Loughborough University
Browse
Thesis-2007-Li.pdf (1.53 MB)

Hierarchical risk assessment of water supply systems

Download (1.53 MB)
thesis
posted on 2007-03-23, 11:18 authored by Huipeng Li
Water supply systems are usually designed, constructed, operated, and managed in an open environment, thus they are inevitably exposed to varied uncertain threats and conditions. In order to evaluate the reliability of water supply systems under threatened conditions, risk assessment has been recognised as a useful tool to identify threats, analyse vulnerabilities and risks, and select proper mitigation measures. However, due to the complexity and uncertainty of water supply systems and risks, consistent and effective assessments are hard to accomplish by using available risk techniques. With respect to this, the current study develops a new method to assess the risks in complex water supply systems by reconsidering the organisation of risk information and risk mechanism based on the concepts of object-oriented approach. Then hierarchical assessments are conducted to evaluate the risks of components and the water supply system. The current study firstly adopts object-oriented approach, a natural and straightforward mechanism of organising information of the real world systems, to represent the water supply system at both component and system levels. At the component level, components of a water supply system are viewed as different and functional objects. Associated with each object, there are states transition diagrams that explicitly describe the risk relationships between hazards/threats, possible failure states, and negative consequences. At the system level, the water supply system is viewed as a network composed of interconnected objects. Objectoriented structures of the system represent the whole/part relationships and interconnections between components. Then based on the object states transition diagrams and object-oriented structures, this study develops two types of frameworks for risk assessment, i.e., framework of aggregative risk assessment and framework of fault tree analysis. Aggregative risk assessment is to evaluate the risk levels of components, subsystems, and the overall water supply system. While fault trees are to represent the cause-effect relationships for a specific risk in the system. Assessments of these two frameworks can help decision makers to prioritise their maintenance and management strategies in water supply systems. In order to quantitatively evaluate the framework of aggregative risk, this thesis uses a fuzzy evidential reasoning method to determine the risk levels associated with components, subsystems, and the overall water supply system. Fuzzy sets theory is used to evaluate the likelihood, severity, and risk levels associated with each hazard. Dempster-Shafer theory, a typical evidential reasoning method, is adopted to aggregate the risk levels of multiple hazards along the hierarchy of aggregative risk assessment to generate risk levels of components, subsystems, and the overall water supply system. Although fuzzy sets theory and Dempster-Shafer theory have been extensively applied to various problems, their potential of conducting aggregative risk assessments is originally explored in this thesis. Finally, in order to quantitatively evaluate the cause-effect relationships in a water supply system, fuzzy fault tree analysis is adopted in this study. Results of this analysis are likelihood of the occurrence for a specific event and importance measures of the possible contributing events. These results can help risk analysts to plan their mitigation measures to effectively control risks in the water supply system.

History

School

  • Architecture, Building and Civil Engineering

Publication date

2007

Notes

Submitted for the Degree of Doctor of Philosophy from Loughborough University

EThOS Persistent ID

uk.bl.ethos.515622

Language

  • en

Usage metrics

    Architecture, Building and Civil Engineering Theses

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC