Ultrasonic Flaw Classification Using a Quasi-Pulse-Echo Technique

Abstract

In solving ultrasonic flaw characterization problems, flaw type information is often needed in order to pursue succeeding tasks such as flaw sizing. In a typical inspection, the interaction of the incident ultrasonic pulse with the flaw results in a series of signal trains. A variety of signal features are extracted from these flaw signals and then used as the basis for the classification process. This classification process is made difficult by the large number of possible scattered waves. For example, typical ultrasonic signals from a planar crack-like defect consist of reflected responses, surface traveling waves, edge diffracted waves and head wave components. For a volumetric void-like defect, the returned signal pattern similarly contains reflected waves of the same mode as well as mode-converted reflections and “creeping” waves. However, in pulse-echo testing a fundamental difference exists between a crack-like flaw and a volumetric flaw that can be used for classification purposes. This difference is reflected in the fact that a significant mode-converted diffracted wave component can exist for a crack-like defect (Fig. 1(a)) which does not exist in pulse-echo testing for a volumetric defect (Fig.1(b)).