Answers are grounded in this paper and cite their sources.
A computational tool for reliability analysis of corroded pipelines is proposed, which accounts for uncertainties in inspection-collected data. The tool combines an efficient mesh generator, an in-house finite element-based plasticity software, and four different options for reliability analysis. Failure pressure results obtained with the in-house software were first benchmarked against a widely used commercial finite element package and experimental data. Then, the sensitivity of the probability of failure to the geometry of corrosion-related defects was investigated in two reliability analysis studies. Initially, semi-empirical expressions were used as failure functions for the comparison of five different approaches: traditional Monte Carlo (MC), importance sampling (IS), first-order reliability method (FORM), subset simulation (SS) and the successive Pareto simulation (SPS), the latter proposed by the authors. SPS has the advantage of delivering results that are as accurate as MC for problems with capacity and demand variables, while significantly reducing the number of function evaluations. Significant reductions in the number of function evaluations were achieved for both SPS and IS. SPS proved to be as accurate as MC for the corroded pipeline problem. As expected, FORM was more efficient than the MC-based approaches, but it was also less accurate. Finally, a reliability analysis using a finite element-based failure function was performed with SPS and FORM only. It was found that these two methods produced similar results, but some differences were noted. SPS is recommended, as it provided the best accuracy-speed balance.
No takes yet. Share an insight, caveat, or question.
Motta et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: