Assessment of stochastic passenger comfort in railway vehicles using detailed 3D multibody models

Passenger comfort in railway vehicles has traditionally been assessed through deterministic metrics such as the Sperling Index or the NMV Comfort Index. Nevertheless, as will be shown in this paper, using a single comfort value may lead to unfair vehicle comparisons. One way to overcome this issue is to enrich comfort description by treating it as a stochastic variable. While a few decades ago this randomness might not have been particularly relevant, the current trend of lightweight carbodies makes it a key factor. Recent studies have shown that the flexible modes of lightweight trains are among the main causes of discomfort, and that passengers do not behave as dead masses but rather as dynamic vibration absorbers. Both facts contribute to random variation of comfort indexes, as the accelerations from which they are obtained largely depend on the number and position of passengers sharing a car. In previous works, the authors defined a stochastic variable, that they named ‘Compound Comfort’ (CC), and proposed a methodology to infer its distribution from a minimal set of tests. However, those studies were carried out with simplified 1D models. More sophisticated 3D models are called for to confirm and refine the results and procedures presented earlier. Results using advanced virtual models obtained through VI–Rail/Simulink co–simulation will be shown in this paper. They confirm that the variance of CC is sufficiently large so as to deprive the single deterministic vehicle index of any meaning. It will also be shown that the probability density function of CC may be derived without knowledge of passenger distribution, and that it can still be easily inferred (as was the case with simplified models) from simple tests: one tare and a five half laden. We believe that Compound Comfort is a strong candidate for a future comfort standard for manufacturers and operators.