Numerical verification of an in-situ methodology for rail fasteners mechanical characterization

Abstract

Nowadays the industry spends massive amounts of money on the supervision and replacement of rail fasteners. This happens due to the fact that damaged fasteners generate vibrations and noise, which is uncomfortable and disturbing to the inhabitants of the area, in addition to the safety risks that can be associated with it. These costs could be reduced by incorporating the in-situ characterization method. This method is based on the excitation and obtention of measurements at specific locations on the track to be able to recover the stiffness of each of the fasteners. The objective of this thesis is to verify the experimental feasibility of the in-situ characterization method for obtaining the fasteners' stiffness in railway tracks using numerical models. For this purpose, some of the measurements provided by the simulation will be discarded (thus emulating the lack of sensors, since theoretically there should be as many as the number of fasteners on the track) and it will be attempted to obtain the stiffness of the fasteners. The thesis will consider two simulations: a simplified model, including 1D elements, and a more realistic model, including 3D solids. Afterwards, a MATLAB code will process the results and show the stiffnesses obtained for each fastener. The results indicate that for both the simplified model and the 3D model, the in-situ method can recover each stiffness when using all the measurements. For both cases, they also show that the closest fasteners to the one that has been discarded obtain a less accurate stiffness, while the more fasteners are between the studied and discarded fastener, the more accurate the stiffness is. In addition, the closest fasteners to the discarded one obtain a higher stiffness at low frequencies, considering the loss of stiffness. Another important fact is that the retrieved fasteners’ stiffness at high frequencies usually tends to the right value. Finally, when disregarding measurements, the stiffness values for the 3D model are more accurate than the values obtained in the simplified model. The conclusion is that the thesis has partially fulfilled its objective. Studies beyond the scope of the thesis are needed to find the required number of sensors to recover the stiffness of the fasteners with acceptable accuracy, but the results of the thesis conclude that the method is feasible and presents a good starting point for future lines of work.