Skip navigation

Development of advanced reliability assessment models with applications in integrity management of onshore energy pipelines

Development of advanced reliability assessment models with applications in integrity management of onshore energy pipelines

Pesinis, Konstantinos (2018) Development of advanced reliability assessment models with applications in integrity management of onshore energy pipelines. PhD thesis, University of Greenwich.

Konstantinos Pesinis 2018 - secured.pdf - Published Version
Available under License Creative Commons Attribution Non-commercial No Derivatives.

Download (4MB) | Preview


This thesis addresses some of the predominant engineering challenges involved in the reliability-based integrity management of energy pipelines. The aim is to conjointly consider realistic safety threats and integrity management strategies, such as in-line inspections, criteria for excavation and direct assessment of energy pipelines. Towards this, advanced algorithmic model-based approaches are developed and proposed, based on fundamental principles of structural reliability analyses, stochastic degradation processes, machine learning through Bayesian statistics, multivariate data analysis, hazard modelling and interval probabilities, in an effort to quantify uncertainties that impose threats and define risks to the integrity of energy pipelines.

First, the quantification of failure probabilities for onshore gas pipelines subjected to external metal-loss corrosion is addressed. The probabilistic methodology proposed is based on a robust integration of stochastic processes within a structural reliability analysis (SRA) framework. It comprehensively accounts for the temporal uncertainty of multiple metal-loss corrosion defects and efficiently predicts long-term time-dependent reliability at the pipe segment level. The application of the methodology is illustrated through two case studies, based on two distinct inspection and maintenance strategies. In specific, an industry-consistent maintenance strategy is considered in one of them, namely External Corrosion Direct Assessment (ECDA). The reliability, originally evaluated at the segment level, is incorporated in an investigation of the influence of imperfect ECDA actions at the system level. The methodology is also applied considering a realistic maintenance and repair strategy based on in-line inspections (ILI). Again, it deals with multiple metal-loss corrosion defects, facilitates the identification of the critical ones and provides expected reliability forecasts for the whole lifecycle of the pipe segment.

Second, two distinct statistical models are proposed that can account for multiple integrity threats, since historical failures form an integral part of informed integrity management strategies. For the implementation actual incidents are employed, derived from the Pipeline and Hazardous Material Safety Administration (PHMSA) database, which contains data of incidents of existing gas transmission pipelines, providing useful insights into their state at the time of the analysis. In both statistical models, a non-repairable system approach is considered, as opposed to the repairable system approach commonly adopted in energy pipeline studies. In the first one, a well-established approach from reliability and survival analysis is employed, known as nonparametric predictive inference (NPI). This method provides interval probabilities, also known as imprecise reliability, in that probabilities and survival functions are quantified via upper and lower bounds. The focus is on the rupture of a future pipe segment, due to a specific cause among a range of competing risks. The second statistical methodology adopts a parametric hybrid empirical hazard model, complemented with a robust data processing technique, i.e. the non-linear quantile regression, for reliability analysis and prediction. It provides inferences on the complete lifecycle reliability of the average pipe segment of a region under study. For the purpose of cross-verification, the results of the second statistical model are compared with these of the second aforementioned structural reliability model, which is based on the ILI maintenance and repair strategy.

Finally, a robust methodology for estimation of small posterior failure probabilities for gas pipelines based on available inspection data is presented. The analysis of the data is based on the BUS (Bayesian Updating with Structural reliability methods), which sets an analogy between Bayesian updating and a reliability problem. The structural reliability method adopted is Subset Simulation (SuS) and the whole analysis is referred to as BUS-SuS. Two case studies are carried out to illustrate and validate the proposed methodology. In the first case study, hierarchical BUS-SuS is implemented on an existing gas pipeline containing metal-loss corrosion defects and is validated against field data. The reliability of the pipe segment is evaluated in terms of three distinctive failure modes, namely small leak, large leak and rupture. In the second case study, the Bayesian updating is conducted by using BUS-SuS in conjuction with the data augmentation (DA) technique. Simulated data, corresponding to an existing gas pipeline with high-pH stress corrosion cracking (SCC) features and constant internal pressure loading, are employed to illustrate and validate the proposed model. Furthermore, the dependence among the growths of different crack features is taken into account, using the Gaussian copula. At the end, the sensitivity of both the stochastic growth model and pipe segment reliability to different dependence scenarios is investigated. All the aforementioned proposed methodologies aim to assist pipeline operators in decision making and informed implementation of integrity management strategies.

Item Type: Thesis (PhD)
Uncontrolled Keywords: Energy pipelines; structural reliability analysis; statistical analysis; stochastic process; metal-loss corrosion; stress corrosion cracking; Bayesian updating; competing risks; historical failure data; non-repairable systems approach;
Subjects: T Technology > TJ Mechanical engineering and machinery
Faculty / Department / Research Group: Faculty of Engineering & Science
Faculty of Engineering & Science > School of Engineering (ENN)
Last Modified: 14 Aug 2019 14:27
Selected for GREAT 2016: None
Selected for GREAT 2017: None
Selected for GREAT 2018: None
Selected for GREAT 2019: None
Selected for REF2021: None

Actions (login required)

View Item View Item


Downloads per month over past year

View more statistics