been found to increase the fairlead tension amplitude and induces higher load in the lowfrequency region, visible on the PSD of the fairlead tension. Lastly, different wind directions are analysed to assess the impact of aerodynamic load to the mooring system, keeping the same wave direction. The wind-wave misalignment induces higher peak tensions in the lower frequency region, close to its surge natural frequency. The datasets containing a total of 1152 simulation cases results are made publicly avai lable. 1. Introduction Capacity factor is a ratio of the average energy produced within a certain time period divided by its maximum possible output (i.e., full capacity) that cou ld have been generated. Therefore, it is an i mportant parameter to measure the efficiency of a wind turbine. Recent data from 2024 suggests that offshore wind turbines i n European Union countries generate an annual average of a 35% capacity factor [1]. Furthermore, the Hywind Scotland Floating Offshore Wind Turbines (FOWTs) can generate up to 57% of capacity factor [2]. As of 2023, offshore wind makes up 75.2 GW of total capacity globally where 14% of them were installed during that year [3]. Out of this number, around 270 MW installed capacity comes from the floating offshore wind turbines [4]. Floating technology is a recent emerging development that wil l allow wind turbines to be installed EERA DeepWind Conference 2025 Journal of Physics: Conference Series 3131 (2025) 012024 IOP Publishing doi:10.1088/1742-6596/3131/1/012024 2 in a water depth deeper than 60.0 m where 80% of the offshore wind potential lies [5]. Thus, FOWT can unlock the untapped potential of harnessing electricity in deeper waters with greater capacity factor than nearshore or onshore. To this end, several coll ective efforts have been made in the past decades to investigate the different types of floaters and understand the physical aspects involved in modelling the dynamics of a floating wind turbine system [6 - 11]. Due to its complexity, numerical modelling of an FOWT system generally requires coupling between several modules that simulate different aspects: aerodynamics, hydrodynamics, structural dynamics, turbine control system, mooring dynamics. There exists several numerical tools that can be used to perform the global response time-domain analysis of an FOWT system such as OpenFAST, OrcaFlex, OPASS , Bladed, HAWC2, aNySIM, PHATAS, 3DFloat, DeepLines Wind, SAMCEF, Sesam, UTWInd [12]. The output of these tools may include hundreds of different channels e.g. 6 Degrees of Freedom (6-DOF) platform’s motions, mooring line tension, sum of forces at a reference point, tower base/top forces and moments, power generated, thrust forces and moments. The aforementioned software are considered as mid-fidelity models which require a fine-tuning of coefficients obtained from the experiment or computational fluid dynamics (CFD) [13]. The experimental validation study and the use of higher-fidelity Computational Fluid Dynamics (CFD) and Finite Element Method (FEM) are performed at the detailed design phase, after the optimisation study is done using the lower to mid fidelity models [14]. At this phase, detailed structural checks for the platform are necessary to c onduct. However, the aforementioned global response analysis software generally considers the platform as a rigid body [15]. Thus, it neglects the stress distribution in the substructure. The lack of accurate representation of the distributed hydrodynamic pressure loads poses a challenge when determining the fatigue life of the critical regions in the platform [16]. One of the methods that can be utilized to tackle this issue is called pressure mapping[16-18]. This approach uses the results of the global response time-domain analysis and the Boundary Element Method (BEM) frequency-domain hydrodynamic analysis. The results of these analyses are then mapped into a quasi-static Finite Element (FE) model. Thus, the development of such a numerical tool requires the global response analysis results. Other than the physics-based numerical tools, the use of neural network with more than two hidden layers, Deep Learning (DL) techniques to model FOWT systems have been gaining traction in the recent years [19-22]. Due to its computational efficiency, this approach is commonly used to create a digital twin model of a particular FOWT system. However, the development of such a model requires different datasets for training, testing and validation. These datasets can come from the aforementioned mid-fidelity software. The aim of the study conducted in thi s paper is to aid the development of different numerical tools, whether to build shell element for pressure mapping, DL, black-box models or any tools related to the modelling of FOWT system. This is done by providing benchmark simulation datasets of global response time-domain analysis. 