Nondestructive evaluation of FRP-reinforced structures bonded joints using acousto-ultrasonic: Towards diagnostic of damage state

Abstract In civil engineering, structural adhesive bonding is increasingly used for the reinforcement or rehabilitation of degraded structures using adhesively bonded composite materials. However, the presence of defects within the bonded joint may affect the effectiveness of the assembly, and their detection is still an issue. The purpose of our study is to evaluate the acousto-ultrasonic (AU) method to detect and identify most common types of defects encountered in adhesively bonded joint and more specifically in FRP reinforced structures. Two types of assembly (made of carbon fibre reinforced polymer pultruded plates bonded on steel or concrete support) were investigated considering three types of defect (voids, kissing-bond defects simulated using a layer of grease at the composite-adhesive interfaces, and weakly polymerized zones implemented through a softer adhesive). The objective of this work is to investigate the ability of acousto-ultrasonic method to detect all those types of defects in the bonded joint with a data-driven methodology. It is expected that the extracted features from the received signals represent the damage state in the joint. For the methodology, a data-driven modelling by means of a Principal Component analysis is performed. Additionally, a Random Forest (RF) is trained using extracted parameters.