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

571

2

(a)

(b)

Fig. 1. AE testing: (a) technique application scheme (MISTRAS Limited), and (b) main AE features (Ercolino et al., 2015).

AE testing has been widely applied in recent years, and several techniques have been developed for the analysis of the AE data. The data analysis techniques range from basic historical / comparison plots (Aggelis et al., 2011; Gostautas et al., 2005) and correlation analysis (Al-jumaili, 2016; Carpinteri et al., 2009), to multidimensional analysis and clustering (Al-jumaili, 2016; Ercolino et al., 2015). AE testing has been shown to be reliable for structural damage assessment in the case of laboratory testing and time-continuous monitoring (Al-jumaili, 2016; Carpinteri et al., 2009). However, when the testing conditions are not well-controlled (e.g., monitoring of in-service bridges) or in the case of time-discontinuous data detection, AE testing is not necessarily reliable and valid (Chai et al., 2018; Chen et al., 2017; Nair and Cai, 2010). Even if novel processing techniques have been recently developed for noise disturbance reduction, their features depend on the specific application, and field monitoring is often needed (Schultz, 2015). Furthermore, the intrinsic chaotic nature of the AE waves makes even more difficult the analysis and the interpretation of AE data (Kahirdeh et al., 2017a). Recent studies directly addressed the chaotic nature of the AE phenomena in order to improve the damage assessment by means of AE testing. This was motivated by the known phenomenological correlation between the structural damage and the systemic disorder , which can be quantified by the evaluation of the Entropy of the system (Amiri and Khonsari, 2011; Moreno-Gomez et al., 2018). Entropy is an extensive property of a thermodynamic system, and its concept is used in statistical mechanics and information theory to assess the evolution of systems having a large number of degrees of freedom. One of the first formulations of the information Entropy is due to Shannon (1948); Shannon Entropy defines the measure of uncertainty contained within a random variable, or equivalently, the amount of information that a variable contains. Shannon Entropy and thermodynamic Entropy are not theoretically correlated; however, their applied formulations are equivalent. Shannon Entropy of the AE waves (i.e., AE Entropy or acoustic Entropy ) was found to be promising for structural damage assessment, but it is still far from the application to (a) real-time structural health monitoring, and (b) time-discontinuous data detection/analysis (Kahirdeh et al., 2017a, 2017b; Stavrakas et al., 2016). The paper presents AE testing of fatigue fracture in metal plates performed according to the parameter-based approach (Grosse and Ohtsu, 2008). The detected AE data were post-processed and filtered using the latest literature techniques (Ercolino et al., 2015; Yu et al., 2011). The information Entropy of the AE data ( AE Entropy ) was evaluated according to both Shannon (1948) and Kullback-Leibler (1951) formulations. The experimental crack initiation, crack propagation, and fracture failure were correlated to the AE Entropy evolution. Damage criteria based on AE Entropy were also identified for time-discontinuous data detection, simulating realistic structural health monitoring of fracture critical components of bridges. 2. Experimental testing and AE analysis Acoustic Emission (AE) testing of fatigue crack initiation and propagation was performed on Compact Tension (CT) specimens (D’Angela and Ercolino, 2018) made of structural metals. The testing set-up is shown in Fig. 2.a, and the geometry of the testes samples is shown in Fig. 2.b/c.