Over the centuries, the assessment of risk has become an integral part of the decision process. Assessment techniques have developed to meet different applications, but all have problems and none is entirely suited to the assessment of risks relating to fire. This dissertation examines the development of risk assessment processes and frameworks, identifying common features and problems and key differences in approach. Despite generically similar approaches, different applications have led to the development of many different paradigms, none of which appears to be entirely suitable for application to building and infrastructure fires. Current fire risk assessment methods which incorporate important advances in fire modelling and Monte Carlo simulation, rely on a fire engineering approach. They tend to consider only the limited range of fire safety systems that are directly involved in construction, failing to address many of the procedural and other activities that can overwhelm traditional controls, and taking insufficient account of interactions between different controls and the factors that influence them. Further, comparative risk levels are generally evaluated against the ill-defined scenario of current practice, as defined in outdated prescriptive regulations. The result is that catastrophic consequences continue to occur, despite the presence of traditionally accepted controls. The problem is to find a framework that evaluates the sensitivity of levels of risk in fire against a defined, uncontrolled state, taking into account the effects of a comprehensive range of factors and controls. A new approach to risk assessment that addresses a comprehensive set of factors and controls and evaluates the event without, and with, controls, is considered. The framework, together with the steps for its implementation, appears to provide a versatile and flexible method of risk assessment. It is likely that the framework can be applied to all risk assessment situations. A study is undertaken to investigate the impact of factors and interactions that are not commonly taken into account in fire risk assessment. The chosen situation is a fire in the driver???s cab of a train. Current driver procedures are examined, and fire growth rates for specified materials are considered. Using the fire spread model CFAST, conditions in the cab for a range of ventilation conditions and fire growth rates are calculated. Threshold levels that determine response times for engineered and human controls and tenability, and common factors that influence consequences, can play a critical role in modelling the decision process. A driver???s decision model is proposed that determines the impact of the driver???s decisions to adjust ventilation by opening or closing windows and doors, and to extinguish the fire. The model takes into account time to respond and time to perform the necessary activities. The study shows that, even with a limited choice of actions, the decisions of the driver can have a critical effect on the outcome of a fire in the driver???s cab, altering the situation from a controlled to an uncontrolled state. Recommendations are made for further development of the new risk assessment framework, and for the development of fire modelling for risk assessment purposes. Finally, recommendations are made for the continuation of the development of the train driver response model that would result in the generation of driver decision support software.
| Identifer | oai:union.ndltd.org:ADTP/258530 |
| Date | January 2004 |
| Creators | Blackmore, Jane, Safety Science, Faculty of Science, UNSW |
| Publisher | Awarded by:University of New South Wales. School of Safety Science |
| Source Sets | Australiasian Digital Theses Program |
| Language | English |
| Detected Language | English |
| Rights | Copyright Jane Blackmore, http://unsworks.unsw.edu.au/copyright |
Page generated in 0.0022 seconds