Upscaling of the standard three-bladed upwind design is one of the main options for reducing the levelized cost of energy for floating offshore wind. As a result of this upscaling, the rotor-nacelle assembly (RNA) is subjected to higher aerodynamic loads and weight. Simultaneously, efforts to reduce material costs are leading to lighter floaters. These concurrent trends are challenging the traditional assumption of a rigid floater, now proven to be unrealistic. While extensive studies exist on both blades and tower flexibility, introducing floater flexibility changes the dynamic behavior of the structure. This study aims to improve the understanding of interactions between floater elasticity and the rest of the structure. A key objective is to identify the deformation modes of the platform and link them to certain RNA modelling parameters. To investigate these interactions, simulations were conducted using two models: one with a rigid hull and another with a fully flexible hull. For each model, a sensitivity analysis of the eigenmodes was performed by varying the blade bending stiffness and the nacelle orientation. Results of the eigenvalue analysis were compared using the Modal Assurance Criterion, taking the initial flexible model as a reference. Doing so helped identify pure platform modes as well as coupling modes influenced by nacelle orientation and blade bending stiffness. These results provide a foundation for more accurate fatigue load predictions and support the development of cost-effective floating wind turbine designs.

Large floating offshore wind turbines (LOWTs) rated at 15–25 MW are under active development worldwide to exploit deep-water wind resources. In the North Sea, they are expected to play a key role in decarbonizing power generation, particularly for Norway, which has the world’s second-largest offshore wind potential (≈15 000 TWh yr⁻¹), mostly in deep waters. Along Norway’s west coast, especially at Utsira Nord, which will host one of the largest floating offshore wind farms in the world, hindcast data indicate significant wind–wave misalignment during extreme weather. However, the influence of such misalignment on the dynamic response of floating LOWTs remains unclear.

This study focuses on the Utsira region, located off the coast of Norway, with the meteorological and oceanographic data obtained from the NORA3 hindcast database. These are used to statistically characterize the joint distribution of wind and wave directions and to identify representative environmental states for the simulation campaign (Cheynet et al., 2024).

This study uses OpenFAST, the NREL open-source aero-hydro-servo-elastic simulator, to perform a parametric analysis of the IEA 15-MW reference floating wind turbine on a semi-submersible platform. The wind-wave misalignment angle is systematically varied to assess its impact on the turbine’s global motions and structural responses. Previous work (Barj et al., 2014), based on a 5 MW turbine, showed that neglecting misalignment can underestimate tower base bending moments by up to 50%. This study extends that investigation to modern 15 MW-class turbines, focusing on the resulting tower base bending moments and blade-root loads. Furthermore, the simulation campaign covers different wind speeds along the normal operating conditions, as well as an extreme load case corresponding to a 50-year return-period wind event. By combining the simulated load cases with the joint probabilistic distribution of the local wind and wave resources, a fatigue analysis is performed to evaluate the long-term structural response under realistic environmental variability. The probabilistic fatigue results are then compared with the ultimate loads obtained under the extreme event, enabling a consistent assessment across both fatigue and ultimate limit states. This approach allows reconstruction of a realistic load history and provides insight into how the cumulative fatigue behaviour relates to the short-term extreme response, for a LOWT deployed in the Utsira region.

The outcomes will provide quantitative insight into how realistic environmental influences the structural response of next-generation floating wind turbines under operational and extreme conditions. The results are expected to inform load-case definition, design standards, and safety-factor selection, supporting reliable and cost-efficient deployment of floating LOWTs at sites such as Utsira Nord.

In an attempt to lower greenhouse gas emissions and transition to a more sustainable energy production, offshore wind turbines have proven to be a key solution. However, the installation of monopiles remains limited to about 60 meters water depth, which limits the suitable locations. To expand the working area of offshore wind turbines, floating offshore wind turbines (FOWTs) are needed. The design and optimization of these complex systems depend on accurate numerical simulation of their hydrodynamic and structural responses. Especially low-frequency loads from the second-order mean drift forces and difference frequency loads are of importance. Those forces can induce resonance in the catenary mooring system and long, slender tower. This leads to a reduction in both performance and lifetime of the FOWT. Both industry and academia rely on potential-flow-based Boundary Element Method (BEM) solvers, yet a comprehensive benchmark of their second-order predictions is lacking.

