Improved Probe-Flaw Interaction Modeling, Inversion Processing, and Surface Roughness Clutter

In Reference 1 a first comparison was made of measured eddy current signals with calculations based on nonuniform probe-field interaction theory. These calculations followed the basic analysis developed in Reference 2. They used interrogating field distributions calculated by Dodd and Deeds theory for the air core coils of Reference 3. (Note that Fig. 6 in Reference 1 and Fig. 7 in Reference 3 should be interchanged). In Reference 1 theoretical and experimental plots of the flaw profile curve (a plot of △Z versus distance along the mouth of a surface breaking flaw) were found to be in good agreement, with regard to shape, for several selected EDM notch samples in aluminum. An iterative procedure was also developed for systematically varying the length, depth, and opening width to obtain a best fit to the experimental data.4 In the present paper a full inversion procedure is developed and illustrated for approximately rectangular-shaped EDM notches. The mathematical structure of the inversion problem is first examined and a solution is proposed. Physical reasoning, based on the form of the flaw profile curves, is then used to simplify the approach and to provide guidance in selection of the most suitable probe geometry. Other topics briefly addressed include, possible improvements in the theory for the region with a/§ close to unity and for more realistic flaw shapes (i.e., semi-elliptical, rather than rectangular), inaccuracies due to errors in the probe scan path, and background clutter due to surface roughness, machining marks, and micro-structure