Bayesian analysis of acoustic emission data for prediction of fatigue crack growth in concrete

Abstract

Acoustic emission (AE) is a valuable technique for non-destructive testing, enabling real-time monitoring and measurement of fatigue crack growth in engineering structures. The present study focuses on investigating the fatigue crack growth behaviour in plain concrete specimens through Bayesian analysis of AE data collected during the testing of beam specimens under three-point bending. A log-linear relationship is established between parameters extracted from AE waveforms and the fatigue crack growth in concrete. Additionally, AE energy is identified as the most suitable parameter for characterising the fatigue behaviour of concrete. Bayesian regression is employed for estimating model parameters and their posterior distributions. The proposed model is validated through post-processing and Bayesian analysis of experimental data from the literature. Furthermore, it is demonstrated that the estimated model parameters remain unaffected by the frequency of fatigue loading and the size of the specimen.