Subscribe to our newsletter to receive the latest news and events from TWI:

Subscribe >

Acoustic Emission Monitoring of Train Axle Fatigue Cracking

Summary

Train axles failures have caused major loss of life, with a consequential increase in both non-destructive testing (NDT) and manual inspection that is expensive and disruptive to efficient rolling stock maintenance. TWI plays a key role in an InnovateUK project, aiming to use complimentary methods of high frequency vibration and Acoustic Emission (AE) monitoring of axles.

Project Background and Objectives

MONAXLE “Live monitoring of train axles with autonomous wireless systems” aims to research the feasibility of detecting axle cracks on trains, continuously monitoring axles and replacing expensive and disruptive inspections between major overhauls. TWI’s main objective is to investigate axle crack growth during fatigue loading by creating cracks at locations most likely to naturally occur, thus simulating the behaviour of the asset.

Figure 1. Technical drawing of the tested axles
Figure 1. Technical drawing of the tested axles
Figure 2a. Test set-up
Figure 2a. Test set-up
Figure 2b. AE sensors mounted on the axle
Figure 2b. AE sensors mounted on the axle

Work Programme

Axles European grade 15 230.9 (Figure 1) were designed, manufactured and tested. An elliptical notch was introduced, of 3-4mm deep and 8-12mm long, simulating the shape of a fatigue crack. The equipment used for this process is electrical discharge machining.

Axles were instrumented with strain gauges and AE sensors and tested in air, at ambient temperature, in a servo-hydraulic test machine under constant amplitude loading using three-point bend testing (Figure 2).

Achievements and Future Work

Different signal processing techniques have been assessed, taking into account noise filtering requirements and the capability of the algorithm to differentiate between damage-related signals and signals generated by other sources (Figure 3).  

Both acoustic emission transient (waveform) and wave (wavestream) analysis have been considered.

Signal processing methods investigated include established clustering methods (Figure 4) such as k-Means algorithm and Kohonen self-organising map (SOM). This technique has shown good potential for the sorting of damage AE signals from noise. Similar tests will be performed on axles grade A4T, in accordance with British railway standards. Efforts will be focused on further refining the signal processing technique applied to the upcoming tests.

 

Figure 3. AE raw data, (a) Amplitude vs time of hits
Figure 3. AE raw data, (a) Amplitude vs time of hits
Figure 3. AE raw data, (b) Cumulative energy vs time with measured crack length
Figure 3. AE raw data, (b) Cumulative energy vs time with measured crack length
Figure 4. Clustering of data (a) U-matrix
Figure 4. Clustering of data (a) U-matrix
Figure 4. Clustering of data (b) U-matrix clustered by K-means algorithm
Figure 4. Clustering of data (b) U-matrix clustered by K-means algorithm
Avatar Emilie Buennagel Principal Project Leader - Fatigue Integrity

Emilie joined TWI in August 2010 after spending more than three years in the engineering and technology division of a large materials testing company in the UK. Previously she completed her MSc studies at the University of Orleans in France, including two work placements at Nexans in Germany.

Emilie brought to TWI experience of mechanicals testing and metallurgy, primarily for the aerospace industry. In her previous employment she was responsible for managing projects and developing new test programmes. At TWI she manages a range of projects, including failure investigations, fatigue assessments and resonance fatigue testing of full-scale pipe.

Among the research projects Emilie has been involved with has been a core research programme investigation comparing approaches to design fatigue assessment, with reference to a pressure vessel designed to an old standard. A published paper based on this research is available on the TWI website.

Media /

Related Insights

View all insights
}