Quantitation of damping is of great significance for the design and condition assessment of wind turbines. The authors’ previous theoretical and numerical studies showed that compared to damping ratios, a modal aerodynamic damping matrix can better describe the damping coupling in the fore-aft (FA) and side-side (SS) tower motions. In the present study, an improved damping identification method was first proposed to identify this damping matrix with artificial exciters and then verified by using OpenFAST simulations under different excitation frequencies, excitation force amplitudes, and different turbulent wind fields. Following the numerical study, a scaled wind turbine model with a geometric scale ratio of 1/75 was carefully designed based on the NREL 5 MW wind turbine prototype, in which the scaled blade design follows the rule in thrust coefficient similarity. An identification study was performed with this scaled model by a series of wind tunnel tests. The modal aerodynamic damping matrix was identified under steady-state harmonic excitation in the operating state and compared with the identified results by a free decay method and the theoretical values. The results experimentally confirm the correctness of the aerodynamic damping matrix theory under uniform wind and the feasibility of the improved identification method in practice.

not-yet-known not-yet-known not-yet-known unknown With increasing wind energy capacity and installation of wind turbines, new inspection techniques are being explored to examine wind turbine rotor blades, especially during operation. A common result of surface damage phenomena (such as leading-edge erosion) is the premature transition of laminar to turbulent flow on the surface of rotor blades. In the KI-VISIR (Künstliche Intelligenz Visuell und Infrarot Thermografie – Artificial Intelligence-Visual and Infrared Thermography) project, infrared thermography is used as an inspection tool to capture so-called thermal turbulence patterns (TTP) that result from such surface contamination or damage. To compliment the thermographic inspections, high-resolution photography is performed to visualise, in detail, the sites where these turbulence patterns initiate. A convolutional neural network (CNN) was developed and used to detect and localise the turbulence patterns. A unique dataset combining the thermograms and visual images of operational wind turbine rotor blades has been provided, along with the simplified annotations for the turbulence patterns. Additional tools are available to allow users to use the data requiring only basic Python programming skills.

Floating offshore wind (FOW) is a renewable energy source that is set to play an essential role in addressing climate change and the need for sustainable development. However, due to the increasing threat of climate emergency, more wind turbines are required to be deployed in deep water locations, further offshore. This presents heightened challenges for accessing the turbines and performing maintenance, leading to increased costs. Naturally, methods to reduce operational expenditure (OpEx) are highly desirable. One method that shows potential for reducing OpEx of FOW is LIDAR-assisted pitch control. This approach uses wind velocity measurements from a nacelle-mounted LIDAR to enable feedforward control of floating offshore wind turbines (FOWTs), and can result in reductions to the variations of structural loads. Results obtained from a previous study of combined feedforward collective and individual pitch control (FFCPC+FFIPC) are translated to OpEx reductions via reduced component failure rates for future FOW developments, namely in locations awarded in the recent ScotWind leasing round. The results indicate that LIDAR-assisted pitch control may allow for an up to 5% reduction in OpEx, increasing up to 11% with workability constraints included. The results varied across the three ScotWind sites considered, with sites furthest from shore reaping the greatest benefit from LIDAR-assisted control. This work highlights the potential savings and reduction in the overall levelised cost of energy for future offshore wind turbine projects deliverable through the implementation of LIDAR-assisted pitch control.

This study presents results from a long-term measurement campaign on a research wind turbine in the field. Pressure measurements are conducted at 25% blade radius over several months. Together with inflow measurements provided by a LiDAR system, they form an extensive dataset which is used in the validation of numerical aerodynamic models.  The model validation is conducted based on both ten-minute average data as well as time-resolved unsteady data. Initially, it is investigated how representative ten-minute average pressure measurements are of the underlying unsteady aerodynamics. Binned ten-minute average pressure distributions are then analysed together with their numerical counterpart, consisting of a combination of rotor and airfoil level aerodynamic/aeroelastic simulation results using average environmental and operating conditions as input. Finally, time-resolved measurements and simulation results are compared, validating the aeroelastic tools’ capability to reproduce unsteady aerodynamics.  Overall, reasonable agreement is found between numerical simulations and field experiment data showcasing two aspects: Low-fidelity numerical tools remain relevant for simulating modern multi-megawatt wind turbines and long-term pressure measurements provide invaluable means for validating such tools.

