To date, despite advancements in the design of offshore wind turbines, the as-designed and identified as-built natural frequencies of offshore wind turbines still show discrepancies. These discrepancies are partially rooted in modelling uncertainties, as well as uncertain input parameters, related to e.g. aero-, fluid- or soil-structure interaction. The first objective of this article is to present a wind farm wide comparison of the first and second, modelled and identified, fore-aft natural frequencies for turbines in parked conditions for a wind farm located in the Belgian north sea. Secondly, the effect of different model parameters on the computed natural frequencies will be assessed using wind farm wide sensitivity studies, with the aim to describe the potential of each considered parametrization in reducing the discrepancy between modelled and measured resonance frequencies. The in-depth considered parametrizations are aimed at assessing the effect of the linearization of the p-y curves, soil stiffness, local scour as well as the mass of the rotor nacelle assembly, whereas results for wall thickness, marine growth, added mass coefficient and sea water level variations will be presented without further discussion. In order to perform this study, turbine specific finite element models have been prepared and verified based on detailed design documents; subsequently updated best-estimate soil data has been used to model the foundation for two different design scenarios. Furthermore, modal parameters have been identified for each turbine, based on vibration data collected in parked condition and state of the art operational modal analysis tools. The results show that the discrepancies between the modelled and identified first fore-aft natural frequencies could potentially be bridged by adjusting combination of the investigated parameters, whereas the discrepancies observed on the second natural frequency cannot be bridged by making changes to the investigated parameters. As such, future work will entail a more detailed investigation on modelling uncertainties.
Read full abstract