Towards Model-Based Condition Monitoring of Railway Switches and Crossings

Licentiate thesis, 2021

Railway switches and crossings (S&C, turnouts) connect different track sections and create a railway network by allowing for trains to change between tracks. This functionality comes at a cost as the load-inducing rail discontinuities in the switch and crossing panels cause much higher degradation rates for S&C compared to regular plain line track. The high degradation rates create a potential business case for condition monitoring systems that can allow for improved maintenance decisions compared to what can be achieved from periodic inspection intervals using measurement vehicles or visual inspection by engineers in track.  

To this end, this thesis addresses the development of tailored processing tools for the analysis of measured data from accelerometers mounted adjacent to the crossing transition in crossing panels. With the presented tools, a condition monitoring framework is established. The analysis procedures showed robustness in processing large datasets. The framework includes the extraction of different crossing panel condition indicators for which the interpretation is supported by multi-body simulations (MBS) of dynamic train–track interaction. Additionally, a demonstrator is presented for MBS model calibration to the measured track responses.

A particularly important signal processing tool is the development of a novel sleeper displacement reconstruction method based on frequency-domain integration. Using the reconstructed displacements, the track response is separated into quasi-static and dynamic domains based on deformation wavelength regions. This separation is shown to be a promising strategy for independent observations of the ballast condition and the crossing rail geometry condition from a single measurement source.

In addition to sleeper acceleration measurements, field measurements have been performed in which crossing rail geometries were scanned. The scanned geometries have been implemented into a MBS software with a structural representation of the crossing panel, where analyses have been performed to relate the concurrently measured accelerations and crossing rail geometries. To address the variation in operational conditions in the MBS environment, a sample of measured wheel profiles was accounted for in the analysis. This MBS study showed that there is a strong correlation between the crossing rail geometry condition, wheel–rail contact force, and crossing condition indicators computed from the dynamic track responses. Contrasting measured and simulated track responses from the six investigated crossing panels showed a good agreement. This observation supports the validity of the simulation-based condition assessment of crossing rail geometry.

Based on the work in this thesis, a foundation is set for developing methods for automatic calibration of S&C MBS models and subsequent damage evolution modelling based on operational online condition monitoring data. This development aims to address S&C service life in a digital environment and presents a key component for building a Digital Twin prototype for S&C condition monitoring.