Annual energy production (AEP) is commonly used in objective functions for wind farm layout optimization. AEP is proportional to wind farm power production integrated over an annual distribution of free-stream wind conditions. Physics-based estimates of wind farm power production typically rely on low-fidelity engineering wake models that approximate the steady-state wind farm flow field. AEP estimates are then obtained by performing independent simulations for discrete wind conditions and using rectangular quadrature to account for each condition’s expected frequency of occurrence. Depending on the number of simulated discrete wind conditions, this numerical integral could be hampered by poor accuracy or high computational costs. The FLOWERS AEP model instead poses an analytical integral of the engineering wake model over the variable wind conditions, yielding a closed-form, analytical function for wind farm AEP. This paper derives the analytical functions for FLOWERS AEP and its derivatives with respect to turbine position, which are useful for gradient-based wind farm layout optimization, in non-dimensional form. We then analyze the benefits of the FLOWERS AEP model over conventional reference models, focusing on its low cost, adequate wake loss predictions, and smooth design space. We find that FLOWERS predicts AEP within about 10% of the baseline AEP models in less than 10% of the computational time. Furthermore, we illustrate how the FLOWERS design space at relatively low resolution is qualitatively similar to the reference and yields comparable optimal layouts.

The Additively-Manufactured, System-Integrated Tip (AMSIT) project is leveraging the flexibility of 3D printing to integrate several technologies in a wind turbine blade tip, while reducing the levelized cost of electricity (LCOE) produced. The design integration is demonstrated for a 200 kilowatt-scale turbine with 13-meter blades, with the outer 15% of the blade replaced with a 3D-printed design. Aerodynamic performance is enhanced through inclusion of a winglet and surface texturing, both challenging for traditional manufacturing. Longevity and durability is improved through integrated lightning and leading edge erosion protection. Increased power, reduced repair frequency, and ease of repair through blade modularity all contribute to reduced LCOE. Cost models are are extended to modern megawatt-scale designs to estimate the impact of the technology at scale, demonstrating the potential to reduce LCOE very significantly for modern onshore turbines, with even higher potential savings offshore.

In efforts to reduce the operation and maintenance cost of wind turbines, there is an increasing interest to monitor key turbine quantities such as the torque load on the gearbox. Monitoring the torque paves the way for the calculation of remaining useful lifetime, leading to cost reductions through improved reliability and maintenance planning. In order to avoid expensive direct torque sensors, this paper investigates the potential of virtual torque sensing, a technique based on 3 basic components: firstly, a set of non-intrusive sensors installed on the gearbox. Three groups of strain gauges on the gearbox as well an angular encoder are considered in this paper. Secondly, a physics-based model, capable of predicting the response of aforementioned sensors. These models are constructed with a purposeful balance between accuracy and computational cost. Model validation and updating are performed to ensure efficient and accurate prediction of the sensor output. Finally, an Augmented Extended Kalman Filter (AEKF) is used to combine the measured response with predictions from the model to infer the gearbox input torque. Since a key factor determining the performance of the AEKF is the tuning of the AEKF covariance matrices, multiple methods are introduced to systematically tune the covariance matrices. Experimental validation results show that the virtual torque sensor can detect the load torque with a Normalized Mean Absolute Error (NMAE) between 3 .41% to 7 .47% , depending on the sensor set. The influence of the amount of sensors used and the tuning method are also investigated.

