Floating wind energy is a relatively new area that consists of harnessing wind energy from wind turbines that are supported by a floating foundation. This enables the installation of offshore wind turbines in deep seas, which means tapping into offshore wind resources that are unreachable with bottom-fixed wind turbines. Up to now, the feasibility of floating wind turbine technology has been demonstrated in small pilot farms. However, floating wind turbines are still subject to unexpected failures. Therefore, a better fundamental understanding of these turbines is needed to improve the technology to accelerate its deployment and reduce the cost of energy. Furthermore, the dynamics of floating wind turbines is different from those of their bottom-fixed counterparts. This presents challenges and opportunities across the different phases of their development and operation. This position paper addresses the fluid mechanics community and presents key challenges and research needs in the field of floating wind energy. Building on the grand challenges identified in the wind energy community, the manuscript addresses three focus areas and their interactions: the met-ocean conditions, the wind turbine, and the wind farm. Five groups of fluid mechanics driven challenges are highlighted: unsteady aerodynamics, high-speed flows, non-linear hydrodynamics, flow-induced vibrations, and wake dynamics. In addition, the kind of research methods and infrastructure needed to address these challenges are discussed, including cross-cutting themes such as digitalisation and co-creation across stakeholders and disciplines. Finally, the conclusions provide overarching recommendations to solve the upcoming challenges in floating wind energy and highlight the role that the fluid mechanics community could play.

Within a wind farm, each wind turbine extracts kinetic energy from the flow to convert it into electric energy. Unavoidably, this reduces the downstream availability of kinetic energy, diminishing the power generation of turbines operating in the waked region. These wake-induced power losses cumulate throughout the wind farm, posing a risk to its economic feasibility. One method that mitigates these power losses is helix active wake control. By leveraging individual blade pitch control, it induces an uneven thrust distribution over the rotor plane, which rotates either in clockwise (CW) or counterclockwise (CCW) direction around the rotor center. The wake deforms into a helical shape that recovers faster than the wake of a conventionally controlled turbine and thereby increases the total generated power. Notably, the CCW helix consistently outperforms the CW helix across all available studies. This work investigates the physical principles underlying these wake recovery enhancements using large eddy simulations (LES) of a wind turbine exposed to laminar, uniform flow. We observe a spatially coherent helical vortex structure in the wake boundary, which actively transports mean kinetic energy into the wake and, therefore, poses a fundamental contributor to the wake recovery enhancement. The opposing rotational directions of CW and CCW helixes result in distinct interactions of the helical vortex with the hub vortex, leading to different wake recovery mechanisms. In the investigated laminar inflow, the CCW helix has transported 44.8% more mean kinetic energy into the wake than the CW helix up to a streamwise position of 5D, explaining their differing efficacies observed in previous studies.

Power losses at waked turbines due to the energy extraction of upstream turbines from the flow pose a major risk to the economic feasibility of wind farms. Helix active wake control has proven its potential to mitigate these wake-induced power losses by accelerating the recovery of the individual turbine wakes. This method leverages individual pitch control to induce a non-uniformly distributed force perturbation that rotates either in a clockwise (CW) or counterclockwise (CCW) direction around the rotor center. This deforms the wake into a helical shape that recovers faster than the wake of a conventionally controlled turbine. The CCW-oriented helix achieves higher power gains than the CW helix. Previous studies have identified a system of counter-rotating vortices to drive the wake recovery enhancement and the difference between CW and CCW helix. Nevertheless, a causal explanation for the creation of these vortices is still pending. This work contributes to understanding their creation by isolating the effect of the helix force perturbation on a symmetric wake from the impact of blade-related features like tip-vortices, hub vortex, or wake swirl. For this purpose, we perform Particle Image Velocimetry (PIV) measurements of a porous disc (PD) model in a wind tunnel. The PD is modified to mimic the helix but does not inherit the blade-related features present in a wind turbine wake. We observe the formation of two counter-rotating vortices in the far wake that deform the wake cross-section into a kidney shape, analogous to the structures present in the wake when helix active wake control is applied to a wind turbine. A conceptual comparison of PD wake and wind turbine wake implies that the wake swirl present in the turbine wake causes asymmetric reactions in several characteristics of the vortex system to changes in the rotational direction of the helix perturbation. Consequently, the dynamic, non-uniform helix perturbation alone is sufficient to activate the governing mechanisms that enhance the wake recovery when using helix active wake control, while blade-related phenomena are not fundamental to the principal processes.

