Training dataset for planar delamination identification and quantification

The project aims to develop HPC‑scalable models for structural health prediction, focusing on wave propagation through elastic and piezoelectric media. The research involves coupling elastodynamic and electrostatic problems, accompanied by a systematic review of computational models for piezoelectric materials.High‑performance computing (HPC) is employed to perform simulations of ultrasonic guided wave propagation on the selected demo case. These simulations are implemented using the Finite Element Method (FEM) within the open‑source code dolfin‑hpc.The training dataset combines different damage sizes and locations within the web area of the project’s demo case — manufactured from 3R resin — corresponding to a section of a wing spar. Eight MFC transducers (M‑0714‑P2) are positioned around the region of interest where the damage (planar delamination) is present. A round‑robin operating mode is used to interrogate the structure with five Hann‑windowed sinusoidal pulses at 120 kHz.