Impact of the correlation between random variables on the reliability assessment of the corroded pipelines

Abstract

In the present work, an Algerian pipeline made of API X70 steel, used for transporting gas and subjected to corrosion attack for three decades, was investigated. Using an intelligent inspection tool, the sizes and orientations of the defects were determined to analyze and improve the accuracy of reliability and/or probability of failure estimates for the damaged pipelines. A probabilistic methodology has been developed to assess the reliability index and the sensitivity of each random variable, within the limit state function, of a gas-transporting steel pipeline that has been corroded for three decades. The reliability analysis utilizes the Second Order Reliability Method, which is based on the BFGS algorithm. In addition, the corrosion defect depth, pipe wall thickness, and corrosion defect length are the most dominant variables within the limit state function, followed by ultimate stress and operating pressure, respectively. Meanwhile, for a failure scenario, thus, it is essential to include a potential parameter such as the correlation coefficient between random variables within the objective function during reliability assessment. For instance, not taking into account the correlation between random variables can lead to an overestimation of the reliability index by 8.89% for a defect depth to wall thickness ratio d/t of 20%. This overestimation increases to 19.16% when the d/t ratio increases to 70%. If the correlation coefficient is significant, it can cause an increase of 8.86% in the error of reliability index assessment for a defect with a depth to wall thickness ratio of 20%, and an increase of 14.58% for a defect with a ratio of 50%. Besides, these avoid the conservative practice resulting from the negligence of the existing correlation between random variables within the limit state function. Hence, considering the correlation coefficients during reliability assessment contributes to setting optimal maintenance schedules for pipelines made of steel under localized corrosion attack.