Predicting the Long Term Distribution of Extreme Loads From Limited Duration Data: Comparing Full Integration and Approximate Methods

Abstract

In this paper, we present a methodology for proceeding from the short-term observations of extreme loads to the long-run load distribution of these extreme events, for both flap and edge loading in both operating and parked wind turbine conditions. First, a general approach utilizing full integration, where numerical routines are used to directly integrate the conditional short-term load distribution over the annual occurrence of wind speeds and turbulence intensities, is presented. Then starting from this general approach, a qualitative analysis is undertaken to explore the extent of the contribution of each of the three variables in the governing equation to the variability in the long-term extreme load distribution. From this analysis, lower-order models are considered, where instead of using the entire distribution of the variables, a constant fractile of the short-term extreme load distribution, turbulence intensity distribution, or both are used. Finally recommendations are given to guide the analyst to decide when simpler, yet robust, methods which account for sufficient variability in extreme load events may be employed with confidence.