Identifying Critical Nodes in Fault Tree Safety Models with Limited Data

Decision makers often use fault trees to identify ways to most effectively reduce risk in a system. However, when little data are available, a large degree of uncertainty may exist in the probabilities in the tree. This paper presents a method to quantify uncertainty in a fault tree considering both statistical uncertainty (due to low observed event counts) and unavailable data (events for which no supporting data are available). The uncertainty quantification is integrated with a method to assess importance metrics associated with events in the tree. This provides decision makers with a degree of confidence in identifying the most critical nodes and in allocating resources in the most effective way. The method is applied to a case study on runway incursions for a fault tree from the Integrated Safety Assessment Model.