Condition monitoring and evaluation of wind turbine drivetrain hold great importance. However, the implementation of real-time monitoring often faces challenges in efficiency and accuracy, as the drivetrain typically operates under harsh conditions. In order to resolve this, this paper proposes a vibration-based damage monitoring digital twin (VBDM-DT) that enables the online intelligent evaluation of wind turbine drivetrain. The VBDM-DT integrates a random wind load model, a high-fidelity dynamics model, and a fatigue damage model. The random wind load model takes the wind speed from the supervisory control and data acquisition (SCADA) as input to estimate the input torque of the drivetrain in real time. Simultaneously, VBDM-DT uses the vibration signals from the condition monitoring system (CMS) to intelligently calibrate the dynamics model, allowing it to be continuously adjusted and optimized in response to actual vibrations. And the fatigue damage model takes the real-time dynamic load estimated by the high-fidelity dynamics model as input to realize real-time fatigue damage monitoring of key components of the drivetrain. The VBDM-DT model is applied to a 2 MW wind turbine drivetrain to verify the effectiveness of the proposed method. In addition, a visualization platform is developed to vividly and intuitively display the real-time operating information, dynamic loads, and damage levels of the key components of wind turbine.

Recent studies have revealed the large potential of vertical-axis wind turbines (VAWTs) for high-energy-density wind farms due to their favorable wake recovery characteristics. The present study provides an experimental demonstration and proof-of-concept for the wake recovery mechanism of the novel X-Rotor VAWT. The phase-locked flowfield is measured at several streamwise locations along the X-Rotor’s wake using stereoscopic particle image velocimetry (PIV) with fixed-pitch offsets applied to the blades. The streamwise vortex system of the upper half of the X-Rotor is first hypothesized and then experimentally verified. The induced wake deformations of the vortex systems are discussed in comparison with previous studies concerning traditional H-type VAWTs. The results suggest that positive blade pitch is more favorable for accelerated wake recovery due to the dominant tip-vortex generated on the upwind windward quadrant of the cycle. Utilizing theoretical blade load variations along the span explains distinct unsteady flow features in the near wake generated at select quadrants of the rotor rotation, shedding light on the potential of the two pitch schemes.

The actuator line method (ALM) is a widely-used tool for the modelling of horizontal axis turbines and wind and tidal farm flows. The method uses a virtual blade representation to simulate the dynamics of wind turbines without the computational expense associated with resolving the blade geometry. Within the ALM the flow is first sampled at each blade section, allowing for the calculation of the sectional lift and drag forces that are in turn imposed on the flow domain using a smearing kernel. In this work the effect of the flow sampling method on the robustness of the ALM is discussed. Implementations of two widely-used types of methods for horizontal axis wind turbines are tested: point sampling, where the flow is sampled at or near the collocation point, and volume average sampling where flow field information from multiple points centered at the collocation point is averaged into a representative blade-local velocity vector. A third method, line average sampling, that was initially proposed for sampling flows from blade resolved simulations is adapted for the ALM framework. This method samples the flow symmetrically around a control element in order to eliminate the interference of the bound circulation, thus allowing the inflow velocity at the aerofoil to be determined. When evaluating power and thrust coefficients as well as the flow field at the rotor plane predicted by the ALM, the line average sampling approach is demonstrated to be more robust and converges faster with time step for two different rotors: a small wind rotor and a high solidity tidal rotor. This suggests that carefully selecting the sampling method could be key in alleviating the very strict time step restriction imposed by the ALM, a widely acknowledged limitation of the method. Such an advancement could contribute towards improving the computational efficiency and tractability of the method.

The risk of missing data and subsequent incomplete data records at wind farms increases with the number of turbines and sensors. We propose here an imputation method that blends data-driven concepts with expert knowledge, by using the geometry of the wind farm in order to provide better estimates when performing nearest neighbors imputation. Our method relies on learning Laplacian eigenmaps out of the graph of the wind farm through spectral graph theory. These learned representations can be based on the wind farm layout only, or additionally account for information provided by collected data. The related weighted graph is allowed to change with time and can be tracked in an online fashion. Application to the Westermost Rough offshore wind farm shows significant improvement over approaches that do not account for the wind farm layout information.

