Modeling rare events under uncertainties

El-Gheriani, Malak Ali (2017) Modeling rare events under uncertainties. Masters thesis, Memorial University of Newfoundland.

[img] [English] PDF - Accepted Version
Available under License - The author retains copyright ownership and moral rights in this thesis. Neither the thesis nor substantial extracts from it may be printed or otherwise reproduced without the author's permission.

Download (2MB)

Abstract

The development of the oil and gas industry is accompanied by high risks that increase the potential for major accidents. Improving safety through implementing safety measures maintains the risk within an acceptable level and helps to prevent the occurrence of accidents. Identifying and treating uncertainty is the main challenge in performing risk analysis. This uncertainty reflects the lack of information about the accident scenario and its potential causes, as well as the absence of a modeling technique used to model accident scenarios. In most situations, there are either few or no data available to perform risk analysis. Gathering the required data from other relevant sources is one of the solutions to overcome this challenge. In the presented work, the first part of the developed methodology considers Hierarchical Bayesian Analysis (HBA) as a robust technique for an event’s frequency estimation using data collected from several sources. Results demonstrate the power of HBA in treating the uncertainty within the gathered data and providing the appropriate estimation of an event’s frequency. The estimated event’s frequency is then integrated into Bowtie (BT) analysis, one of the modeling techniques, in order to predict the occurrence of a major accident. Due to their limitations, the standard modeling techniques are unable to capture the variation of risks as changes take place in the system. Therefore, their results involve a degree of uncertainty, considered as model uncertainty. In the second part of the presented study, the developed methodology has been improved by integrating HBA and Bayesian Network (BN) into one framework to cope with data and model uncertainties simultaneously. HBA handles the uncertainty within the multi-source data, while BN is used to model the accident scenario in order to treat model uncertainty. Using HBA along with BN provides more accurate estimations and better handling of uncertainties.

Item Type: Thesis (Masters)
URI: http://research.library.mun.ca/id/eprint/12817
Item ID: 12817
Additional Information: Includes bibliographical references.
Keywords: Modeling, Rare Events, Uncertainties, BN, HBA
Department(s): Engineering and Applied Science, Faculty of
Date: October 2017
Date Type: Submission
Library of Congress Subject Heading: Oil industries--Risk management--Mathematical models; Uncertainty--Mathematical models; Risk--Mathematical models; Bayesian statistical decision theory

Actions (login required)

View Item View Item

Downloads

Downloads per month over the past year

View more statistics