This research presents a rigorous code-to-code comparison of four prominent BEM solvers: open-source Nemoh is compared to the closed-source proprietary solvers Wamit, OrcaWave, and HydroStar. A specific focus is given to their capability to calculate second-order effects through Quadratic Transfer Functions (QTFs). The study assesses two benchmark platforms: the well-documented 5MW DeepCWind and the larger 15MW VolturnUS-S platform, the latter is representative of current industry trends.

The benchmarking methodology is structured in a two-step approach within the frequency domain. First, a comparison is performed between the first-order hydrodynamic properties, which are not only dependencies for the second-order calculations but are also important inputs for coupled dynamic time-domain solvers like OpenFAST. This includes a comparison of hydrostatic stiffness, frequency-dependent added mass, radiation damping, and first-order wave excitation forces, as well as the resulting Response Amplitude Operators (RAOs). Subsequently, the research focuses on the second-order induced loads. The analysis evaluates the mean drift forces and the difference-frequency QTFs at frequency pairs corresponding to the surge, heave, and pitch resonance periods of the platforms, moorings, and towers.


The comparisons show a high agreement for the first-order properties between all solvers on both platforms. The comparison of calculations, taking into account second-order forces, shows larger discrepancies. These variations are more pronounced for the larger VolturnUS-S platform and tend to increase at higher difference frequencies.

The significance of this work is twofold. First, it serves as a crucial validation study for Nemoh, currently the only open-source BEM code capable of calculating the full QTFs. This validation is a vital step towards establishing a fully open-source simulation methodology, enabling the use of an open-source BEM code with solvers like OpenFAST for time-domain analysis. Second, by quantifying the variability in second-order load predictions, this study highlights an area of uncertainty in the FOWT design process. These insights are essential for optimizing the performance and cost of the floater and mooring systems, contributing to more reliable and economically viable floating wind energy.

Aero-elastic models offer a wide range of use cases for the operation of wind farms. If available, mechanical load and fatigue analysis or operational optimization can be performed. Especially, in digital twin use cases, models that represent the dynamic behaviour of the physical asset are necessary. However, operators of wind farms or researchers do not have access to aero-elastic models of the industry turbines. This leads to a complicated parameterization and development problem of aero-elastic models. In general, generic open-source models can be utilized, however the power and dimensions do not necessarily match the industry turbine. Scaling methods could be applied to correct dimensions, but not all turbine quantities can be matched by scaling. This can lead to a mismatch between the real wind turbine and the derived aero-elastic model of it. To overcome this limitation, a procedural framework is developed for designing and estimating aerodynamic blade parameters from reference data, which are then integrated into aero-elastic wind turbine models.
The developed framework allows for the creation of aero-elastic models capable of reproducing not only the power curve of existing turbines, but also dynamic aerodynamic behaviour. The blade design constraints are derived from operational information, e.g. the tip-speed-ratio and produced power of the turbine. A multi-objective generic optimization process using the Non-dominated Sorting Genetic Algorithm II (NSGA-II) is utilized to determine the aerodynamic design parameters like distributions of chord, twist, and relative thickness. All structural properties are interpolated from existing data at this stage. The aerodynamic performance of the blades is evaluated through steady Blade Element Momentum (BEM) simulations. The methodology incorporates literature-based constraints and introduces an additional aerodynamic constraint aimed at locating the maximum global aerodynamic efficiency at a desired operating point.
The developed workflow was rigorously tested in various case studies and compared to an industry loads model of the Adwen AD 8–180 turbine as a reference model. This includes optimization parameters such as population and generation size and design parameters like airfoil selection, inflow profile, design tip-speed ratio, and drivetrain efficiency. The results indicate that the estimated blades successfully replicate the aerodynamic performance of the reference rotor, achieving a mean absolute error of 2–4% across the partial-load region.
In summary, a framework that estimates aerodynamic parameters of wind turbine blades using publicly available data was developed. The Adwen AD 8-180 wind turbine served as a reference and validation case, supporting the analysis across multiple conditions. The framework demonstrated its capability to generate blade designs that closely match the aerodynamic performance of the reference rotor.