Model-based damage reconstruction in composites from ultrasound transmission

Composites are high performance advanced materials with a growing applicability due to their extreme strength, rigidity to weight ratios. However, usual kinds of damage are hardly visible and require monitoring techniques to guarantee their reliability. Impact-type damage, which may occur during manufacture, service or maintenance, can induce severe degradation of the mechanical properties to composite materials by delamination, matrix cracking and fiber breakage while remaining invisible from the surface. This work aims to design and develop experimentally a non-destructive monitoring technique based on ultrasonic transmission through thickness combined with a model-based inverse problem for detecting variations in structural parameters due to impact damage. A model-based inverse problem is proposed using a semi-analytical model of the ultrasonic wave propagation in the layered composite material. Once calibrated, this model simulates the response of the excitation - propagation - measurement system and the wave-damage interaction for the assumed types of continuum damage. The damage parameters and the mechanical constants of the laminated composite material are obtained by minimizing a cost functional that quantifies the mismatch between experimental and simulated measurements. These parameters can be correlated with the damage condition. Genetic algorithms are used as parameter search algorithms due to their capability of finding a global solution where the cost functional has several local minima that may give false answers, at the cost of larger computational resources than gradient-based algorithms. Finally, a sensitivity study to the uncertainties of the parameters is performed for establishing the feasibility of this technique. The reconstruction procedure using genetic algorithms is proved feasible to quantify mechanical damage properties from a single measurement. (c) 2012 Elsevier Ltd. All rights reserved.