Previous numerical studies suggested the motions of Floating Offshore Wind Turbines (FOWTs) may enhance their wake recovery rates due to having different modes of wake dynamics from the bottom-mounted counterparts. However, the majority of previous research were conducted with models having relatively low fidelities and/or focusing on laminar inflow conditions. Models with lower fidelities are not able to capture the dynamics of tip-vorticies reliably while inflow conditions without turbulence are unrealistic out in the fields. In light of this, this paper performed high fidelity numerical simulations (large eddy simulation with actuator line technique) using full scale surging (prescribed and harmonic) FOWT rotor with different inflow turbulence intensities and multiple surging settings systematically to better understand the wake dynamics of FOWT. The results showed that the differences of wake structures between fixed and (harmonic) surging rotors were pronounced when under laminar inflow conditions, where the Surging Induced Periodic Coherent Structures (SIPCS) could be detected straightforwardly; while the differences were much less significant when under inflow conditions with realistic turbulence intensities, and the SIPCS were clearly revealed only after phase-locked averaging. Moreover, when under laminar inflow conditions, the values of mean disk-averaged streamwise velocity at x/D=8 could be above 30% larger for the surging cases than the fixed case, while the increases were down to around 0 .5∼2% when under inflow conditions with realistic turbulence intensities.

Satellite SAR provides ocean surface wind fields at 10 m above sea level. The objective is to investigate the capability of Synthetic Aperture Radar (SAR) satellite StriX observations for mapping offshore wind farm wakes. The focus is on the conditions under which an apparent wind speed surplus is generated, measured in 48% of the 67 images available. The results compare well to Sentinel-1 observations, showing a 34% surplus rate during several years based on 1171 images. Three wind speed surplus cases have been studied in detail using the mesoscale Weather, Research, and Forecasting model (WRF) model with two wind farm parametrizations. The SAR-based observations and WRF model compare for most cases, though only when Turbulent Kinetic Energy (TKE) is included in the wind farm parametrization. The TKE mixes higher momentum downwards in a stable atmosphere, causing surface wind speed surplus near the surface.

Short-term wind speed prediction is essential for economical wind power utilization. The real-world wind speed data is typically intermittent and fluctuating, presenting great challenges to existing shallow models. In this paper, we present a novel deep hybrid model for multi-step wind speed prediction, namely LR-FFT-RP-MLP/LSTM (Linear Fast Fourier Transformation Rank Pooling Multiple-Layer Perception/Long Short-Term Memory). Our hybrid model processes the local and global input features simultaneously. We leverage Rank Pooling (RP) for the local feature extraction to capture the temporal structure while maintaining the temporal order. Besides, to understand the wind periodic patterns, we exploit Fast Fourier Transformation (FFT) to extract global features and relevant frequency components in the wind speed data. The resulting local and global features are respectively integrated with the original data and are fed into an MLP/LSTM layer for the initial wind speed predictions. Finally, we leverage a linear regression layer to collaborate these initial predictions to produce the final wind speed prediction. The proposed hybrid model is evaluated using real wind speed data collected from 2010 to 2020, demonstrating superior forecasting capabilities when compared to state-of-the-art single and hybrid models. Overall, this study presents a promising approach for improving the accuracy of wind speed forecasting.

This study presents a data-driven approach to extrapolate and predict Reynolds number effects on wind turbine airfoil polars. For this purpose a database is created using experimentally obtained aerodynamic coefficients from open literature for airfoils with thickness-to-chord ratio (t/c) values in the range of 15% to 30% in order to be more relevant to wind turbine blade design applications. All available airfoil geometries are parameterized using PARSEC methodology, and a Pareto analysis is performed to understand the sensitivity of maximum lift coefficient ( C l max ) and minimum drag coefficient ( C d min ) variations to geometrical inputs as well as to the Reynolds number using statistical tools. Based on this analysis, response surfaces are generated to predict C l max and C d min of a given airfoil operating at a given Reynolds number. These predicted values are then utilized in a power-law based estimation methodology to obtain predictions for full polars, which are then compared to experimental data for selected airfoil test cases. The results show that the response surfaces generated through the current data-driven approach as well as the full polar prediction methodology show better agreement with experimental results compared to those obtained using numerical simulations-based extrapolation schemes. For airfoil types that might be under-represented in the database (such as symmetrical airfoils), the prediction results can be erroneous, especially at very high Reynolds numbers that are not in the constructed database. In the power-law based polar prediction methodology, utilizing a reference Reynolds number that is closer to the target value provides more accurate predictions for both lift and drag polars.

