Railway bogies of the Y25 family are the most common type of freight bogies used in Europe. However, this family of bogies presents poor curving performance and a derailment history. Although numerical simulations are a powerful tool to study railway dynamics, derailment scenarios involve complex operational and vehicle conditions that increase the multivariate aspect of the problem. Thus, simulating all possible scenario variants is unfeasible from the computational perspective due to time and computational constraints. This paper proposes an approach based on a Response Surface Methodology to study the combined influence of uncertain parameters on the derailment potential of a railway vehicle. The potential of this approach is demonstrated using a real derailment as a case study. A set of scenarios is identified using a Design of Experiments approach, and is simulated on a commercial software. The response functions of the quantities used to assess derailment are generated, and the conditions that maximise the derailment potential are identified. The results reveal a combination of asymmetric loading, excessive speed, and Lenoir link failure may cause extreme wheel unloading in the study developed. This work reveals the advantages of a Response Surface Methodology to identify the conditions that maximise the derailment potential.

Unlike regular automotive vehicles, which are designed to travel in different types of roads, railway vehicles travel mostly in the same route during their life cycle. To accept the operation of a railway vehicle in a particular network, a homologation process is required according to local standard regulations. In Europe, the standards EN 14363 and UIC 518, which are used for railway vehicle acceptance, require on-track tests and/or numerical simulations. An important advantage of using virtual homologation is the reduction of the high costs associated with on-track tests by studying the railway vehicle performance in different operation conditions. This work proposes a methodology for the improvement of railway vehicle design with the objective of its operation in selected railway tracks by using optimisation. The analyses required for the vehicle improvement are performed under control of the optimisation method global and local optimisation using direct search. To quantify the performance of the vehicle, a new objective function is proposed, which includes: a Dynamic Performance Index, defined as a weighted sum of the indices obtained from the virtual homologation process; the non-compensated acceleration, which is related to the operational velocity; and a penalty associated with cases where the vehicle presents an unacceptable dynamic behaviour according to the standards. Thus, the optimisation process intends not only to improve the quality of the vehicle in terms of running safety and ride quality, but also to increase the vehicle availability via the reduction of the time for a journey while ensuring its operational acceptance under the standards. The design variables include the suspension characteristics and the operational velocity of the vehicle, which are allowed to vary in an acceptable range of variation. The results of the optimisation lead to a global minimum of the objective function in which the suspensions characteristics of the vehicle are optimal for the track, the maximum operational velocity is increased while the safety and ride quality measures of the vehicle, as defined by homologation standards, are either maintained in acceptable values or improved.