Failure probability of corroded pipeline considering the correlation of random variables

• Published in: Engineering failure analysis 2019-05, Vol.99, p.34-45
• Main Authors: Zhang, Peng, Su, Lingbo, Qin, Guojin, Kong, Xinhai, Peng, Yang
• Format: Article
• Language: English
• Subjects: Correlation; Corroded pipeline; Failure probability; Multi-mode failure; Reliability
• ISSN: 1350-6307; 1873-1961
• DOI: 10.1016/j.engfailanal.2019.02.002

In view of the shortcomings of independence assumption for random variables in the reliability analysis of corroded pipeline, an analytical method is put forward for the reliability of corroded pipeline considering the correlation among random variables. The models of corrosion perforation, local burst and rupture as well as their composite failure probability are also established. Based on JC method and orthogonal transformation, the multidimensional normal distribution function is used to propose a computational method for the multi-mode failure probability of corroded pipeline considering correlated random variables. This paper elaborates on the correlation of pipeline reliability analysis. Meanwhile, based on case studies, a study is conducted on how the pipeline failure probability is affected by the correlation between four pairs of random variables, including pipe diameter and wall thickness, the depth and length of the defect, the radial rate and axial rate of corrosion, as well as yield strength and tensile strength. The analysis results suggest that the correlation between random variables has no influence on the failure probability of pipeline corrosion perforation. The probability of local burst, rupture and composite failure increases with the increase of correlation coefficient between the defect depth and length as well as that between the radial and axial corrosion rate, and decreases with the increase of correlation coefficient between the pipe diameter and wall thickness. The larger the correlation coefficient between yield strength and tensile strength is, the greater the probability of local burst and rupture will be, while the composite failure probability remains the same. Besides, the impact of variables correlation on the failure probability of corroded pipeline decreases as the corrosion continues. The failure probability of corroded pipeline is most sensitive to the correlation coefficient between the radial and axial corrosion rate, and least sensitive to the correlation coefficient between the yield strength and tensile strength. •An analysis method of corroded pipeline considering correlations between random variables.•The failure probability model is built for three distinctive failure modes.•A method is proposed to calculate the failure probability of corroded pipeline.•The impact of the random variable correlation on the pipeline failure probability is investigated.