A Risk Based Model for the Switch Derailment Using Bayesian Network

Safety is amongst the superiorities of the railways in comparison to other modes of transportation. Assessing the level of safety (LOS) in a railway network requires the up-to-date risk model which considers the hazards participating in an accident while analyzing the severity of their consequences. Bayesian networks are used to analyze the conditional probabilities of the chain causes and demonstrate it as a visual graphical model for the relations between the involved factors. The accuracy of calculations depends on the recorded data and the main contributor of the model validity. The switch derailment that is one of the most repeated accidents in railways is under scrutiny in this research. For the analysis different reasons separately or combined with other factors are evaluated and categorized in six severity categories based on the Iranian railway accident severity guidelines. The outcome of the model demonstrates the human factors and its subdivisions as the main factor participating in the switch derailment. The predictions are that the risk of the third grade of severity is the most probable of severities.