The design of floating wind turbines and several of their key components tend to be governed by fatigue loading, caused by the interaction with random wave and wind processes throughout the operational lifetime. Accurate representation and synthetization of the environmental conditions for a given project location is thus fundamental for reliable calculation of fatigue loading, with significant implications for the evaluation of project feasibility and profitability.
The traditional approach to fatigue analysis, inherited from design of O&G floaters, has been to use 2D wave scatter diagrams. I.e. bins of significant wave height Hs and spectral peak period Tp with associated mean wind speed. Alternatively, and due to the increased importance of wind loads, 2D scatter diagrams of wind speed and Hs with associated Tp have been used or bottom-fixed wind. For floating wind, both wind and wave contributions and their relative directions can be important, which suggests more complex (higher dimensional) scatter diagrams or alternative approaches such as sampling cases directly from hindcast, as discussed in the update to DNV-OS-E301 Position Mooring from 2024.
The authors of this paper have developed, independently and separately, two novel approaches for load case selection tailored to floating wind design, one based on higher dimensional scatters, and one based on hindcast sampling, within their respective companies specializing in floating wind engineering and development. Each method has its own distinct advantages and distinct limitations. One of the drawbacks, at least from the perspective of project certification, is that neither method corresponds to the recommended approach in the mentioned DNV-OS-E301 standard, a variety of hindcast sampling referred to as ‘importance sampling’.
In this paper, the two floating wind specific methods of fatigue load case selection are described and discussed in terms of their merits and limitations and compared to the DNV recommended approach. Fatigue load case matrices are established using all three methods, in a case study for a specific floating wind farm in the North Sea basin, and simulated based on time domain fully coupled analysis. Results are compared in terms of aggregate fatigue lifetime results of both floater structure (tower) and mooring lines. In addition, more granular analyses are presented to shed light on the relative contribution to fatigue loading from the different components (directionality, wind speed distribution, wave height distribution, wave period distribution, etc.) within the different approaches. All three approaches are compared to each other, and none is considered ‘correct’.
The aim of the work is to give practical guidance for fatigue load case selection that allows efficient engineering analyses and avoids unnecessary conservatism with quantifiable uncertainties, for development and certification of future full-scale floating wind projects.

Floating offshore wind turbines (FOWTs) are subjected to complex environmental and operational loads throughout their lifetime, leading to various structural damage and failure scenarios. Operation and maintenance (O&M) of these systems are particularly challenging and costly due to limited accessibility, harsh weather conditions, and the increasing size and complexity of turbines. Specialized vessels, motion-compensated access systems, and skilled personnel further contribute to high operational expenditures and extended downtimes.

Structural Health Monitoring (SHM) offers a promising approach to mitigate these challenges by enabling continuous condition assessment and early damage detection, which allows for the optimization of maintenance schedules and reduced inspection frequency. To develop an effective SHM framework, it is essential to identify and understand the potential damage scenarios that may occur over the lifetime of FOWTs.

This study presents a systematic framework for identifying and modelling damage scenarios in FOWTs. Likely failure modes affecting the platform, tower, and mooring system are reviewed and categorized based on their physical origin and potential impact on global dynamics. A practical approach on how to simulate a damage scenario in OpenFAST is proposed, highlighting how such faults can be represented through changes in stiffness, mass, or hydrodynamic properties. The OC4 Semi-submersible platform is adopted as a case study, and OpenFAST simulations are performed to generate healthy and damaged time-domain responses. The findings enhance understanding of how damage affects the global dynamics of floating wind turbines and contribute to the development of reliable, cost-effective SHM systems for future floating offshore wind applications.

Offshore wind energy has seen remarkable development in recent year, and floating offshore wind turbines (FOWTs) in particular have attracted attention as a renewable energy source in deep-water areas where bottom-fixed offshore wind turbines are not suitable. FOWTs are subject to variable environmental loads such as wind, waves, and currents, which necessitates structural design based on accurate prediction of cumulative fatigue damage over their service life. The international standard IEC 61400-3-2 requires that fatigue design considers the long-term joint distribution of wind speed, significant wave height, peak spectral period, and their directions as metocean conditions. Fatigue assessment requires performing a huge number of coupled wind-wave-current analysis cases using these metocean conditions as inputs, posing a major challenge.

