Defect Characterization Using Guided Wave Technology, 18-R8436
Jay L. Fisher
Adam C. Cobb
Inclusive Dates: 01/01/14 – 10/01/15
Background — Guided wave inspection technology is most often applied as an inspection survey tool, in which relatively low-frequency ultrasonic waves, compared to those used in conventional ultrasonic nondestructive evaluation (NDE) methods, propagate along the structure. Discontinuities cause a reflection of the sound back to the sensor for flaw detection. Although the technology can be used to accurately locate a flaw over long distances, the flaw sizing performance, especially flaw depth estimation, is much poorer than other, localized NDE approaches. Estimating flaw depth, as opposed to other parameters, is of particular interest for failure analysis of many structures. At present, most guided-wave techniques to estimate flaw depth require many assumptions to be made, such as weld geometry and flaw shape, and are highly dependent on the flaw reflection amplitude, which can vary based on many factors other than flaw depth.
Approach — To estimate flaw depth in a way that is not strictly amplitude-dependent, an approach based on a multimodal analysis was developed. The response of the fundamental shear horizontal wave mode (SH0), which is uniform through the part thickness, is compared with other higher order modes, primarily SH1, which are not uniform through the thickness. A specially designed meander wound electromagnetic acoustic transducer (EMAT) probe was used to generate SH0, SH1, and SH2 modes. Additionally, a model was developed to predict beam energy profiles (Figures 1 and 2) and flaw amplitude response based on part thickness, sensor size, probe distance to the flaw, and wave mode. This model was used to develop a relationship between the SH1 to SH0 mode amplitude ratio and the flaw depth, for a range of flaw widths.
Accomplishments — The approach was verified with a test set of 96 defects on plate specimens with flaws of different widths, depths ranging from 5 percent to 100 percent of total wall thickness, and different sensor-to-flaw spacings. Of these 96 flaws, 69 had SH1/SH0 amplitude ratios in a portion of the amplitude ratio curve that was not single-valued, so that they could be classified only as having depths greater than 40 percent of total wall thickness. Of these 69 flaws, five had depths that were in the range of 30 to 40 percent of wall thickness. Therefore for these 69 defects, 93 percent were classified correctly. Of the remaining 27 flaws, all had SH1/SH0 amplitude ratios in a single-valued portion of the curve, which meant that their depths could be estimated. Overall, the average depth sizing error of these flaws was only 4.3 percent, with a tendency to slightly oversize the flaw depth. The standard deviation of the error was 7.6 percent, and the maximum error was approximately 19 percent. In conclusion, this project showed that it was possible to characterize flaw depth independent of factors that cause the absolute amplitude of an individual wave mode to vary, such as non-uniform coupling of the probe to the inspection part surface, as well as (to a large extent) flaw width, and without the need for geometric reflectors for calibration. Figure 3 summarizes these sizing results.