Open Collections

Woldesellasse, Haile

University of British Columbia

This dissertation proposes a comprehensive decision support framework for risk assessment of oil and gas pipelines exposed to external pitting corrosion. The study begins by performing a bibliometric analysis to identify the top cited research publications on this area. After that, several imputation and class imbalance handling techniques are proposed to handle the missing values and class imbalance that are frequently observed problems in a corrosion dataset. The usefulness of both proposed techniques was assessed using a neural network model. The model's prediction performance improved when it was trained using the cleaned corrosion dataset. Then, the risk assessment framework is developed using the corrosion dataset in a bow-tie approach, which represents an entire accident scenario starting with causes and ending with consequences. The proposed framework was developed by integrating Bayesian belief network (BBN) with a geographic information system (GIS). The framework part for evaluating the failure probability of buried pipelines is developed using three basic components: corrosion model, failure pressure model, and limit state function to estimate the probability of failure due to burst and leak. Lastly, the framework part for evaluating the consequence assessment is developed using an existing event tree and additional factors including failure probability of a pipeline, igniting probability, population density, and the presence of protected or conserved areas. The consequence of gas pipeline failure are assessed based on casualty and environmental impact. The reasoning power of the BBN and the spatial capability of the GIS combine to form a powerful tool to aid decision makers manage a spatially distributed aging pipeline. Additionally, it aids pipeline operators in making timely, and cost-effective decisions to avoid accidents in high consequence areas.

Text

2023-04-21

Attribution-NonCommercial-NoDerivatives 4.0 International

http://hdl.handle.net/2429/84396

Civil Engineering

University of British Columbia

UBCO

http://creativecommons.org/licenses/by-nc-nd/4.0/