This study proposed a method for assessing fatigue damage of FOWTs that takes into account the damage equivalent wave height, current velocity, as well as wind, wave and current directions. This study also proposed a method for evaluating cumulative fatigue damage over the entire service life, considering low-cycle fatigue under extreme conditions.

The annual fatigue damages obtained using two methods, the IEC 61400-3-2 method and the proposed method, are compared. IEC 61400-3-2 specifies that the long-term joint distribution of metocean conditions should be represented using resolutions, such as wind speed intervals of 2 m/s or less and significant wave height intervals of 0.5 m or less. Based on observed or simulated data, hourly values of the seven metocean parameters such as wind speed, wind direction, significant wave height, peak spectral period, wave direction, current speed, and current direction, are binned at specified resolutions. Extracting combinations are extracted, resulting in 6,675 cases off the coast of Fukushima. In the proposed method, when dividing hourly metocean parameters over a one-year period, significant wave heights within the same bin are treated as a single equivalent wave height for fatigue damage calculation, current speeds within the same bin are averaged, and infrequent wave and current directions are aggregated into the predominant directions. As a result of extracting these combinations, 906 cases are obtained. Coupled analyses are then performed using the IEC method (6,675 cases) and the proposed method (906 cases). the annual fatigue damage is calculated taking into account the occurrence frequency of each case. The results of both methods are found to be consistent.

Furthermore, fatigue damage due to low-cycle loads under extreme conditions that cannot be adequately captured by analyses based on one year of metocean conditions is calculated. Under the extreme conditions, the 50-year recurrence significant wave height is calculated from 30 years of metocean data. the cumulative fatigue damage over the entire service life, including contributions of low-cycle fatigue, is then calculated, taking into account the occurrence time of the 50-year recurrence significant wave height based on the exceedance of defined threshold. The proposed methods make it possible calculate cumulative fatigue damage over the service life using both coupled analyses based on one year of metocean conditions and coupled analyses under extreme conditions.

A digital twin is a virtual counterpart of a physical asset, such as an offshore wind farm, allowing to monitor the asset closely and identify potential opportunities for improving O&M strategies through scenario analysis. Still, to perform scenario analysis, a proper calibration of sub-models is a crucial component of a digital twin, as these should reflect reality accurately in a transparent manner to generate reliable and interpretable insights.

To this end, an integrated framework for constructing digital twins of offshore wind farms is presented, which can be calibrated using multiple types of field measurements.
First, scalable hyperparameter tuning is used to calibrate analytical wake models, e.g., the Gauss-Curl Hybrid model, using field data. Calibration methods using SCADA anemometer data and scanning lidar measurements are compared. For the scanning lidar, the wind field is reconstructed to estimate horizontal wind speeds from line-of-sight velocity data. These estimates are combined with SCADA wind direction measurements and the wind farm layout to identify wake-induced velocity deficits, which are then used for hyperparameter tuning. In the case of SCADA data, a time-series-based calibration approach is used on anemometer wind speeds and active power of the turbines.
Second, accelerometer sensors are installed on drivetrain components. The vibration measurements are combined with SCADA to perform operational modal analysis to extract the eigenfrequencies and compare modal behaviors. The resulting modes are used to calibrate a turbine model in OpenFAST, obtained by scaling open-source reference turbines, such as the LEANWIND 8MW and DTU 10MW turbines. The calibrated wake model is used to generate a wind field based on initial inflow wind conditions, and create local observations for each turbine.
Third, the wake and load models are integrated into a single digital twin of the wind farm, which allows predicting both wear-out component loads and produced powers under various scenarios, such as curtailments, stops or normal operation. The full digital twin is calibrated on data of multiple offshore wind parks within the Belgian-Dutch North Sea area, covering data from in-house accelerometers, lidars, scada and reanalyses.

To demonstrate the framework, alternative control strategies are investigated and optimized using the calibrated digital twin. An AI-driven wind farm controller is optimized that considers grid constraints and health of turbines. Specifically, power set-points are selected per turbine within the wind farm in order to match the energy production requested by an external party (TSO or BRP), while reducing the impact of the derations on wear-out component loading. The resulting controller is compared with current practice and a state-of-the-art analytically-derived controller. While the baselines consider the imposed grid constraints, it is challenging to include anything related to turbine health metrics. The controller learns the optimal set-point allocation strategy through repeated interactions with the digital twin, effectively optimizing asset lifetime.