Probabilistic residual strength assessment of smart composite aircraft panels using guided waves

•This contribution presents a framework for the translation of SHM indications to reliability estimates in composite structures.•Fusion of numerical with experimental observations is proposed to reduce the required number of large-scale tests, improve prediction accuracy, and propagate uncertainties.•A method is presented for the estimation of disbond damage area in composites through Bayesian inference.•The performance of the framework is demonstrated for a 1.6 m flat stiffened panel. Typical prognosis methods in Structural Health Monitoring (SHM) link the degradation mechanisms to the extracted health indicators. Such approaches however do not provide estimations on the current strength loss when a new damage is detected. This study presents a framework that enables the SHM-informed asset integrity management of composites. First, the integrated guided wave based SHM system is used to inform on the existence, location and size of the damage and subsequently, the residual strength of the structure is predicted. The framework combines physics-based and data-driven models to mitigate the incompleteness of the former and address issues relating to the representativeness in the training datasets of the later. Detailed finite element models are used to create a digital representation of the structure while a building-block program is initiated to quantify and propagate the uncertainties observed experimentally. A 1.6 m composite panel with a skin-stringer delamination, equipped with a network of 24 piezoelectric transducers, is used as a case study to demonstrate the prognostic capabilities of the framework. The framework estimated the damage size with a mean absolute percent error (MAPE) of approximately 10% while the residual strength of a destructively tested damaged panel was predicted with a MAPE of 5%.