Statistical Uncertainty Analysis in Time-Domain Fatigue Assessment of Steel Risers

Abstract

This paper addresses the statistical uncertainty in long-term fatigue damage in offshore structures due to the short-term simulation length used in time domain analysis of stresses. The paper focuses on steel risers applications. A new simulation-based estimator for the variance of the short-term fatigue damage is presented. The proposed estimator is based on a variation of the original nonparametric bootstrap. It works with blocks of data instead of discrete values, in order to better account for the autocorrelation of the stress cycles in the stress time series. This versatile estimator can be applied in time-domain fatigue analyses to assess the variance of the fatigue damage using a single stress time series and does not require any previous assumptions on the stochastic stress process.