PSI - Issue 52

Ilias N. Giannakeas et al. / Procedia Structural Integrity 52 (2024) 655–666 Ilias N. Giannakeas/ Structural Integrity Procedia 00 (2022) 000 – 000

660

6

illustrated in Fig. 3B. In the same figure, the locations of the impacts have been translated to the corresponding positions in the 4-sensor subnetwork.

Fig. 3: A) Geometry of the flat panel and locations of the impact events and B) Subnetwork of 4 sensors and equivalent impact locations.

Such approaches are particularly helpful to improve the efficiency of the numerical investigations as they reduce the problem domain and avoid modelling the full flat panel geometry. An Abaqus model was used here to simulate the interaction of propagating waves with delamination damage. During simulations, to avoid the generation of fictitious reflections from the domain boundaries an Absorbing Layer of Increased Damping (ALID) is introduced that attenuates the outward traveling waves. Delamination was modelled by locally removing the tie constrains between the skin and the stringer. Further details on the definition of the numerical model can be found in (Giannakeas et al. 2023). The HI computed from the 4 impact events and the results from the numerical simulations were used to construct the and matrices and train the stochastic model in Eq. 6. 2.4. Uncertainty Propagation It is important to understand the uncertainties associated with the operation of a SHM system to understand its response and behaviour. Uncertainties may arise during the operation of the SHM system due to the EOC, such as changes in temperature and due to the natural variability of the physical system (e.g. material properties, damage geometry, interaction of GW with delamination). These uncertainties can affect both the detection ( ̃ ) as well as the localization results ( ̃ , ̃ ) . The uncertainty in these parameters is propagated to the ̃ estimation. It is of interest therefore to assess the sensitivity of ̃ to these uncertainties.