Clustering of interlaminar and intralaminar damages in laminated composites under indentation loading using Acoustic Emission

Abstract

This study focuses on the clustering of the indentation-induced interlaminar and intralaminar damages in carbon/epoxy laminated composites using Acoustic Emission (AE) technique. Two quasi-isotropic specimens with layups of [60/0/-60] 4S (is named dispersed specimen) and [60 4 /0 4 /-60 4 ] S (is named blocked specimen) were fabricated and subjected to a quasi-static indentation loading. The mechanical data, digital camera and ultrasonic C-scan images of the damaged specimens showed different damage evolution behaviors for the blocked and dispersed specimens. Then, the AE signals of the specimens were clustered for tracking the evolution behavior of different damage mechanisms. In order to select a reliable clustering method, the performance of six different clustering methods consisting of k-Means, Genetic k-Means, Fuzzy C-Means, Self-Organizing Map (SOM), Gaussian Mixture Model (GMM), and hierarchical model were compared. The results illustrated that hierarchical model has the best performance in clustering of AE signals. Finally, the evolution behavior of each damage mechanism was investigated by the clustered AE signals with hierarchical model. The results of this study show that using AE technique with an appropriate clustering method such as hierarchical model could be an applicable tool for structural health monitoring of composite structures.

Barely visible impact damage assessment in laminated composites using acoustic emission

“…In our previous research [32], the performance of six different clustering methods to classify the AE signals of interlaminar and intralaminar damage mechanisms in specimens S D and S B under quasistatic indentation loading is compared and finally, three damage mechanisms containing matrix cracking, fiber breakage and delamination were clustered using the hierarchical model. The clustered AE data of these damage mechanisms is shown in Fig.…”

“…This fact shows that there Table 5). As can be seen, despite the dynamic and quasi-static nature of LVI and indentation tests and also the utilized method (clustering indentation AE data by hierarchical model [32] and analyzing impact data by WPT) there is a good consistency between the percentage of different damage mechanisms for LVI and indentation tests that it shows AE is a powerful tool to…”

“…They did not directly monitor the impact process by AE and used the AE to monitor the quasi-static Compression After Impact (CAI) test on the impacted specimens. Saeedifar et al [32] studied the performance of six different clustering methods containing k-Means, Genetic k-Means, Fuzzy C-Means, Self-Organizing Map (SOM), Gaussian Mixture Model (GMM), and hierarchical model to classify AE signals of the interlaminar and intralaminar damages in carbon/epoxy laminated composites under quasi-static indentation loading. The results showed the hierarchical model has the best performance to cluster the AE signals of the damage mechanisms.…”

Using passive and active acoustic methods for impact damage assessment of composite structures

“…Whereas, the initial matrix crack AE activities at load cycles LC3 and LC4 illustrate the fact that matrix crack has an unstable evolution behavior in these load cycles. However, due to the fact that matrix cracking is not usually considered as a critical damage mode individually [18], this point is not considered as the critical damage state of the plate. At the initial stage of load cycle LC5, some delamination AE activities are found that show the unstable behavior of delamination growth for this load cycle.…”

“…These studies almost focused on two subjects; 1) Damage detection and classification, and 2) Damage localization. Saeedifar et al [17,18] employed AE to detect, classify and also trace the evolution of different damage mechanisms in CFRP plates under quasi-static indentation and LVI loading conditions. There are different approaches to localize damage in a composite plate using AE such as triangulation [19,20], delta T-mapping [21,22], multi-steps and optimizing techniques [23], etc.…”