On modelling techniques for the numerical simulation of Acoustic Emission signals

Abstract

Acoustic Emission (AE) is a highly sensitive technique for detecting the initiation and evolution of damage in materials, widely used in structural diagnosis and health monitoring. However, its sensitivity results in large datasets, making it essential to distinguish damage-related signals from environmental noise and other unrelated acoustic events. In practice, experiments conducted under fully controlled conditions, where specific damage mechanisms such as micro-crack initiation or fibre breakage can be directly correlated with recorded AE signals, are often extremely difficult or impossible. In this context, numerical simulations provide valuable support. A reliable numerical model enables the simulation of individual damage events and the prediction of the corresponding signals detected by AE sensors. Developing such models is challenging, as they must capture the entire physical chain: (i) elastic energy release due to locale sub-critical crack formation and / or propagation, (ii) wave propagation through the material, particularly including damping, (iii) transmission of surface vibration to the sensor, (iv) piezoelectric conversion of mechanical vibrations into electrical signals, and finally (v) signal processing into the amplitude-time-frequency representation, typically expressed in decibels (dB). This contribution presents recent advancements achieved by our team, in the development of numerical tools capable of addressing these challenges.