In a crosswind scenario, the risk of high-speed trains overturning increases when they run on viaducts since the aerodynamic loads are higher than on the ground. In order to increase safety, vehicles are sheltered by fences that are installed on the viaduct to reduce the loads experienced by the train. Windbreaks can be designed to have different heights, and with or without eaves on the top. In this paper, a parametric study with a total of 12 fence designs was carried out using a two-dimensional model of a train standing on a viaduct. To asses the relative effectiveness of sheltering devices, tests were done in a wind tunnel with a scaled model at a Reynolds number of 1x10(5), and the train's aerodynamic coefficients were measured. Experimental results were compared with those predicted by Unsteady Reynolds-averaged Navier-Stokes (URANS) simulations of flow, showing that a computational model is able to satisfactorily predict the trend of the aerodynamic coefficients. In a second set of tests, the Reynolds number was increased to 12x10(6) (at a free flow air velocity of 30m/s) in order to simulate strong wind conditions. The aerodynamic coefficients showed a similar trend for both Reynolds numbers; however, their numerical value changed enough to indicate that simulations at the lower Reynolds number do not provide all required information. Furthermore, the variation of coefficients in the simulations allowed an explanation of how fences modified the flow around the vehicle to be proposed. This made it clear why increasing fence height reduced all the coefficients but adding an eave had an effect mainly on the lift force coefficient. Finally, by analysing the time signals it was possible to clarify the influence of the Reynolds number on the peak-to-peak amplitude, the time period and the Strouhal number.

High-speed vehicles are exposed to crosswinds that can result in the overturning of the vehicle. As part of the vehicle homologation process, the characteristic wind curve (CWC) must be calculated to determine the maximum vehicle velocity under different wind conditions. European Standard EN 14067-6 presents different approaches for studying this matter, from a two-dimensional (2D) model to a full multi-body model. The key to the problem rests in the uncertainties inherent in a multi-body model: characterizing its suspension elements and obtaining its aerodynamic coefficients. This paper compares a 2D model with a multi-body model of the same vehicle, and presents the advantages and disadvantages of each one. Results show that a 2D model is valid for graphing the CWC and sufficient for studying the basics of the overturning process. Furthermore, a wheel unloading time signal is also studied. It is concluded that the simplification process can modify the transitory response but that the maximum wheel unloading remains constant. Finally, a study of the wheel/rail contact is performed to check if the contact model has an influence on the calculation of the CWCs.