LAETA, INEGI, Faculdade de Engenharia, Universidade do Porto, Porto, Portugal
Palavras-chave:
reliability analysis, corroded pipelines, selective Monte CarloResumo
In the present paper assessment of corroded pipelines taking into account uncertainties are considered. The possibilities for pipelines assessment, after inspection, are standards, and numerical simulations. Here, a computational tool to perform the reliability analysis of corroded pipelines is created. The tool combine together an efficient mesh generator, an in-house finite element (FE) based plasticity software and four different options for reliability computations. The corroded pipelines are modeled by axisymmetric finite elements. Deterministic validation studies are conducted considering eleven specimens from literature. Then, reliability analysis (RA) studies to assess the sensitivity of the reliability index against the geometry of corrosion-related defects are performed. This later is also accomplished with the purpose to find out the most efficient RA procedure from different methodologies such as traditional crude Monte Carlo (MC), Importance sampling (IS), selective Monte Carlo (SMC) and the first-order reliability method (FORM). To conduct this comparison, initially, semi-empirical equation from the ASME B31G standard are used. As typical, large number of function evaluations were needed for MC simulation method. Big reductions were achieved for both IS and SMC. However, SMC prove to be as accurate as CMC and simpler than IS. FORM method has demonstrated to be very computationally efficient when compared to the other approaches, slightly less accurate than MC based methods. Due to this observed behavior, FE studies were performed on SMC and FORM only. Good comparison was found. Recommendation is the use of the SMC in the assessment of corroded pipelines as it provided the best accuracy-speed compromise.