Accounting for rotor flexibility in MDE-based virtual sensing for OWTs: A model comparison

Offshore wind turbines (OWTs) are becoming increasingly susceptible to low-frequency excitation, making them vulnerable to fatigue damage. Virtual sensing methods like modal decomposition and expansion (MDE) are valuable for monitoring their lifetime fatigue damage. However, existing studies applying MDE typically model the rotor and nacelle assembly (RNA) as a lumped inertia, thus disregarding the effects of blade flexibility on the MDE accuracy. The present paper addresses the limitations of using a lumped inertia model for representing the RNA for MDE-based virtual sensing in monopile-founded OWTs. The primary goal is to accurately estimate load and stress histories, including fatigue damage equivalent stresses (DESs) and lifetime fatigue damage, within the full supporting structure of an OWT during normal operation. Using the dataset [1], the paper compares fatigue predictions derived from MDE moment time histories based on a lumped inertia RNA model and two different flexible rotor models. The study also considers the influence of a locked-idle rotor configuration and changes to the blade pitch angle. The study reveals that MDE predictions are particularly sensitive to blade flexibility in the tower top region. Crucially, the introduction of a flexible rotor model has significantly improved accuracy, thereby reducing the relative damage error in the tower top to below 10% in the fore-aft (FA) direction and approximately 22% in the side-side (SS) direction. This underlines the necessity of detailed, flexible rotor representation for reliable MDE-based fatigue monitoring.