2003

Rosqvist, Tony

The increased use of risk assesment in governmental and corporate decision-making has increased the role of expert judgement in providing information for safety related decision-making. Expert judgements are required in most steps of risk assessment: hazard identification, risk estimation, risk evaluation and analysis of options. The use and elicitation of expert judgement is therefore subject to on-going research. Furthermore, expert judgement is also required in the quality assurance or quality verification of risk assessment. The research presented in the thesis addresses qualitative and probabilistic methods supporting the use of expert judgement in specific decision contexts; introduces a conceptual and procedural framework for quality verification of risk assessment; and presents techniques for the aggregation of probability distributions specified by experts' percentile information. The methodological view to risk assessment adopted in the thesis is requisite modelling, where a decision and risk model is developed for a certain decision context to support decision-making under uncertainty, and refined until the decision-maker has confidence in the results and prescriptions obtained from the model. The decision and risk modelling approaches presented in the thesis are related to specific decision contexts in maritime safety, maintenance management, and software reliability. The modelling approaches presented can, however, be utilised for risk-informed decision-making in other application areas as well.