Statistical Uncertainty Assessment in Time-Domain Fatigue Analysis of Steel Risers
Rodrigo Carneiro da Costa1; Luis Volnei Sudati Sagrilo1
1 Universidade Federal do Rio de Janeiro - COPPE/UFRJ
doi:10.20906/CPS/CILAMCE2015-0224
Resumo
The structural safety of risers is one of the major concerns related to offshore oil exploration. Risers are complex structures and proper models for determining their long-term structural behavior are difficult to obtain. Due to this limitation, most fatigue analyzes rely on single-point estimates and safety factors to determine the fatigue life of risers. The problem with single-point estimates is that they do not contain any information about the uncertainty of the outcomes. The simplest way to avoid the problem with single-point estimates would be solving the problem several times, under the same conditions, in order to capture its intrinsic randomness. Nonetheless, this alternative is unrealistic since it would reflect into over-costly analyzes. For this reason, the current literature covering this subject present a few alternative methods to assess the variance in the long-term fatigue damage based on limited information. Among them, the methods proposed by Low (2012), and Vasconcelos Neto (2009) and Steinkjer et al (2010) are of especial relevance. The downside of these methods is that Low's method is limited to Gaussian narrowband process, while the method proposed by the other two authors do not contemplate the stress process autocorrelation, what may lead to unreliable estimates. Considering the aforementioned, this work seeks to develop a method to estimate the variance of the long-term fatigue damage that requires a single stress time series, considers the process autocorrelation and has wide applicability. Thus, two methods are proposed here. The first of these methods is based on a variation of the block bootstrap method. The idea behind it is resampling blocks of data from a single sample, in order to properly account for the autocorrelation of the process. This method is here called cycle-counting block bootstrap (CCBB) estimator. The second proposed method is a simplified manner of facing the block data used in the CCBB. Instead of using a bootstrap estimator, the second method treats each bl
Palavras-chave: Statistical uncertainty; Fatigue analysis; Steel risers