Uncertainty-based mode selection for closely spaced modes in operational modal analysis of a wind turbine during assembly

Abstract

This study investigates the dynamic behaviour of an onshore wind turbine tower throughout the entire assembly process, with particular emphasis on the challenge of identifying closely spaced modes in operational modal analysis. A comprehensive measurement campaign, involving up to 23 accelerometers, was conducted to capture vibration responses under environmental excitation across seven construction stages. Modal parameters and their associated uncertainties were identified using the covariance-driven stochastic subspace identification (SSI-COV) within a robust operational modal analysis scheme. A novel uncertainty-based mode selection approach was introduced and applied to reliably extract the modes of the first two bending mode pairs. Additionally, a bending mode indicator was developed to assess the purity of the identified bending shapes. The results show that the natural frequencies of the first bending mode pair decrease continuously as construction progresses. A similar trend was observed for the fore-aft mode of the second bending mode pair, while the side-to-side mode remained largely unaffected. Modal contribution analysis reveals that, particularly in the later assembly stages, the structural dynamics are dominated by the first bending mode pair. These findings highlight the effectiveness of the uncertainty-based mode selection framework and provide new insights into the dynamics of wind turbine towers during assembly. The results support model validation for structural health monitoring and may inform future design and construction practices.