For many design applications in offshore engineering, including offshore wind turbine foundations, engineers need accurate statistics for kinematic and dynamic quantities, such as hydrodynamic forces, whose statistics depend on the stochastic sea surface elevation. Nonlinear phenomena in the wave--structure interaction require high-fidelity simulations to be analyzed accurately. However, accurate quantification of statistics requires a massive number of simulations, and the computational cost is prohibitively expensive. To avoid that cost, this study presents a machine learning framework to develop a reliable surrogate model that minimizes the need for computationally expensive numerical simulations, which is implemented for the monopile foundation of an offshore wind turbine. This framework consists of two parts. The first focuses on dimensionality reduction of stochastic irregular wave episodes and the resulting hydrodynamic force time series. The second of the framework focuses on the development of a Gaussian process regression surrogate model which learns a mapping between the wave episode and the force-on-structure. This surrogate uses a Bayesian active learning method that sequentially samples the wave episodes likely to contribute to the accurate prediction of extreme hydrodynamic forces in order to design subsequent CFD numerical simulations. Additionally, the study implements a spectrum transfer technique to combine CFD results from quiescent and extreme waves. The principal advantage of this framework is that the trained surrogate model is orders of magnitude faster to evaluate than the classical modeling methods, while built-in uncertainty quantification capabilities allows for efficient sampling of the parameter using with the CFD tools traditionally employed.

{The use of state estimation techniques offers a means of inferring rotor effective wind speed from standard measurements of wind turbines. Typical wind speed estimators rely upon a pre-computed quasi-steady aerodynamic mapping, which describes the relationship between pitch angle and tip-speed ratio and the power coefficient. In practice, the static mapping does not capture the influence of turbine structural dynamics and atmospheric turbulence, inevitably resulting in poor performance of the wind speed estimation. In addition, the turbine aerodynamic properties might not be easily accessible. Thus, this paper presents a rotor effective wind speed estimation method that obviates the requirement for prior knowledge of turbine power coefficients. Specifically, the proposed method exploits a simple actuator disc model, where the aerodynamic power and thrust coefficients can be characterised in terms of axial induction factors. Based on this insight and standard turbine measurements, real-time estimation of rotor effective wind speed and axial induction factors can then be achieved using a simplified turbine drive-train model and an extended Kalman filter. In addition, the actuator disc model can be updated easily over time by calibrating solely two correction factors. Thus, the proposed algorithm presents an alternative for estimating the rotor effective wind speed, which is valuable for numerous applications, for example, LiDAR-assisted control and coherence studies.

Turbine wake and local blockage effects are known to alter wind farm power production in two different ways: (1) by changing the wind speed locally in front of each turbine; and (2) by changing the overall flow resistance in the farm and thus the so-called farm blockage effect. To better predict these effects with low computational costs, we develop data-driven emulators of the ‘local’ or ‘internal’ turbine thrust coefficient CT* as a function of turbine layout. We train the model using a multi-fidelity Gaussian Process (GP) regression with a combination of low (engineering wake model) and high-fidelity (Large-Eddy Simulations) simulations of farms with different layouts and wind directions. A large set of low-fidelity data speeds up the learning process and the high-fidelity data ensures a high accuracy. The trained multi-fidelity GP model is shown to give more accurate predictions of CT* compared to a standard (single-fidelity) GP regression applied only to a limited set of high-fidelity data. We also use the multi-fidelity GP model of CT* with the two-scale momentum theory (Nishino & Dunstan 2020, J. Fluid Mech. 894, A2) to demonstrate that the model can be used to give fast and accurate predictions of large wind farm performance under various mesoscale atmospheric conditions. This new approach could be beneficial for improving annual energy production (AEP) calculations and farm optimisation in the future.