This experimental characterization study shows that for most configurations, the specular model for geometrical echo calculations provides results close to experimental measurements and better that those obtained with Kirchhoff model.
It improves the obtained echo predictions during the water path variation, probe aperture, probe tilt and in the case of a complex « wavelike » as-machined surface.
In addition to improvement of specimen geometrical echo amplitudes, storage problems have been solved with specular model, simulations time are shorter and computations are less sensitive to the accuracy variation than with Kirchhoff model.
The specular model has its own limitations as well. The case of the « weld bead » surface points out that diffraction phenomena not taken into account by this model imply surface echo predictions clearly different from experimental results.
Finally, in some configurations (the more divergent probe, 6.35 mm diameter), the specular model tends to overestimate amplitudes when the reference is a side drilled hole – discrepancies reaching 6 dB. At this time, this phenomenon remains unexplained.
Back to geometrical echoes