Evaluation of different simplified crash models: application to the under-frame of a railway driver's cab

Due to the high cost and the practical difficulties of full scale rolling stock vehicle crash tests, numerical simulation proves to be a good alternative. Finite element codes provide very satisfying results. Nevertheless, long modelling and computation times implied by this local approach do not allow this technology to be fully efficient in the early design stages. In these cases, a more adequate solution is to use simplified models based on multibody systems. This approach requires the input of specific data for non-linear bars and springs which are determined via experiments, finite element calculations, or in a more efficient way by analytical models. In a classical way, rigid multibody systems are used. We propose here to evaluate the influence of the flexibility in simplified models, in terms of global responses and computation times. The application to an under-frame of a railway driver's cab shows that a multibody model using non-linear springs and elastic-plastic beam elements dedicated to thin-walled structures carry out the best compromise.