A Dynamic Wheel/Rail Interaction Model Based on an Euler-Bernoulli Beam Rail Model

Abstract

Early detection of rail defects is necessary for preventing derailments and costly damage to the train and railway infrastructure. A rail surface flaw can quickly go from a small fracture to a broken rail after only a few train cars have passed over it. Rail defect detection is typically performed by using an instrumented car or a separate railway monitoring vehicle. Such systems can either be costly or time consuming to implement. This paper presents a dynamic wheel/rail interaction model for generating training data for a defect detection algorithm. The defect detection algorithm works with an onboard data acquisition system and sensor set consisting of accelerometers mounted to the bogie side frame. Side frame vertical acceleration signatures are analyzed by the algorithm to detect defects ranging from small surface fractures to rail breaks with complete rail separation. Proper training of the defect detection algorithm requires a large data set. This data set should consist of a wide range of train operating conditions and defect characteristics. Operating conditions of interest include various forward speeds and payloads, and defect characteristics of interest include rail fracture and break geometric parameters. To generate this training data, a dynamic wheel/rail interaction model was developed that relates crack/break geometry to the side frame vertical acceleration signature. The model was generated by treating the rail as an Euler-Bernoulli beam with appropriate boundary conditions and the pad-tie-ballast systems as discrete, evenly spaced lumped parameter forcing inputs into the rail system. The pad and ballast were each separately modeled as a non-linear spring and damper in parallel, and the sleeper was treated as a mass fixed between the spring and damper of the pad and ballast. The model was validated by comparing the simulated vertical acceleration response to actual bogie side frame vertical acceleration data.