This study explores optimization-based strategies for the design of morphing wind turbine blades to adapt to changing wind conditions. The suction surface of the NACA-63418 airfoil is optimized through a Genetic Algorithm, utilizing the ANSYS Design of Experiments toolkit for control of the algorithm’s execution and Fluent solver (ANSYS suite) for evaluation of the objective function. Optimization process was conducted at angles of attack (AoA) of 0° and 8°, under inflow wind speeds of 6 m/s, 8 m/s, and 10 m/s to reflect prevailing wind conditions in South Africa. The resulting optimum geometries differed between the two angles of attack, underscoring the importance of AoA in airfoil optimization. Suction surface modifications yielded improvements in lift-to-drag ratio of 9.6%, 11.1%, and 1.2% at the respective wind speeds. These optimized profiles are subsequently integrated into the design of a three-dimensional rotor using Q-Blade, an open-source Blade Element Momentum Theory code, from which power curves are derived. Morphing schemes are evaluated across wind speed conditions ranging from 4–16 m/s, and performance is benchmarked against a baseline rotor employing the rigid NACA-63418 airfoil. Results demonstrate a 6.9% increase in Annual Energy Production for rotor when morphing airfoil strategies are used, highlighting the potential of optimization-driven morphing blade design to enhance wind turbine power output.
To optimise the economic and environmental performance of offshore wind turbines, it is necessary to know and predict the material health of the blades in order to extend their service life. When exposed to an aggressive environment, the materials constituting a complex structure such as a blade gradually degrade and become damaged. We focus here on the sandwich material used in the manufacture of the blade shell. During accelerated water ageing by immersion in seawater heated to 40°C, the material health of batches of test specimens is assessed at regular intervals using four-point mechanical bending tests with acoustic emission monitoring of internal damage. Over 88 weeks, the study reveals a three-phase evolution in damage resistance, correlated with acoustic emission signatures. Phase I: during water absorption until saturation, stiffness decreases by approximately 15%, while changes in the acoustic signatures reflect only modifications in acoustic impedance, related to water saturation of the material porosity, with no detected ageing. Phase II: although the mechanical properties remain stable, the acoustic signatures indicate ageing, with a weakening of interfacial strength. Phase III: structural failure is marked by a sudden drop in mechanical properties, and the acoustic signatures reveal a significant weakening of the material resistance to interfacial and volume damage. The main cause is associated with the polyester matrix, which undergoes plasticisation and hydrolysis.
Co-located wind and solar energy facilities provide more reliable power production due to the complementary nature of wind and solar resources. One potential design consideration for such facilities is the possibility that ice pieces released from turbine blades may impact solar panels and cause damage. This paper describes a novel methodology for quantifying the risk of ice throw impacts on solar panels at co-located wind and solar energy facilities. First, an ice throw model is used to predict impacts via Monte Carlo simulation. A probability density function is fitted to the Monte Carlo simulation results and integrated within the bounds of the solar panel locations to determine the probability of panel impact. Local climate data is used to determine the expected number of panel impacts each year over an entire facility. The methodology is demonstrated for a single example turbine, showing that impact risk is a function of panel distance and direction to a turbine, icing frequency, and panel area, among other factors. Results for a notional facility consisting of multiple turbines and PV panel arrays demonstrate that PV panel arrays located within a setback distance of 1.1 times the tip height of the turbine generally exhibit impact risks between 1 in 1,000 years and 1 in 10,000 years. The proposed method may be used by hybrid wind-solar energy facility designers to assess ice throw risk and optimize the site layout accordingly.
Wind power forecasts can have low average error while producing negative outputs or values above a documented turbine curve. This study presents Physio-S3M, an LIF-gated selective state-space forecaster with a horizon-specific cap inferred solely from information available at issue time. After chronological development on public T1 SCADA data, six frozen checkpoints were transferred without adaptation to ENGIE turbine R80711. Across 7625 inferential origins, Physio-S3M reduced the combined negative-or-above-curve rate from 60.20% to 52.68%, a 12.50% relative reduction, while macro nRMSE improved from 0.2189 to 0.2112. Against the same reference predictions clipped to rated power, the reduction remained 10.38% (95% confidence interval, 8.20–11.50%). Controlled ablations showed that removing output bounding, cap supervision, or the pinball loss increased the combined diagnostic rate. Removing the inter-horizon loss increased forecasting error without a clear change in that rate. The diagnostic reduction was strongest at 6 h (30.05%) and weakened at 72 h (4.24%). Because the model uses historical SCADA without issue-time numerical weather prediction, the evidence favors further evaluation for short-term forecasting that may support dispatch; multi-day weather changes remain a limitation. These exploratory results concern one external turbine-year and a transferred reference curve. They do not establish physical feasibility, operational safety, broad generalizability, or measured energy efficiency.
This paper presents a comprehensive review of the fault detection and diagnosis (FDD) techniques that exist for wind turbine systems, which are classified into three categories: model-based methods, signal-based methods, and knowledge-based methods. The analysis of the reviewed papers shows that the performance of the diagnosis process is the highest in the case of advanced AI models, especially deep learning hybrid models. CNN-LSTM with attention mechanism achieves an accuracy of 95% for converter fault detection, one-dimensional CNN based on acoustic vibrations achieves an accuracy of 96.8%, Bayesian-GAN-LSTM achieves an accuracy of 97.6% for bearing fault diagnosis, and Graph LSTM achieves an accuracy of 95.2% for multi-component fault diagnosis. The accuracy of the tree-based random forest models for gearbox temperature-based classification is 86% while the enhanced random forest models on SCADA derived features is 99.72%. The multi-scale dense fusion networks and synchronized wavelet transform-CNN models achieve gearbox fault and planetary gearbox fault accuracy larger than 98%. Even with these developments, there are some key challenges: the lack of standardized benchmarking procedures, low model generalizability across a range of operating conditions, and computational limitations in real-time applications. The findings have led to this review concluding that data-driven and hybrid approaches are the most promising diagnostic capabilities that may be integrated into wind turbine condition monitoring systems, and that future research directions to make these systems more robust, more interpretable, and more deployable in practice include physics-informed learning, federated learning architectures, multi-modal data fusion, and explainable AI systems.
