A reliability-based approach to determine the probability of Loss of Containment of Dangerous Goods in a rear-end collision in the automated marshalling process, using a multi-body dynamic model

Abstract

The Dutch government provides a calculation method to determine the risks to society outside the boundaries of institutions handling Dangerous Goods (the risk affecting External Safety). The current method to calculate the risk affecting External Safety for train marshalling yards is deterministic, outdated, and not transparent. This thesis aims to improve the calculation method regarding rear-end collisions in automated marshalling. It provides a reliability-based approach to the study case Kijfhoek and uses a multi-body dynamic model to simulate the collisions. A Monte Carlo simulation is performed implementing over twenty-five parameters to determine the probability of damage in case of a rear-end collision. From the results, a simplified method is designed to assess the probability of damage to the wagon’s structure. This method uses a newly found relation to determine the absorbed energy at the impact interface in case of a collision. Furthermore, the simplified method reduced the calculation to five parameters and an analytically solvable Limit State function. A first-order reliability method shows that the impact velocity is the single most important parameter in determining the probability of damage.
The simplified method also allows for site-specific values and gives insight into the system and the most important parameters. It is recommended that this method, to determine damage to especially tank wagons in rear-end collisions, is used to improve the existing calculation method for determining the risk affecting External Safety.