In this paper, a numerical study regarding the influence of the wind speed and orientation on the train running safety against crosswinds is presented. The wind profiles are defined through the Chinese Hat Model taking into account multiple combinations of vehicle speed, wind speed and wind orientation, resulting in 180 dynamic analyses solved using an in-house developed software named Vehicle-Structure Interaction analysis (VSI). The running safety is evaluated by analysing three safety criteria based on factors related to the vehicle contact forces: the Nadal criterion, the Prud’Homme criterion and the Unloading criterion. Moreover, the behaviour of the vehicle undergoing different wind profiles is assessed throughout a 1400 m straight track section with real irregularities of one of the Portuguese Railway Line. The results reveal that both the wind speed and orientation have a preponderant impact on the vehicle running safety, contrary to the vehicle speed which influence is minor.

This article presents an efficient methodology for the calibration and validation of a numerical model of a freight wagon based on a dynamic test under real operation conditions. The dynamic test takes place during a regular journey of the train and involved the installation of on-board accelerometers and LVDTs, whose number and location was conditioned by the space occupied by the goods and constraints associated with loading and unloading the wagon. The data derived from the dynamic test was used for the identification of carbody's modal parameters, namely the frequencies, mode shapes and damping coefficients and to extract accelerations and displacements time-histories. A three-dimensional (3D) FE numerical model of the freight wagon was developed and calibrated using a genetic algorithm. The methodology proves efficiency and robustness in precisely estimating three numerical parameters, besides a significant upgrade in relation to the model before calibration. Model validation involved the comparison between numerically simulated results, based on a vehicle-track dynamic interaction analysis, and the experimental observations. An excellent agreement between experimental and numerical after updating time-histories was obtained, especially for the carbody responses, whose behaviour is governed by lower frequencies (below 3.5 Hz), in which the calibration process was focused.