Event tree and fault tree analysis in tunneling with imprecise probabilities

This paper presents the methodologies of event tree analysis and fault tree analysis in tunneling by using imprecise probabilities. In traditional risk assessment and analysis, uncertainties are measured by precise probabilities. However, due to the limited information in geological and underground conditions, assigning a precise value as to the probability of an event may not be practical. Probability is often evaluated imprecisely in tunneling. The International Tunnelling Association has published guidelines for tunneling risk management, in which the likelihood of occurrence is recommended to be evaluated by several predefined intervals rather than a crisp probability. The goal of the methodologies proposed in this study is to deal with imprecise information without forcing the experts to commit to assessments that they do not feel comfortable with or the analyst to pick a single distribution when the available data does not warrant such precision. Several case histories of risk analysis in tunneling were revisited by using the methodologies developed in this study. All results obtained based on imprecise probabilities are discussed and compared with the results from precise probabilities. © 2012 American Society of Civil Engineers.