Abstract. Blade erosion of wind turbines causes significant performance degradation, impairs aerodynamic efficiency, and reduces power production. However, traditional monitoring systems based on supervisory control and data acquisition (SCADA) data, which rely on operational data from turbines, lack effectiveness at early detection and quantification of these losses. This research builds on an established turbine performance integral (TPI) method with a sensor-augmented aeroelastic modelling approach to enhance wind turbine performance assessment, focusing on blade erosion. Applying this approach to a distinct multi-megawatt turbine model, the study integrates multibody aeroelastic simulations and real-world operational data analysis. The study identified readily available sensors that were sensitive to blade surface roughness changes caused by erosion. Operational data analysis of offshore wind turbines validated the initial sensor selection and approach. Refined simulations using further virtual sensors quantified the effect size of these sensors' output under different turbulence levels and blade states, employing Cohen's d – a dimensionless metric measuring the standardised difference between two means. For the turbine investigated, findings indicate that sensors such as blade tip torsion, blade root flap moment, shaft moment, and tower moments, especially under lower turbulence intensities, are particularly sensitive to erosion. This confirms the need for turbine-specific, controller-informed sensor selection and emphasises the limitations of generic solutions. This research provides an approach for bridging simulation insights with operational data for turbine-specific performance assessment, contributing to the development of condition monitoring systems (CMSs), resilient turbine designs, and maintenance strategies tailored to specific operating conditions.
Read full abstract