Presentations

Abstract

In a fast-evolving sector with a growing installed capacity, the wind industry is continuously seeking ways to increase energy yield and maximise returns on the already operating wind assets. Wind turbine upgrades, such as advanced control algorithms or hardware retrofits, are an increasingly popular option to improve energy production. However, these enhancements increase the Annual Energy Production (AEP) by narrowing the operational margins of wind turbines, which can impact availability due to added strain on components or by not being well adapted to the specific local wind conditions.  This study presents an innovative methodology for accurately evaluating the impact of the upgrade on AEP, while also considering the potential trade-offs between performance, operating conditions and turbine availability. The methodology is based mainly on SCADA data analysis and is split into two independent parts: the Performance impact quantification and the Turbine Operation analysis.  Quantifying performance on operating wind farms is challenging, especially after an upgrade, as it alters turbine operation and interactions with external conditions, making it hard to isolate true performance gains. In order to tackle this, a normalisation of SCADA measurements is performed based on multiple wind speed and power predictive models: detecting and correcting any changes in the wind sensor, in the measurements and on turbine operation allows for a reliable comparison of performance before and after the upgrade. The performance is assessed using complementary methods including power curve, power to power comparison and power prediction machine-learning models. Turbine Operation analysis is performed by reverse-engineering the key control laws of the turbine (pitch, yaw and rotor speed) and comparing the control parameters before and after the upgrade. The control parameters are calculated with a combination of machine-learning and statistical regression models. To complete the analysis, turbine logs are also integrated to detect any increase in frequency of events that might force a turbine stop or increase wear on components.  This innovative methodology was applied in multiple cases of turbine upgrades on operating wind farms. Results showed that for the wide majority, the impact in the AEP was positive but lower than initially expected: they ranged from 0.5% for smaller upgrades up to 4% gains with vortex generators. A particular focus is provided on cases where the upgrades had negative effects on turbine operating including significant increase of turbine stops from higher vibrations caused by higher rotor speeds and more frequent yaw activations created by the upgrade.  The presented work provides a reliable and comprehensive method to fully assess the impact of turbine upgrades focusing on both the AEP and the overall turbine operation for effective decision-making.