Effect of parameter uncertainty on the numerical estimate of a railway vehicle critical speed

The paper describes a joint study carried out by SNCF and Dipartimento di Meccanica Politecnico di Milano, aimed at investigating how parametric uncertainty can be treated in the framework of virtual homologation of railway vehicles in respect to vehicle dynamics. In railway vehicle sources of parameter uncertainty may arise from inaccuracy in the modelling of a vehicle component or from a scatter in the behaviour of nominally identical components, on account of the variability implied by the component manufacturing process. The approach proposed in this paper, completely new in the railway field, is to use statistical methods having different complexity (and entailing a proportional computational effort), to analyse the propagation of uncertainty from the parameters input in the vehicle mathematical model to the results of running dynamics, in terms of the assessment quantities used for vehicle homologation. The problem is treated by numerical means, being the dependency of simulation outputs from the input parameters typically non-linear, and not defined in an analytical form.