In order to evaluate the effects of the curvature of the bottom and the surface of the specimen on the multiple echoes of the notch, inspections of a backwall breaking vertical notch located in a flat mock-up and a cylindrical specimen of the same thickness (12 mm) were performed with T waves produced by a flat probe. In each specimen, the “multiskip” echoes reflected by the notch after "n" skips of the beam on the backwall (n between 1 and 7) were measured. The notch has 2x10mm dimensions, the probe diameter is 12.7mm and its center frequency 5MHz. For reproducibility reasons, inspections were performed in immersion. The simulations were performed with CIVA11.1.
For this study, 3 mock-ups were used: one with SDHs intended for the probe calibration and the measurement of the attenuation, a flat specimen and a cylindrical mock-up of the same thickness. Particular attention was required during the production of these three blocks in order to ensure a same attenuation in each of them (all of them are extracted from the same steel ingot and the notch and the SDH were machined so that in each specimen the fibers are oriented in the same way relatively to the direction control). Despite these precautions, the attenuation measurements have shown that attenuation was not identical in the three blocks. This has complicated the study but we succeeded in reaching conclusions regarding the prediction of multiskip echoes of the notch with CIVA.
The flat specimen was inspected with T45° mode while the cylinder inspections were performed with 4 angles of incidence included between 16° and 19°.
Regarding the evaluation on the flat mock-up, comparisons have shown a very good prediction of the amplitudes of the notch multiskip echoes, the echodynamic curves and the Ascans.
Regarding the evaluation on cylindrical part, comparisons showed differences between experimental multiskip echoes and the ones predicted by CIVA. The differences were analyzed at first for corner echoes with the help of ATHENA2D. They are due to the fact that CIVA does not take into account the head waves on the surface of the cylinder (for angle of incidence of 16°) and doesn't take into consideration creeping waves on the surface of the notch (for angle of incidence of 19° ) which both have an impact on the notch corner echo. This causes differences between CIVA and measurement for both amplitudes and echodynamic curves shapes.
The differences obtained between measurement and CIVA for the amplitudes of the multiskip echoes after the corner echo were analyzed in a second time. Their evolution when the number of skips increases greatly depends on the attenuation coefficient of the cylindrical specimen which was not determined precisely. For this reason, simulations of multiskip echoes of the notch were performed with several coefficients. The best agreement in amplitude was obtained for the coefficient of 0.025 dB/mm. With this coefficient, the differences between measurement and CIVA hardly increase with the number of skips. By extrapolation of observations made on corner echoes, these differences as well as those obtained between CIVA and experiments for the shapes of echodynamic curves were attributed to phenomena which are not taken into account by CIVA (head wave on the cylinder surface and / or creeping waves on the surface of the notch).
Thus, in the flat specimen, during the inspection with T45° waves, no head wave or creeping wave have appeared. Moreover, the CIVA predictions for multiskip echoes of the notch are in good agreement with the measurement. In the cylindrical specimen, due to the curvature of the surface and of the backwall, it seems difficult to find an angle of incidence for which no head or creeping wave is generated. Therefore, CIVA predictions deviate from the measurement. However, it was not possible to quantify with certainty these differences during successive skips due to the uncertainty of the determination of the attenuation coefficient in the cylinder.