Ph.D. Thesis Defense by Zhiqiang Shi
Tuesday, April 3, 2001

(Dr. Jacek Jarzynski, advisor)

"Studies on Quantitative Acoustic Emissions and Applications to Material Fatigue Testing"


Acoustic emission (AE), or the spontaneous release of transient elastic energy in solids due to the sudden change of localized stresses or incremental crack growth, provides information of the internal integrity of in-service structures. While applications of the technique have increased tremendously in recent years, quantitative characterization of AE signals, and more importantly, the correlation of the AE observations with the underlying mechanisms of AE generation, remains a problem.

The objective of this research is to develop a quantitative waveform-based AE technique for studying mode-I fatigue of PH13-8 stainless steel. There are two major challenges in successful application of the technique, i.e., the characterization of wave propagation and the correlation of detected AE signals with the underlying fatigue mechanisms. This work studies the elastic wave propagation in 2D and 3D plate structures, as well as the source effect on the produced AE waveform using finite element method (FEM) and laser-ultrasonics. Comparison of the results from the numerical simulation and the experiments confirms that the FEM is capable of simulating and revealing the features of AE pulse propagation in plate structures.

A series of high-cycle mode-I fatigue tests were conducted using on-line AE monitoring. The observed acoustic activities were found to be reproducible, from which three stages of fatigue were identified acoustically. The AE activities that correspond to the crack growth only further revealed two stages of fatigue crack growth. The AE activities and the crack growth behavior in these two stages were found to follow different relations. A transition between these two stages was identified acoustically as the transition from stable to unstable crack growth. Based on the differences between the AE signals due to crack growth and the noise-related events, a feature-based statistical signal processing technique was developed for automatic separation and clustering of AE signals from the noise-mixed record.