Design methodology for track systems considering the long-term ballast behaviour: application to crossings

Switches and Crossings (S&Cs), also called turnouts, refer to the components of the railway that provide flexibility to the system in terms of possible routes of a rail vehicle. Nevertheless, they are the asset which experiences the highest number of failures due to a number of reasons, including discontinuities in the rail geometry and track properties. Impact forces along the S&C panels cause faster degradation than in plain line track and, hence, the turnouts incur very high maintenance and renewal costs, in addition to significant investment costs. From a modelling perspective, S&C are also a considerably complex part of the railway system due to physical non-linearities, including rapidly changing rail geometry and trackbed characteristics, sharp radius with little or no transitions in the diverging route and rail profiles subject to heavy wear.

This research proposes a scientific based design methodology that is able to assess different assets and support the infrastructure manager’s decision, based on knowledge of the long-term trackbed evolution, which is considered one of the most significant causes of whole track degradation. Special attention is given to the crossing panel, but the methodology can be applied to the switch panel as well as plain line and other discrete features.

To support the methodology, a three-dimensional vehicle-track interaction model has been developed using Matlab. This model includes a comprehensive Finite Element (FE) model of a ballasted track section, various vehicle configurations and a contact model solved in the normal direction using the non-linear Hertzian theory with a single point of contact and in the tangential direction using the linear Kalker theory modified according to the Shen & Hendrick correction. The modelled track response is validated against experimental data from a UK site equipped with Under Sleeper Pads (USPs) achieving a very good agreement in terms of bearer vertical displacement. The long-term ballast behaviour in plain line is firstly assessed using a vertical two-dimensional track model under a point load with various settlement models available in literature for different load characteristics (i.e. amplitude and loading frequency) and trackbed stiffness values. A new settlement equation is then proposed in order to overcome the limitations of the current approaches. The results from the simplified model are applied at the crossing panel. The iterative process is validated against the calculated rate of longitudinal level defect growth using two consecutive Track
Recording Coach (TRC) data signals.

The results from the simplified plain line track model show that the settlement rate, according to the equations available in literature, not only varies in magnitude but also in trend. For instance, the Guérin’s and the Fröhling’s predictions decrease non-linearly with increasing trackbed stiffness (i.e. stiffer supports will settle less than softer ones), while the Sato’s prediction shows the opposite tendency (i.e. stiffer supports will settle more than softer ones). This can be explained with the opposite behaviours revealed by the physical quantities involved, i.e. displacement for the Guérin’s equation, the ballast force for the Sato’s equation and variation in load and stiffness for the Fröhling’s equation. It is felt that none of these results are entirely satisfactory and there is a need for a more comprehensive equation that can capture the trends seen on track as well as physical quantities that can be easily recorded in situ. A new settlement equation is therefore proposed based on the maximum energy transmitted to the ballast layer, which has been demonstrated to be the only physical quantity that reflects all the main track modes as well as the vehicle characteristics. Through this research, it has been proven that it is necessary to include the longitudinal variation of trackbed stiffness in the numerical model in order to correctly capture the long-term ballast behaviour. The tool developed can also be (but not limited to) used to assess (1) the modification of the resilient track layers, such as rail-pads and potential USPs; (2) the change of the crossing geometry design on the long-term system degradation; (3) the variation of the crossing layout due to a change in bearer spacing; (4) the modification of sleeper design and sleeper material design.

As further work, additional validation of the VTI model can be achieved using available on-track acceleration data not only at the bearer level but also at the rail level. Further verification of the proposed settlement equation should also be carried out against laboratory tests in order to capture the overall behaviour under controlled conditions and for low loading frequencies. Finally, measurement sites should be also considered in order to assess the influence of higher loading frequency.