Prediction of rolling contact fatigue caused by full train operation by using a digital twin method

Knowledge of quantified information on rail wear helps better scheduling of rail grinding operations and leads to increased rail life and cost-effective operation. Rail grinding activities are planned based on prior physical measurement of rail wear over a long time and known traffic haulage. The assessment does not cover the effect of newly introduced rollingstock or the upgrading of rail on the network. Simulation tools reduce the cost of physical testing. However, high-fidelity models for wear study are still computationally expensive, and single vehicles in a train are modelled to solve a specific requirement. The continuous assessment of a full train transiting along a rail network to assess rail wear has not been trialled due to the lack of a framework to translate the simulation results into outcomes in the physical world. In this paper, a method has been proposed to analyse rail wear for a train based on a full 4D simulation of a train. In a 4D simulation, the space variables in three dimensions (longitudinal, lateral, and vertical vectors) are measured along with time or track position as the fourth dimension to allow predictions over time. A digital twin of a full train and the complete track structure has been used to predict rail wear for a full train. A case study has been used to demonstrate the application of the method and results are discussed. Rail wear induced by locomotives and wagons in a train has been compared. It is found that rail wear can vary depending on vehicle type, vehicle position, and driving strategy. The variation in rail wear due to vehicle dynamics has been discussed.