Fatigue crack initiation detection in ductile cast iron crankshaft under rotating bending fatigue test using the acoustic emission entropy method

• Published in: Engineering failure analysis Vol. 144; p. 106981
• Main Authors: Hosseini, Seyed Morteza, Azadi, Mohammad, Ghasemi-Ghalebahman, Ahmad, Jafari, Seyed Mohammad
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.02.2023
• Subjects: Acoustic emission entropy; Ductile cast iron; Field Emission Scanning Electron Microscopy; Real-time health monitoring; Rotating bending fatigue test; Acoustic emission entropy; Real-time health monitoring; Rotating bending fatigue test; Ductile cast iron; Field Emission Scanning Electron Microscopy
• ISSN: 1350-6307; 1873-1961
• DOI: 10.1016/j.engfailanal.2022.106981

•The AE entropy method is more practical for real-time structural health monitoring applications.•In the AE entropy method, the maximum entropy was found to be constant during the degradation period.•The Shannon entropy is more useful for detecting fatigue fractures under different loading conditions.•The logarithmic entropy curves have more suitable for detecting fatigue crack initiation points.•Fatigue crack initiation life estimation based on crack density is a method for verifying the results of the AE entropy. The subject of this study was fatigue crack initiation detection in a ductile cast iron (EN-GJS-700–2) crankshaft. The specimens extracted from the crank web of a car crankshaft were exposed to the four-point rotating bending fatigue test with fully-reversed loading as per iSO-1143. The fatigue crack initiation estimation was performed by (1) using acoustic emission features, (2) utilizing the acoustic emission entropy method, and finally, (3) using crack density measured in field emission scanning electron microscopy images. In this research, the sampling rate in the acoustic emission features method was 1 million data per second, which was reduced to 10 data per second in the acoustic emission entropy method. Therefore, in the acoustic emission entropy method, the data volume was greatly reduced and it was much more practical and cost-effective for real-time health monitoring. As a remarkable issue, the average difference percentage between the Shannon entropy and the Kullback-Leibler relative entropy was 32% and 35%, respectively, compared to AE features for the fatigue crack initiation prediction.