Probabilistic Bayesian approach for delamination localization in GFRP composites using nonlinear guided waves

Abstract Nondestructive evaluation (NDE) techniques that use nonlinear wave–damage interactions have gained significant attention recently due to their improved sensitivity in detecting incipient damage. This study presents the use of finite element (FE) simulation with the experimental investigation to quantify the effects of guided waves’ propagation through multiple delaminations in unidirectional glass fiber-reinforced polymer (GFRP) composites. Further, it utilizes the outcomes of nonlinear interactions between guided waves and delaminations to locate the latter. This is achieved through probabilistic Bayesian updating with a structural reliability approach. Guided waves interacting with delaminations induce nonlinear acoustic signatures that can be quantified by the nonlinearity index (NLI). The study found that the NLI changes with the interrogation frequency, as confirmed by numerical and experimental observations. By using the numerical outcomes obtained from the nonlinear responses, a Bayesian model-based approach with subset simulation is proposed and subsequently used to locate multiple delaminations. The results indicate that both the log-likelihood and log-evidence are key factors in determining the localization phenomenon. The proposed method successfully localizes multiple delaminations and evaluates their number, interlaminar position, width, and type.