Ballasted Track Safety Monitoring under Operational Conditions by Bayesian Model Updating of the Coupled Train-track System

Project: Research

View graph of relations

Description

Ballasted track system is the most popular and cost-effective railway track system. Extensiveresearch has been carried out on non-destructive evaluation of the rails, and commercializedtechniques are available (e.g., the track and overhead line geometry vehicle and ultrasonic testingvehicle) for inspecting railway level and alignment, rail gauges and corrugation. However, researchwork for assessing the condition of railway ballast is very limited. Ballast damage under sleepers,which can affect track stability and deteriorate riding quality, are impossible to detect through visualinspection. When railway ballast is damaged, its stiffness in supporting the sleeper is reduced. Thisalters the vibration characteristics of the track system. Therefore, it is possible to detect ballastdamage under sleepers by measuring the vibration of the track under the action of a dynamic forceand calculating the ballast stiffness by solving the inverse problem. Several difficulties must beaddressed before this method can be applied. First, it is well-known that the track stiffness dependson the magnitude of the loading on the track. Therefore, the identified track stiffness may not bereliable for track safety monitoring if the magnitude of dynamic force is very different from the trainload. To ensure the usefulness of the identified track stiffness, more complicated train tests (vibrationdue to train passage) are recommended in the proposed project. Second, the properties of railwayballast are uncertain in nature. The probability theory must be adopted to explicitly handle theuncertainty problem. Third, the modeling of the coupled train-track system is complicated.Numerical iterative methods are available for vibration analysis of the coupled system, but theanalysis result is sensitive to many factors (e.g., the suspension systems of vehicles, the imperfectionof the wheels (contact surface) and the rails (vertical profile)). Great care must be taken in theselection of an appropriate train-track model. The proposed project will address these difficulties anddevelop a ballasted track safety monitoring method utilizing an operational train load. The idea is tocontinuously measure the vertical vibrations of the train and track systems, and identify the dynamictrack stiffness by the newly developed Bayesian model updating method.?

Detail(s)

Project number9042336
Grant typeGRF
StatusFinished
Effective start/end date1/10/168/09/20

    Research areas

  • Track stiffness , Train-track vibration , Bayesian model updating , Bayesian model class selection , Markov chain Monte Carlo