2. Methodology 2.1 OpenFAST An opensource Aero-Servo-Hydro-Elastic (AHSE) software called OpenFAST v3.5.3 [23] is used to perform the numerical simulations conducted in this paper. The flowchart consisting of the OpenFAST modules used in this study is shown in Figure 1. Note that WAMIT [24] frequencydomain outputs, consisting of hydrodynamic coefficients (i.e. added mass and damping), wave first order excitation force and total second-order wave forces (sumand difference-frequency) are used by the HydroDyn module to calculate the hydrodynamic force in the time-domain. In EERA DeepWind Conference 2025 Journal of Physics: Conference Series 3131 (2025) 012024 IOP Publishing doi:10.1088/1742-6596/3131/1/012024 3 HydroDyn, the hydrodynamic force can be calculated by combining the Linear Potential Flow (LPF) theory and the Mori son equation [25]. T he LPF theory account for the wave excitation force (diffraction and Froude-Krylov) and the radiation force. Using the frequency-domain coefficients, the Impulse Response Function (IRF) and infinite added mass are calculated to account for the memory effect in the Cummins equation [26, 27]. Viscous effect can be modelled by including the drag term of the Morison equation [25]. As for the wind input, TurbSim [28] is used to generate the Full-Field (FF) turbulent wind. This is used by the InflowWind module to calculate the undisturbed wind velocities at various points around the wind turbine [29]. The output of InflowWind is used by the AeroDyn [30] module to calculate the aerodynamic load ac ting on the turbine blades and the tower. The lumped-mass based mooring line module called MoorDyn [31] is used to simulate the dynamic behaviour of the mooring lines. 2.2 Numerical model setup and verification Figure 2 shows the frontal view (YZ plane) of the reference FOWT system used in this study. The detailed specification and dimension can be found in the documentation for the DeepCwind OC4 semi-submersible definition [31]. The FOWT platform is paired with the National Renewable Energy Laboratory (NREL) 5 MW reference turbine where its specification is described in [32]. The numerical model for this reference turbine and platform is available on OpenFAST GitHub platform. This model is used as a basis for the simulations conducted in this paper. To ensure that the model is working correctly, selected load cases are simulated and compared against the numerical benchmark results published in [7]. Figure 1. OpenFAST modules that were used to perform the simulations in this study, a dapted from [35] EERA DeepWind Conference 2025 Journal of Physics: Conference Series 3131 (2025) 012024 IOP Publishing doi:10.1088/1742-6596/3131/1/012024 4 Numerical simulations of decay tests for the moored system are performed for three different DOFs: surge, heave, pitch. The results from the decay tests simulations are compared with the results of DeepLines and OrcaFlex in [34]. Table 1 s hows the natural frequencies and periods in surge, heave and pitch, obtained from the simulated decay tests. The results are nearly identical to the ones produced by DeepLines and OrcaFlex with the biggest discrepancy found in the surge motion with an error of less than 2%. In addition to the decay tests, the Load Case (LC) 3.1 from [7] that combines wind and wave inputs are used to verify the model in this study. The uniform and steady wind velocity is set at 8.0 m/s. Regular wave height is s et at 6.0 m with the wave period of 10.0 s. Both wind and wave are aligned and perpendicular to the nacelle, propagating in positive x-axis. As shown i n Figure 3 left, the 3-DOF responses are showing good agreement where the peaks and troughs are matching the reference results of DeepLines and OrcaFlex. Similar trend can also be found in the fairlead tensions, shown in Figure 3 right, with a slight underprediction of the minimum tension in fairlead 2 from this study. 2.3 Simulation cases A total of 1152 individual simulation cases were generated and the results were uploaded as open-access datasets to the Zenodo platform. The download link is provided in [34]. Shown in Table 2 are the simulation matrix for the regular wave cases. All the combinations of the simulation variables (i.e., 2 scenario, 4 wave height , 4 wave period, 2 wind speed, 3 wind direction, 2 wind field) are simulated, resulting in a total of 384 unique simulation cases . The irregular wave cases have a total of 768 simulations with the combinations shown in Table 3. Figure 2. XZ plan v iew of the reference FOWT system (7). Table 1. Natural frequencies and periods compared against the simulated decay tests results in [7] Software [-] Surge Frequency - Period Heave Frequency - Period Pitch Frequency - Period OpenFAST v3.5.3 – This study 0.0089 Hz - 111.12 s 0.0566 Hz - 17.653 s 0.0400 Hz - 25.0 s OrcaFlex 0.0092 Hz - 109.09 s 0.0567 Hz - 17.647 s 0.0400 Hz - 25.0 s DeepLines WT 0.0092 Hz - 109.09 s 0.0567 Hz - 17.647 s 0.0400 Hz - 25.0 s EERA DeepWind Conference 2025 Journal of Physics: Conference Series 3131 (2025) 012024 IOP Publishing doi:10.1088/1742-6596/3131/1/012024 5 Figure 3. Platform’s motions (left) and mooring fairlead tensions (right) results compared against LC 3.1. Table 2. Simul ation matrix for the regular waves cases. Scenario [-] Wave height [m] Wave period [s] Wind speed [m/s] Wind direction [deg] Wind field [-] Operational 3.0 8.0 8.0 0.0 Steady Fairlead loss 6.0 10.0 13.0 30.0 Turbulent 9.0 12.0 45.0 12.0 20.0 Table 3. Simul ation matrix for the