Quantitative risk analysis in an uncertain and dynamic environment

Refaul Ferdous, Chy. Md. (Chowdhury Mohammed) (2011) Quantitative risk analysis in an uncertain and dynamic environment. Doctoral (PhD) thesis, Memorial University of Newfoundland.

[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 (46MB)

Abstract

Quantitative risk analysis (QRA) is an integral and essential part of risk analysis, which quantifies the risk of any unwanted events in industrial process facilities. However, the application of QRA in the industrial process facility is still limited. One major barrier is handling uncertainties while performing QRA using available techniques. Other important weaknesses include unrealistic assumptions and the absence of a dynamic aspect in QRA. These weaknesses undermine the credibility and utility of the output results from QRA. -- Fault Tree Analysis (FTA) and Event Tree Analysis (ETA) are two common and important techniques of QRA for evaluating the likelihoods of unwanted occurrences. Traditionally, both techniques impose two major assumptions to simplify the analysis. The first assumption is related to the likelihood values of input events, and the second assumption is concerned about interdependence of events (for ETA) or basic-events (for FTA). FTA and ETA both use crisp probabilities; however, to deal with uncertainties, the probability distributions of likelihoods of input events can be assumed. These probability distributions as well as the crisp probabilities are often hard to come by, and even if available, they are subjected to different types of uncertainties including incompleteness (partial ignorance) and imprecision. Furthermore, both FTA and ETA assume that events (or basic-events) are independent. In practice, these assumptions are often unrealistic and introduce data and model uncertainties while performing FTA and ETA. -- Bow-tie analysis has recently gained popularity as another important technique for QRA. It can combine both FTA and ETA techniques and describe the total accident scenarios for an unwanted event, also called a critical event (CE), in a common diagram with two parts: the first corresponds to a fault tree defining possible causes leading to the CE and the second represents an event tree to reach possible consequences of the CE. Unfortunately, in spite of having this feature, the application of bow-tie analysis in QRA is still limited to a graphical representation of causes and consequences for the unwanted event. -- To overcome the challenges of QRA, this research explores uncertainty handling approaches for analyzing the fault tree and event tree, which further extends to bow-tie analysis for developing a generic framework utilizing different techniques for QRA. First, fuzzy- and evidence theory- based approaches have been developed to express the uncertainties related to data and model inadequacy of input events (events or basic events) in FTA, ETA and Bow-tie analysis. Second, an updating inference comprised of another two approaches, fuzzy-bayesian and IAE (integrity of available evidence) approaches, has been developed to integrate the dynamic aspect in QRA. In addition to these approaches, a sensitivity analysis method has also been developed for bow-tie analysis to identify the important risk contributors and evaluate corresponding risk reduction. -- Applications of the developed frameworks, approaches and updating inferences have been explored in four different illustrative examples. The first example is the event tree analysis of an LPG release where the likelihoods of different outcomes of the event tree are determined in an uncertain data environment. In the second example, two separate sub-examples, i.e., fault tree of a runaway reaction and event tree of an LPG release are considered to describe the utility of the developed approaches in case of data and model uncertainties. The third example discusses the application of the developed framework and approaches for bow-tie analysis of the BP Texas city accident. In the final example, updating approaches have been used in the bow-tie analysis of an offshore oil & gas process facility. In these examples, the likelihood of occurrence has been estimated for the unwanted event, critical event and outcome events, and the important risk contributors have been also determined. The analysis of these results helps to perform a systematic QRA in uncertain and dynamic conditions, and to measure the risk and likely losses associated with an unwanted occurrence for industrial process facilities.

Actions (login required)

View Item