Statistical modeling of SCADA data for reliable wind resource assessment: The case of Leitwind wind turbine

In the context of wind energy expansion, wind resource assessment is traditionally based on meteorological masts, which require extensive pre-installation campaigns and involve substantial costs. Furthermore, their deployment is limited in spatial coverage, as masts are often dismantled after the completion of measurement campaigns, resulting in temporally and spatially constrained site data. Recent approaches have explored the use of Supervisory Control and Data Acquisition (SCADA) systems for project monitoring and development, owing to their greater availability and the existence of longer-term datasets. To date, SCADA data have been primarily employed for performance evaluation and anomaly detection, with limited application in wind resource assessment due to data confidentiality and access restrictions. Moreover, wind speed and direction measured at the nacelle are inherently affected by rotor-induced flow distortion, nacelle geometry, and turbine operating conditions (such as rotor speed, pitch angle, and yaw misalignment), which introduce measurement biases in resource assessment. In this regard, the adoption of data-driven and machine learning methodologies is increasingly gaining interest in the research community, as these approaches are capable to predict the wind flow distorsion and provide systematic correction of nacelle-based measurements. This thesis aims to address this state-of-art gap by investigating site assessment criteria based on the deviation between free-stream wind characteristics and nacelle-based measurements. Utilizing SCADA data and available type testing measurements conducted upstream of selected Leitwind wind turbines, the objective is to develop and validate a statistical correction model of nacelle-based wind data. The correction method is intended to improve the accuracy of wind assessments for future wind energy projects, reducing the uncertainty in energy yield estimation and site suitability analysis. The methodology includes a quantitative comparison of free-stream and nacelle-based wind parameters, the definition of a correction function based on SCADA variables and the validation of the proposed model through uncertainty analysis.