Open Access Open Access  Restricted Access Subscription or Fee Access

Structural Health Monitoring on Cracked Railway Axle Journals Using Ultrasonic Phased Array Technique

J. BAIK, S. HURLEBAUS

Abstract


The fatigue-prone nature of railway vehicle axles is a concern due to increasing vehicle loads and speeds. The failure of one axle has the potential to derail an entire train. Derailments cause a great danger to the public, can be life threatening, and may lead to thousands of dollars in repair costs and rehabilitation. The most idealistic way to prevent such a disaster is to inspect railway cars at every station. Unfortunately, this is unrealistic due to the cost and time associated with inspection. The middle region of the axle is the most fatigue critical region, as it experiences the largest stress cycles. Previous studies have shown that the axle body can be inspected using a non-contact ultrasonic technique on moving trains. Another fatigue critical region is the axle journal. Accessibility to the axle journal is limited since the journal is covered by the bearing, while surrounding areas are covered by the bearing cap and wheel. The main challenge in this research is to overcome the limited accessibility using ultrasonic techniques. Structural health monitoring on the railway axle journal using ultrasonic phased array (UPA) is introduced in this paper. UPA provides a high probability of detection (POD) and allows for high-speed scanning from a single inspection point. The technology also has a greater flexibility that allows the inspection of complex geometries. UPA transducers are attached on each railroad car axle as part of a continuous monitoring system. A full-scale railway axle specimen with multiple artificial cracks is used in the experiment. An image processing technique and a pattern recognition algorithm are applied to the results obtained from the transducers. The grey-level segmentation technique is used to clarify the resulting C-Scan images, and a supervised learning algorithm, Support Vector Method (SVM), is used to classify the axle condition. The incorporation of post-processing techniques minimizes human involvement. This research will contribute toward the development of a reliable and accurate inspection method to improve railway safety.

Full Text:

PDF