Bayesian analysis of small probability incidents for corroding energy pipelines. (15th June 2018)

• Record Type: Journal Article
• Authors: Pesinis, Konstantinos; Tee, Kong Fah
• Is Part Of: Engineering structures. Volume 165(2018)
• Page Start: 264
• Page End: 277
• Publication Date: 2018-06-15
• Subjects: Pipelines -- Bayesian updating -- Subset simulation -- Structural reliability -- In-line inspection; Structural engineering -- Periodicals; Structural analysis (Engineering) -- Periodicals; Construction, Technique de la -- Périodiques; Génie parasismique -- Périodiques; Pression du vent -- Périodiques; Earthquake engineering; Structural engineering; Wind-pressure; Periodicals; 624.105
• DOI: 10.1016/j.engstruct.2018.03.038
• Languages: English
• ISSNs: 0141-0296

Abstract

Highlights: The growth of the corrosion defects is modelled through an hierarchical Bayesian framework. The methodology is based on the BUS (Bayesian Updating with Structural reliability method). The structural reliability method adopted herein is Subset Simulation. Corrosion data obtained from multiple in-line inspections of an underground pipeline are used. In-line inspections associated measurement errors are comprehensively considered. Abstract: This paper presents a new methodology for estimation of small posterior failure probabilities for underground energy pipelines, based on external corrosion inspection data. The analysis of the data is based on the BUS (Bayesian Updating with Structural reliability methods) that sets an analogy between Bayesian updating and a reliability problem. The structural reliability method adopted herein is Subset Simulation (SuS) and the whole analysis is referred to as BUS-SuS. Corrosion data obtained from multiple in-line inspections (ILI) of an underground natural gas pipeline are used to illustrate and validate the proposed methodology. The growth of the corrosion defects is modelled through an hierarchical Bayesian framework and the ILI associated measurement errors are comprehensively considered. Through this efficient method, it is ensured that the final samples have reached the posterior distribution. It is also more advantageous over other methods typically employed for Bayesian analysis of corroding pipelines, because it allows the estimation of small posterior failure probabilities directly within the same framework. The proposed methodology can be incorporated in a reliability-based pipeline integrity management program to assist engineers in selecting suitable maintenance strategies.