PSI - Issue 18

Danilo D’ Angela et al. / Procedia Structural Integrity 18 (2019) 570–576 Danilo D’Angela et al. / Structural Integrity Procedia 00 (2019) 000–000

572

3

(a)

(b)

(c)

Fig. 2. Testing features: (a) testing set-up, (b) CT specimen geometry (Table 1), and (c) notch detail. The dimensions are in mm.

The mechanical testing was carried out using a servo-hydraulic digitally controlled actuator; AE testing was performed using a single-channel USB AE system (software AEwin™ ); a 300 kHz-resonant ultra-low noise pre amplified sensor ( PKI30 ) was used to detect the AE data. The AE equipment was produced by MISTRAS Limited. The superior end of the fatigue machine was fixed, and the inferior one was moved by the actuator (Fig. 2.a); the AE sensor was coupled to the fixed end of the fatigue machine by a thin layer of silicone glue. The testing plan is reported in Table 1. Four CT samples were tested (Fig 2.b and Table 1): the samples A 1 and A 2 ( S 1 and S 2 ) were made of Aluminum alloy 7075-T651 (Steel S355J0). Two mock tests were carried out prior to the definite tests using similar samples; these were aimed to (a) check the reliability of the whole testing equipment, and (b) calibrate the AE testing parameters. Accordingly, the AE amplitude threshold was set equal to 45dB. The crack initiation was detected during the testing by means of dye penetrant inspection (European Committee for Standardization, 2000); the failure was defined by the complete fracture of the sample (clearly observable).

Table 1. Testing plan. ID test Material

W [mm] H [mm] N [mm] B [mm] P [kN] R [-] f [Hz]

Al 7075-T6 71 Al 7075-T6 63

86 77 77 77

32 29 29 29

4

2.5

0.05 0.05 0.05

10 10 10

A 1 A 2 S 1 S 2

10 10 10

5 5 6

S355 S355

63 63

0.05 10

The AE signals were filtered using the modified Swansong II filtering technique (Ercolino et al., 2015) in order to remove (a) long-duration Hits with low Amplitude, and (b) short-duration Hits with high Amplitude, which are typical features of non-genuine AE signals. The filtering limit values were derived by Yu et al. (2011), who performed similar tests. The traditional AE analysis was preliminary performed considering both historical/comparison plots and correlation analysis (i.e., b-value analysis ). The information Entropy of the AE data was performed according to both Shannon (1948), and Kullback- Leibler (1951) formulations. The former Entropy is referred as Shannon Entropy H S (Equation 1), and the latter is referred as relative Entropy H R (Equation 2). Both Entropies are defined by the probability mass distribution vector p i (Equation 3), which can be evaluated considering the ratio between the number of Counts n i and the cumulative number of Counts c i . � � �∑ � � � � � � ��� (1) � � �∑ � � � � � � ��� � � ��� (2) � � � � � � � ; � � � � ; … ; � � � � � (3)