Establishing robust short-term distributions of load extremes of offshore wind turbines

Abstract

A novel method with a rigorous theoretical foundation is proposed for establishing robust short-term distributions of load extremes of offshore wind turbines. Based on the wind turbine load time series, the proposed method begins with incorporating a declustering algorithm into the peaks over threshold (POT) method and searching for an optimum threshold level with the aid of a Mean Residual Life (MRL) plot. Then, the method of L-moments is utilized to estimate the parameters in the generalized Pareto distribution (GPD) of the largest values in all the selected clusters over the optimal threshold level. As an example of calculation, an optimal threshold level of the tower base fore-aft extreme bending moments of the National Renewable Energy Laboratory (NREL) 5-MW OC3-Hywind floating wind turbine has been obtained by utilizing the novel method. The short-term extreme response probability plots based on this optimal threshold level are compared with the probability plots based on the empirical and semi-empirical threshold levels, and the accuracy and efficiency of the proposed novel method are substantiated. Diagnostic plots are also included in this paper for validating the accuracy of the proposed novel method. The method has been further validated in another calculation example regarding an NREL 5-MW fixed-bottom monopile wind turbine.