A Bayesian analysis of component life expectancy and its implications on the inspection schedule. (May 2017)

• Record Type: Journal Article
• Authors: Mason, Paolo
• Is Part Of: Reliability engineering & system safety. Volume 161(2017)
• Page Start: 87
• Page End: 94
• Publication Date: 2017-05
• Subjects: Degradation modelling -- Markov Chain Monte Carlo -- Approximate Bayesian Computation -- Risk-informed decision making; Reliability (Engineering) -- Periodicals; System safety -- Periodicals; Industrial safety -- Periodicals; Fiabilité -- Périodiques; Sécurité des systèmes -- Périodiques; Sécurité du travail -- Périodiques; 620.00452
• DOI: 10.1016/j.ress.2017.01.006
• Languages: English
• ISSNs: 0951-8320

Abstract

Abstract: A model of crack initiation and residual component life is fitted to the inspection history, inclusive of two in-service failures, of a set of gas circulator impellers at two UK power stations. The model is then used to estimate the probability of future in-service failure of each item in scenarios in which the next opportunity for inspection (i.e. detection of a developing crack) is exploited or forgone. The study takes into account in exact manner both variability and uncertainty. A novel Approximate Bayesian Computation (ABC) methodology is introduced for the quantification of uncertainty, i.e. the sampling of the posterior distribution of the parameters of the model. At the price of the discretization of this distribution, the methodology represents a quickly implemented option in problems where standard ABC rejection sampling is unacceptably inefficient. Abstract : Highlights: A model of crack initiation and residual component life is introduced. The model is fitted to the inspection log of a set of gas circulator impellers. The odds of future failure are computed under different inspection schedules. This prediction informs the choice to perform or forgo future inspection. A novel Approximate Bayesian Computation algorithm is introduced and validated.