Abstract

The estimation of track degradation rate is a challenging step in scheduling preventive maintenance interventions due to its uncertain nature. To tackle this challenge, a large dataset of historic passenger track geometry data, between 2011 and 2021, integrated with Ground Penetrating Radar (GPR), and Light Detection And Ranging (LiDAR) data are investigated. Stochastic models are proposed for estimating the expected track degradation rates based on subsurface and drainage conditions of blocks and also a framework is proposed for automatically evaluating drainage conditions of track from LiDAR data. Results indicate that the probability of having high track degradation rates in highly fouled ballast condition can be 10 times more than in clean ballast condition. In addition, analyzing the effect of drainage conditions revealed that sections with poor drainage conditions are 30% more likely to have high track degradation rates than sections with moderate drainage conditions. The results of this study enhance scheduling geometry maintenance and provide reliable information for decision making on different types of maintenance such as Ballast Cleaning and ditching.