Assessment of the Effects of Scanning Variations and Eddy Current Probe Type on Crack Detection

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

“…f cr E' E'" dV V F (6) This formula is applied to liftoff by considering the "flaw" to be a thin slice of thickness h removed from the surface of the work piece, and taking the perturbed (primed) field to be the same as the unperturbed field. The same approach can be used for surface roughness by allowing h to vary with ~osition on the surface according to the roughness function.…”

“…In this procedure, errors in locating the line of the flaw profile have minimal effect on inversion accuracy, because this line corresponds to the ridge passing through the two peaks of the image plot. 6 In the following section a formal algebraic procedure is described for inverting the data available in the flaw profile. It is shown that an essential initial step in this process is an accurate determination of the surface length of the surface-breaking flaw.…”

Methodology for Analysis and Characterization of Nondestructive Inspection Capability Data

“…The binomial grouping or moving average method of analysis described by Rummel et al [1] is accomplished by passing a large number of flaws of varying sizes, bracketing the anticipated detection capability of the inspection process, through the inspection and recording the output in terms of detection or failure to detect. The probability of detection as a function of flaw size is obtained by \1) ordering the data from largest flaw size to smallest flaw size; ~2j selecting a sample size for binomial statistics that is consistent with the desired reliability and confidence level (for example a sample size of 29 provides a 90% reliablity / 95% confidence level analysis); (3, counting down sequentially starting with the largest flaw to obtain a sample of the selected size; (4) calculating the point estimate for detection probability (ie.…”