Clustering multiple turbines in close vicinity gives rise to efficiency losses due to the energy extraction of upstream turbines, a phenomenon known as the wake effect. The risk wake-induced power losses pose for the economic feasibility of wind farm projects motivated several methodologies aimed at mitigating the wake effect by dynamically exciting one operational parameter of the upstream turbine. Among them are dynamic yawing, which sinusoidally varies the yaw angle of the turbine with the wind, and helix active wake control, which dynamically manipulates the turbine thrust. This study is the first to explore the potential of exciting two operational parameters simultaneously by synergizing dynamic yawing and helix active wake control. Therefore, we conduct wind tunnel experiments using a yawable porous disc model modified to mimic the effect of the helix on the flow. A particular focus is put on the relative orientation between helix and dynamic yawing. Results indicate that wake recovery enhancements achievable by synergizing helix and dynamic yawing are in the same range as both methods individually; however, at 50% lower excitation frequencies than only helix and 10° smaller yawing amplitudes compared to only dynamic yawing.

Two setups are used to investigate differences between modeling a wind turbine nacelle by means of an actuator-line model (ALM) and a wall-model (WM) using large-eddy simulations. One advantage of the ALM is that it requires a lower mesh refinement, making it less computationally costly. In the first setup, the nacelle is in standalone configuration and the ALM results show a much lower turbulence intensity and a significantly slower wake recovery when compared to the WM cases. In the second setup, the nacelle is in a rotor-nacelle assembly configuration and many variations of the ALM are tested in order to match the results from the experiment addressed in the OC6 task phase III. Contrary to previous findings that the nacelle might affect the turbine loads, this study shows that the improved match with the experiment stems from the increased mesh refinement in the nacelle region rather than the actual presence of the nacelle. Nevertheless, the wake profiles in the near-wake show a very good agreement between the ALM and WM, regardless of the refinement in the nacelle region. These cases also show a higher wake deficit than not using any nacelle at all.

This paper presents a surrogate-assisted optimisation approach to speed up the substructure analysis in the preliminary design phase. The approach consists of replacing the radiation-diffraction analysis in a frequency domain analysis model for floating wind turbines with a data-driven surrogate model predicting the hydrodynamic coefficients for parameterised substructure geometries. This procedure is compared with the reference approach of estimating the hydrodynamic coefficients via radiation-diffraction analysis. A representative use case of assessing the trade-off between minimising the capital cost and reducing the wave-induced nacelle acceleration standard deviation for a semi-submersible substructure is presented. The accuracy of the surrogate model is found to increase significantly up to training datasets consisting of 400 designs and less noticeably afterwards. For a dataset consisting of 400 designs, the mean error on the prediction of the hydrodynamic coefficients and the error at one standard deviation from the mean are generally below 7% and 10%, respectively. For the same dataset size, the mean error on the most probable maximum wave-induced pitch over a 3h storm period is below 17%, while the error at one standard deviation from the mean is lower than 27%. The same values for the most probable maximum nacelle acceleration are under 7% and 12%, respectively. The surrogate model can capture the trade-off between the two objective functions, and the optimal designs identified with the surrogate model generally follow the same trend as those obtained with the reference model. However, relying on the surrogate model for performing the analysis of the substructure introduces local minima in the objective function that cause a discrepancy between the optimal designs identified with the surrogate model and those identified with the reference model.

Actuator line modeling of wind turbines requires the definition of a free-stream velocity in a computational mesh and a regularization kernel to project the computed body forces onto the domain. Both choices strongly influence the results. In this work, a novel velocity sampling method—the so-called effective velocity model (EVM)—is implemented in the CFD software SOWFA, validated, and compared to pre-existing approaches. Results show superior method robustness with respect to the regularization kernel width ((Formula presented.)) choice while preserving acceptable accuracy. In particular, the power predicted by the EVM is nearly independent of the (Formula presented.) value.

