SEARCH

SEARCH BY CITATION

Keywords:

  • loads;
  • dynamics;
  • extremes;
  • extrapolation

Abstract

This paper describes the statistical identification of component loading occurring at the same instant as an extrapolated primary 50 year extreme load. As it is not feasible to simulate a full lifetime of fluctuating wind turbine loads to determine the largest extreme value, extrapolation to a 50 year or other desired long-term return period is performed based on limited simulation data. For specific wind turbine component design, it is required to identify other simultaneously applied loading corresponding to this long-term extrapolated load level. This identification is performed using statistical methods, since an appropriate loads simulation time series directly giving loads at the extrapolated load level is unavailable. The statistical behavior of each contemporaneous load in the simulation time series that was used to determine the extrapolated load is illustrated by the histogram. Since this histogram is in general skewed, the mean value of the corresponding data may not be an appropriate value for such contemporaneous loads. Instead, in this paper, a value corresponding to the mode of the estimated probability distribution function of the contemporaneous load is used. The robustness of ascertaining this most probable contemporaneous load is also investigated. The application of this process of identification of contemporaneous extreme loads is also applied to cases wherein the extrapolated load is a resultant load value, for which a suitable vector direction for the extreme resultant load is chosen. Copyright © 2008 John Wiley & Sons, Ltd.