Model-Based Remote Health Monitoring of Ballast Conditions in Railway Crossing Panels

Paper in proceeding, 2023

Railway crossing panels accumulate damage due to impact loads induced in the wheel transition area. To reduce the need for labour-intensive visual inspections of crossing panels, railway administrations are evaluating solutions for remote health monitoring. One such solution is to measure the track response that follow from the wheel–crossing impact using embedded accelerometers mounted on the sleeper at the crossing transition. The challenge that remains is to determine the health of the asset from these measured signals. In this paper, a procedure is developed to identify the ballast condition under a crossing panel via the calibration of a multibody simulation model to measurement data. This model considers the complex wheel–rail interaction in the crossing transition area, while also capturing the dynamic response of the track using a Finite Element representation of the track structure. The calibration procedure has been developed using data from six in situ crossing panels where the crossing geometry and the track response are known via laser-scanned crossing geometries and measured accelerations. Parameters associated with the physical state of the ballast are identified by minimizing the least-squares discrepancy between the measured track response and the corresponding response from simulations of dynamic vehicle–track interaction. The utilised ballast parameterization has been motivated from sensitivity analysis to ensure that each parameter has a clear and observable influence on the track response. The performance of the developed calibration procedure is demonstrated and its suitability for implementation in condition monitoring solutions is discussed.