Bayesian network-based probability analysis of train derailments caused by various extreme weather patterns on railway turnouts

Since multiple failure events associated with derailments could not be identified and derailment probability could not be reached quantitatively by event tree and fault tree analysis for safety assessment in railway systems, applications of Bayesian network (BN) were introduced over the last few years. The applications were often aimed at understanding safety and reliability of railway systems through various basic principles and unique inference algorithms focusing on particular railway infrastructures. One of the most critical engineering infrastructure, railway turnouts (RTs) have been investigated and analysed critically in order to develop a new BN-based model with unique algorithm. This unprecedented study reveals the causal relations between primary causes and the subsystem failures, resulting in derailment, as a result of extreme weather-related conditions. In addition, the model, which is designed for rare events, has been proposed to identify the probability and underlying root cause of derailment. Consequently, it is expected that various weather-related causes of derailment at RTs, one such undesirable event, which can result, albeit rarely, damaging rolling stock, railway infrastructure and disrupting service, and having the potential to cause casualties and even loss of life, are identified to allow for smooth railway operation by rail industry itself. The insight into this weather-derailment will help the industry to better manage railway operation under climate uncertainty