Risk analysis of derailment induced by rail breaks - a probabilistic approach

In this paper, a model is developed to analyse the risk of derailment using a probabilistic approach. The process of rail defects leading to rail breaks and further to derailment is analysed while addressing the factors relating to rail maintenance. The risk of derailment is measured with the expected number of rail breaks multiplied by severity. To evaluate the risk, two models are developed to predict the rate of occurrence of rail defects and breaks. The hazard rate of thermit weld defects is predicted using a birth process with immigration. Imperfect inspections are modelled by employing a filtered non-homogeneous Poisson process. The performance and application of the proposed model are illustrated by an example. It is shown that the risk of derailment can be significantly reduced by more frequent inspections, and grinding can also reduce considerably the risk of derailment relevant to rolling contact fatigue