Prediction of RCF clustered cracks dimensions using an ACFM sensor and influence of crack length and vertical angle

ABSTRACT

Rolling contact fatigue (RCF) cracks are the predominant reason for rail grinding maintenance and replacement on all types of railway system, as they can potentially cause rail break if not removed. To avoid excessive material removal, accurate crack sizing is required. Alternating current field measurement has been used as an electromagnetic method for RCF crack sizing, incorporating with modelling results for single RCF cracks with large vertical angles (>30°). No study using this knowledge to size shallow angled crack clusters has yet been reported. A novel method, the pocket length compensation method, is proposed to determine the length and depth of RCF cracks with shallow vertical angles. For shallow crack clusters, vertical angle predictions are close to the measured values with a deviation of less than 13.6%. Errors in crack pocket length prediction are greatly reduced when the pocket length compensation was included. The predicted vertical depth using the approach developed for clustered angled cracks is accurate with errors <8.3%, which compares to errors of up to 60% if the single RCF crack approach is used and errors of up to 21.4% if a non-compensated prediction for crack clusters is used.

KEYWORDS:

• Modelling
• ACFM
• RCF
• clustered cracks
• pocket length
• vertical angle

Acknowledgements

The authors wish to thank the Railway Research Centre at the University of Birmingham and the Steel Processing Group from Warwick Manufacturing Group (WMG), University of Warwick, for the provision of facilities and equipment. The authors also would like to express their gratitude to TSC Inspection Systems for constructive technical discussions and Eric Magel for the provision of rail samples. The work was supported by a joint scholarship between the Chinese Scholarship Council (CSC) and the University of Warwick, UK.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by the University of Warwick.