Integrated Condition Monitoring of Railway Crossing Noses Using Modelling and Vehicle-Based Inspection Data

Railway switches and crossings (S&C) concentrate geometric discontinuities and stiffness transitions that amplify wheel–rail dynamics, accelerating ballast degradation and promoting the formation of voided (hanging) sleepers. This thesis develops a methodology including (i) a time-domain finite-element (FE) train–track model, (ii) physics-informed force proxies derived from Track Recording Vehicle (TRV) geometry (“Top-to-Force”, T2F), and (iii) a vehicle-borne image-processing (VBIP) pipeline for crossing-nose monitoring. The aim is to diagnose poor support, link it to impact forces, and track surface evolution using commonly collected track data.
The FE model represents a Timoshenko rail with discrete supports, linearised Hertzian contact, and a moving unsprung wheel. Validation against analytical benchmarks and frequency-domain responses reproduces receptance, pinned–pinned effects, and re-contact signatures at voids. Parametric studies show the first impact (P1) scales with dip angle and speed, while the second (P2) is governed by support stiffness and unsprung mass; FE trends are consistent with classical estimates within expected bounds.
To connect the numerical modelling with inspection data, this thesis develops two practical tools. First, the T2F uses TRV longitudinal level (“Top”) data to estimate quasi-static and dynamic loading indicators at crossings, enabling rapid screening of locations with high wheel–rail forces. Second, a VBIP workflow is developed to monitor changes at the crossing nose. This workflow combines consistent region-of-interest selection, illumination correction, and absolute image differencing to produce both visual change maps and a numerical change score.
A case study with a reference crossing integrates these threads. TRV dip-angle/force proxies and FE simulations with multi-sleeper voids explain persistent high P1/P2 loads and re-contact behaviour at the crossing nose despite surface resets. The combined evidence demonstrates a coherent pathway from measurement to mechanism for earlier detection and objective post-work verification in S&C.