A data-driven approach to quantify track buckling strength through the development and application of a Track Strength Index (TSI)

Abstract

Railroads are increasingly adopting advanced technologies to enhance safety and state of good repair of their track infrastructure, some of which employ artificial intelligence-based inspection methods. The objectivity of these inspection techniques, along with their detailed measuring capabilities, has created opportunities for railroads to improve both inspection and operational efficiency. This is achieved through more frequent and precise track data collections and technical data aggregations. This paper explores the potential of these track inspection methods to address one of the few drawbacks of continuous welded rail (CWR): track buckling. While track buckling mechanics and prevention have been the subject of numerous studies, the need for a practical and objective method to assess resistance to buckling remains. Numerous track health indices for geometry parameters and some for railway components have been developed and utilized in various applications. Yet, the opportunity exists to develop a metric that combines geometric parameters with the condition levels of railway components, particularly designed to quantify the resistance of track to buckling. Such a holistic view of the track and network-wide time series analyses of the proposed metric demonstrate whether the buckling resistance improves, remains in a steady state, or declines. In this study, a sensitivity analysis conducted using CWR-Risk software identified misalignment amplitude, track curvature, lateral resistance, torsional resistance, and longitudinal resistance as the main factors contributing to buckling resistance. Based on the sensitivity study and with a focus on the capacity side of the track buckling equation, a methodology was proposed for converting inspection data into 10-point scales that were combined through weighted averaging into a single Track Strength Index (TSI) to quantitatively assess the resistance to buckling at a track-system level. The influence of ballast condition on TSI output was evaluated using 15 ballast configuration scenarios, ensuring that the index accurately reflects ballast deficiencies in a proportionate manner. Lastly, the proposed metric was tested leveraging revenue service data collected from a Class I railroad mainline in the United States.