irregular waves cases. Scenario [-] Significant wave height [m] Wave peak period [s] Wind speed [m/s] Wind direction [deg] Wind field [-] Operational 1.5 8.0 8.0 0.0 Steady Fairlead loss 3.0 10.0 11.0 30.0 Turbulent 4.5 12.0 13.0 45.0 6.0 20.0 20.0 EERA DeepWind Conference 2025 Journal of Physics: Conference Series 3131 (2025) 012024 IOP Publishing doi:10.1088/1742-6596/3131/1/012024 6 3. Results and discussions 3.1 Influence of hydrodynamic models Three hydrodynamic models are c ompared to assess their influence on the floater’s responses: • Linear Potential Flow (LPF) only. • LPF with the inclusion of Morison drag (hybrid). • LPF with the inclusion of Morison drag and second-order wave forces via Quadratic Transfer Function (QTF). The input variables used in the simulations for this comparison are listed in Table 4. Note that the wave kinematics are only calculated up to the mean water level as no kinematic stretching is implemented in the aforementioned hydrodynamic models. During the transient part of the simulation, the inclusion of Morison drag dampened the surge response sooner compared to the LPF-only model, as shown in Figure 4. This is most pronounced for the wave peak period of 8.0 seconds where it takes more than 400 s for the surge response of LPF-only model to converge with the hybrid approach without QTF. In the case of the longest waves (20.0 s peak period), the influence of Morison drag to the surge motion is not apparent. On the othe r hand, influence of the mean-drift from the difference-frequency QTF can be seen in the shorter waves, especially for the peak period of 8.0 s. In this case, the platform is displaced further in the surge direction compared to the LPF-only and hybrid approach. Figure 5 s hows the heave response during the transient zone. This shows that the omission of the Morison axial drag causes the heave motion to be exaggerated throughout the simulation time, not only in the transient zone. This is because of the wave period that is close to the heave natural period of 17.65 s which induces resonance. Figure 4. Time series of the surge motion comparing different hydrodynamic models Table 4. Simul ation matrix for the co mparison of the hydrodynamic models Scenario [-] Significant wave height [m] Wave peak period [s] Wind speed [m/s] Wind direction [deg] Wind field [-] Operational 6.0 8.0, 10.0, 12.0, 20.0 11.0 0 Steady EERA DeepWind Conference 2025 Journal of Physics: Conference Series 3131 (2025) 012024 IOP Publishing doi:10.1088/1742-6596/3131/1/012024 7 Figure 6 and 7 show the Power Spectral Density (PSD) of the surge motion and heave, respectively, considering all simulation data (1 hour). In Figure 6, it can be seen that including second-order wave forces induces the peaks in the lower frequencies region. An exception appears for the wave peak period of 12.0 s where the (lower frequencies) peak is higher for the LPF only model compared to the model that includes Morison and QTF. This is because the omission of Morison drag induces large initial surge motion during the transient part of the simulation. Note that these peaks are close to the natural frequency of the surge motion which is at 0. 00899 Hz. Observing the PSD of the heave motion, it can be seen from the Figure 7 that the heave resonance peak at 0.056 Hz is consistent across all wave peak periods when Morison (axial) drag is not modelled. Therefore, highlighting the importance of a heave disk to dampen the he ave response. Furthermore, the influence of QTF to the heave motion is apparent across shorter wave peak periods (i.e., 8.0 s and 10.0 s ), which can be seen from the peaks in the lower frequencies region (i.e., lower than the wave peak frequency). Lastly, as shown in Figure 8, modelling the QTF induces tower base pitch moment in both lower and higher frequency regions. This is due to the difference-frequency and sum-frequency components of the QTF, respectively. Figure 5. Time series of the heave motion comparing different hydrodynamic models Figure 6. PSD of the surge motion comparing dif ferent hydrodynamic m odels EERA DeepWind Conference 2025 Journal of Physics: Conference Series 3131 (2025) 012024 IOP Publishing doi:10.1088/1742-6596/3131/1/012024 8 3.2 Damaged scenario – loss of fairlead 2 This analysis compares the damaged and operational scenarios for the same wind and wave conditions. The damaged condition is simulated by removing fairlead 2 before starting the simulation. The input variables used in this comparison are summarized in Table 5. The platform’s responses in surge, heave and pitch are shown in Figure 9. It can be seen that the platform drift for more than 800 m before it starts to oscillate around its new position where the remaining mooring lines are fully stretched. Note that the platform pitches more as the mooring Figure 7. PSD of the h eave motion com paring different hydrodynamic models Figure 8. PSD of the tower base pitch moment comparing different hydrodynamic models Table 5. Simul ation matrix for the comparison of the damaged scenario Scenario [-] Significant wave height [m] Wave peak period [s] Wind speed [m/s] Wind direction [deg] Wind field [-] Operational, damaged 6.0 8.0 13.0 0 Steady