In recent years, the relevance of the interaction between neighboring wind farms has grown steadily. As one farm extracts energy from the wind, a downstream one can systematically experience lower wind speeds which threatens the economic viability of the farm. Significant progress has been made in understanding these farm-farm wake interactions, but we still lack methodologies to mitigate their undesired effects. In this study, we introduce Active Cluster Wake Mixing (ACWM). This novel method aims to accelerate the recovery of the cluster wake using dynamic control actions: By exciting the thrust of the individual turbines depending on their relative location, we generate non-uniform patterns of energy extraction. Phase offsets between the individual excitation signals propagate these regions through the wind farm. This results in large-scale velocity gradients inside the farm, which also affect the flow in the cluster wake region. An in-depth exploration and optimization of ACWM requires significant computational effort. Therefore, we compare three different wind farm modeling approaches in Large Eddy Simulations (LES) that differ in their computational costs regarding their suitability for further exploration of ACWM. For this purpose, we use an unoptimized ACWM scheme with two different excitation frequencies. For the first time ever we successfully show that ACWM manipulates the flow inside the wind farm with favorable effects on the wake velocity. We also demonstrate that the modeling of cluster wakes is challenging and has a significant effect on the potential gain.

Floating offshore wind turbines experience different operating conditions, such as wind and wave inflow characteristics. Accurate prediction of the loads acting on the floating wind system is essential for the system design and optimisation. However, there are a lot of uncertainties with the modelling input variables for time domain simulation tools such as OpenFAST to represent various hydro-aerodynamic and structural properties. The primary objective of this work is to identify the critical input parameters for different damage-equivalent load outputs for two substructure types: OC3 Hywind Spar and OC4 DeepCwind semisubmersible. The same rotor-nacelle assembly and tower (the NREL 5MW reference turbine) are used in both case studies. A sensitivity analysis based on the damage equivalent loads of six output quantities was conducted with 8 or 10 input parameters (depending on the floater). The dependent parameters were conditionally parameterised based on the independent inputs, such as wind speed and wind-wave misalignment. The outcomes of this work show that the floater type affects the sensitivity levels of wave characteristics and hydrodynamic drag coefficients with no significant influence on the turbulence intensity, as expected. Further, the drag coefficient for spar-buoy configuration significantly influences mooring line tension compared to the semisubmersible because of their drag-dominant slender structure. The current velocity is the most dominating parameter for the mooring loads, irrespective of the floater type. While wave characteristics also influenced some turbine loads, it was almost independent of the floater type. Furthermore, the choice of the hydrodynamic model does not affect the sensitivity level rankings. A convergence study on the number of starting points was conducted to ensure a global sensitivity approach. As seen in this study, the results are floating platform-specific. This study provides valuable insight into design-driving input parameters, characterising substructure-specific wind-wave influence.

Long Short-Term Memory Recurrent Neural Networks (LSTM) are used to build surrogate models to forecast time-series blade loads for both fixed and floating offshore wind turbines. In this paper, we train surrogate models on datasets generated with OpenFAST on the IEA-15MW-RWT under a range of metocean conditions. The aim of the surrogate models is to generate load forecasts inexpensively and accurately such that they can be used in a model predictive controller. Two cases are investigated with different model inputs: one with only measurements available to typical PI controllers and another one with additional wave elevation and deflection measurements (alongside the endogenous variable). The model performances are evaluated and compared. It was found that for the fixed turbine, the models predicted all three blade loads to a high degree of accuracy. The floating turbine surrogate models performed relatively worse, but edgewise and pitching moments are still reasonably accurate. The surrogate model forecasts the flapwise moment to a satisfactory accuracy only in 58% out of 400 test cases. The addition of wave elevation and blade deflection features did not significantly improve the prediction performance of the surrogate, demonstrating that just the information used by current PI controllers may be sufficient for forecasting blade loads.