Disbond detection and load estimation on a composite test box using Fiber Bragg grating sensors

Abstract

Detection of hidden damages such as disbonds and delaminations are of relevance to the safety and maintenance of composite airframe structures. This study explores damage identification and load estimation using artificial neural networks (ANNs) on a composite test box. The objective is to detect disbond between spar and skin of the composite box using strain measurements made using fiber optic sensors. A novel approach is devised to create disbonds of various lengths using a non-adhesive insert between top skin and flange of centre spar. Initially a large disbond is created and progressively the disbond size is reduced in stages to obtain a healthy structure. The composite box is loaded at every stage and strains are acquired at various locations in the box using fiber Bragg grating (FBG) sensors. The difference in strains between healthy and unhealthy stages of the box is indicative of the presence of damage, as well as its extent and the loads acting on the structure. ANNs are trained using data generated from numerical models of the composite test box for healthy and disbond cases. These trained ANNs are then used to estimate the applied load, disbond size and disbond location using experimental strain data.