Nonlinear Modeling of a Rail-sleeper-ballast System and Its Application in the Detection of Hidden Ballast Damage without Baseline Measurement

A mass transit system is an essential component of a region’s economic growth, and ballasted tracks are the most popular mass transit system worldwide. Extensive research has been carried out on nondestructive evaluation and ballasted track safety. The current methods for monitoring railway tracks are comprehensive. Fast and efficient methods are available for inspecting railway level and alignment, rail gauges and corrugation. However, the detection of railway ballast damage continues to rely significantly on visual inspection and destructive core tests. Clearly, visual inspection is good at detecting surface damage. Hidden defects (e.g., damaged ballast under a sleeper loading area and/or ballast shoulder), which can affect track stability and deteriorate riding quality, are extremely difficult if not impossible to detect through visual inspection. When railway ballast is damaged, its size and stiffness in supporting the sleeper are reduced. This alters the vibration characteristics of the rail-sleeper-ballast system. Therefore, it is possible to detect ballast damage under a sleeper by measuring the vibration of the system and solving the inverse problem for identifying the ballast size distribution under the sleeper. Several difficulties must be addressed before this solution can be applied. First, because the stress-strain relationship of railway ballast is nonlinear, model updating methods that rely on measured natural frequencies and mode shapes are not appropriate, and time-domain nonlinear model updating methods must be developed. Second, the set of measurements for a baseline reference system is not available. Third, the result of the inverse problem is highly uncertain owing to modeling errors and measurement noise problems. The proposed project will address these difficulties and develop a practical ballast damage detection method based on impact hammer tests following the Bayesian probabilistic approach. This method will aim to provide valuable information on ballast size distribution under a sleeper to engineers and inspectors during their visual inspections.