ABSTRACT
The proper treatment of dependency among Human Failure Events (HFEs) has been an ongoing issue in the field of Human Reliability Analysis (HRA). This issue has been recognized and acknowledged in the HRA community but has not been fully addressed. Over the past decade, the use of Bayesian Belief Networks (BBNs) has become increasingly popular in the field of Reliability and Risk analysis and it is gradually finding its way into the HRA domain. Recently, the use of BBN to model HFE dependency issue has been proposed (Mosleh et al. 2012). This paper presents the full methodology for the explicit treatment of dependencies among HFEs using the BBN model and the time slice concept of Dynamic Bayesian Networks (DBN). The BBN model contains the specific contextual factors that are common between multiple HFEs and uses these dependencies to estimate the individual conditional probabilities of those HFEs. We will use an example to demonstrate how this approach can be applied to address this long standing problem in HRA.
