PSI - Issue 75

22

Fabrice Deleau et al. / Procedia Structural Integrity 75 (2025) 392–418 Deleau Fabrice, Emmanuel Persent, Guillaume Coudouel, Guillaume Perrin/ Structural Integrity Procedia (2025)

413

5.2. Acoustic Emission detection The analysis of the Acoustic Emission files focuses on the activity recorded at the level of the connector's two lugs, respectively named lug No 1 “front tooth” located on the acquisition station side and lug No 2 “rear tooth” located on the hydraulic power unit side. A localization filter is applied to display only the bursts that first reach either of the two sensors but are sufficiently energetic to be recorded by the sensors on the outside of the small-sized i-Clip body. This setting ensures that the maximum amount of noise from the fatigue machine and the i-Clip connection areas is filtered out. The EA activity recorded from the start of the test can be seen on Fig. 19, showing the burst energies recorded over the two lugs from 10840 to 50890 cycles. This activity can be observed over approximately 1,650,000 cycles. EA activity over this entire period is equivalent in terms of EA burst characteristics, or even lower in energy than during the first cycles. This activity is characterized by relatively low energies, below 10 3 eu (among other parameters).

Fig. 19: Activity recorded on small-sized i-Clip 987 - 10,840 to 50,590 cycles - Energy as a function of time

All EA files were analysed according to the cracking event detection criteria defined by the Institut de Soudure, during this test and previous experience. The bursts filtered as potentially cracking-like are highlighted in yellow on the following graphs (Fig. 20). A significant change in burst characteristics, consistent with a cracking-type phenomenon, can be observed from 1,651,000 cycles. The bursts recorded on lug No 1 show a significant increase in recorded energy. This activity is characterized by higher energy than the surrounding noise (x10). Based on previous analyses and Institut de Soudure know-how, this activity is interpreted by the EA acquisition as the onset of damage on lug No 1. It should be noted, however, that the number of bursts identified as cracking remains low at this stage of the test.

1 651 000 cycles

Fig. 20: Activity recorded on small-sized i-Clip 987 - 1 610 694 to 1 652 311 cycles - Left: Energy versus time with cracking filter - Right: Energy versus time It should be noted that this number of cycles is entirely consistent with the onset of drift observed on certain front lug deformation gauges installed in parallel (Fig. 18). Equivalent EA activity starts again at around 1,656,000 cycles on lug No 2, similarly within a relatively limited volume of activity and energies (Fig. 21).