ABSTRACT
With the ultimate goal of extending the lifetime of offshore wind turbines, applying monitoring data to develop fatigue damage calculations is of great interest. However, uncertainty in recorded loads and potential underestimation must be considered. This study analyzed the interface loads of a wind turbine in the North Sea using three years of strain and SCADA (Supervisory Control And Data Acquisition) data to predict lifetimes deterministically and probabilistically. The deterministic method calculated short-term damages through Rain-flow cycle-counting and Palmgren-Miner rule. While the first probabilistic method additionally used bootstrapping to obtain confidence intervals, the second, binned short-term damages in wind speed groups. As the chronological order of load cycles is lost by constructing datasets from bootstrapping or binning, to account for the effect of Low-Frequency Fatigue Dynamics (LFFD), an amplifier factor from (Sadeghi et al., 2022) was applied to the total damage. Finally, the deterministic/probabilistic methods with/without LFFD impact were compared.
