On the identification of the failure mechanisms in oxide/oxide composites using acoustic emission

In the present work, a new class of oxide/oxide composites made of Nextel™ 720 fibre reinforcement and a mullite-based matrix, fabricated by using liquid polymer infiltration process, were studied. A fibre coating was applied via sol-gel in order to achieve improved damage tolerant behaviour. Mechanical properties were investigated at ambient temperature under quasi-static loading in the presence of continuous Acoustic Emission (AE) monitoring. Statistical pattern recognition analysis is the proposed tool for the classification of the monitored AE events. Lacking an a priori knowledge of different signal classes, unsupervised pattern recognition algorithms were used. A complete methodology including descriptor selection methods, procedures for numerical verification and cluster validity criteria is followed. Cluster analysis of AE data was achieved and the resulted clusters were correlated to the damage mechanisms of the material under investigation. This process was assisted by systematic microscopic examination. Furthermore, the initiation and evolution of each mechanism is described by plotting the cumulative hits of each class as a function of the applied load.