Monitoring of fatigue of materials from acoustic emission

Fatigue is one of the most dangerous mechanical processes leading to the failure of parts or structures, encountered in a variety of modern industrial situations ranging from nuclear reactors to microelectronic connections. However, we do not yet have non-destructive methods or indicators capable of predicting fatigue life, generally still estimated today from empirical relationships such as the Manson-Coffin law, which relates the number N _R of cycles to fatigue failure to the imposed cyclic plastic strain amplitude Δε _p : $N_{\text{R}}=c\times \text{Δ}{\epsilon }_{\text{p}}^{-\text{β}}$ . The problem with this type of approach is that the parameters c and β depend on the material, as well as the loading mode (triaxiality, for example), and must therefore be determined by numerous calibration tests. This article presents, in various metallic materials, the detection of acoustic emission signals specific to fatigue crack propagation. These signals, known as acoustic multiplets, are characterized by almost identical waveforms, the signature of a single source, and are repeatedly triggered over many successive loading cycles at the same stress level. They mark the slow, incremental propagation of a fatigue crack at each cycle, or friction along fracture surfaces. Being specific to incremental fatigue cracking, they can be used as early warnings of crack propagation, which will ultimately lead to global failure. Their detection and characterization thus pave the way for new, reliable monitoring of the onset of fatigue cracks during mechanical testing or within structures in service.