Protecting oil storage tanks against floods: Natech risk assessment with imprecise probabilities

Abstract

Natechs are technological accidents that are triggered by natural disasters. The increase in the frequency and severity of climatic natural disasters along with the growth of industrialization has accelerated the demand for development of dedicated methodologies for risk assessment and management of Natechs. Due to a lack of accurate and sufficient data, risk assessment of Natechs has largely been based on subjective assumptions and imprecise probabilities, making the assessed risks and the subsequent risk management strategies deficient in terms of cost-effectiveness. In the present study, evidence theory, as an effective technique for dealing with imprecise probabilities, and Bayesian network, as an effective tool for reasoning under uncertainty, are combined to develop a methodology for risk analysis of Natechs based on imprecise probabilities with no attempt to increase the precision of the input data but the accuracy and cost-effectiveness of the outcomes. Flotation of oil tanks during floods has been considered to exemplify the methodology. The methodology is demonstrated to outperform conventional approaches where average probabilities or generic probability distributions are used instead of interval probabilities for risk assessment and management.