The process of the so-called creeping wave is explained on Bill Blanshan's website
I do know how the term arose as it is also given the term "near side creeping wave" and "far side creeping wave". On the near surface you can mesure the time from the exit point to a returned signal from a notch and when displaying the timebase in units of time the velocity is that of a compression mode. The signal that people claim is the "far side creeping wave" can be similarly calculated measuring the actual distance from the probe to the notch and you see it is a shear mode velocity. If there was a "creeping wave" it would be a unique wave-mode and it would have a definite velocity for a given material.
The Corner "trap" is again an unfortunate terminology. You can find "anechoic traps" (see http://www.absoluteastronomy.com/topics/Anechoic_chamber) that are intended to prevent sound returning to the receiver, but this does not seem to be the meaning some people are using in NDT.
Instead, a better wording would be a "corner effect". When a ray strikes a boundary with right angles it is redireted back along a parrallel path as indicated in the uploaded image (regardless of the angle of incidence). Of course there are mode conversion losses when the boundary is struck at subcritical angles but the same mode is preserved on the second skip. When the centre of the beam is directed at the corner (junction point of the perpendicular boundaries) the separation between the incidence ray and reflected ray is indistiguishable one from the other and a maximum return signal pressure results.
just some supplementary remarks to the corner effect:
For standard carbon steel we have a ratio of about 1.8:1 for compression wave velocity to shear wave veloctity. Therefore the 'critical angles' for corner effect are about 33 degrees and 57 degrees.
If we have a shear wave sound beam, where the lower effective boundary angle is larger than 33 degrees and the upper effective boundary angle is below 57 degrees, then for corner effect in half skip insonification we have no losses, neither at the surface reflection opposite to the probe, nor at the reflection on the reflector perpendicular to the surface opposite to the probe.
If part of the beam angle is about 33 degree, then e.g. for a narrow beam we have a signal amplitude loss from 100 percent to about 12 percent, due to the mode conversion losses to compression wave at the surface opposite to the probe. We might have additional losses due to the generation of compression wave at the probe surface. If the effective angle is going down to zero degrees (perpendicular to the probe surface), we will have again 100 percent amplitude (no losses).
If part of the beam angle is about 57 degree, then e.g. for a narrow beam we again have a signal amplitude loss from 100 percent to about 12 percent, due to the mode conversion losses to compression wave at the reflector surface. If the effective angle is going up to 90 degrees (parallel to the probe surface), we will have again 100 percent amplitude (no losses).
You can see a nice diagramme on this in the famous KRAUTKRAMER (in original KRAUTKRAEMER) book on UT basics.
In practice for corner effect this means:
If you use a shear wave probe of e.g. an effective beam spread of plus/minus 5 degrees, you never should use a probe of nominal angle lower than 38 degrees or larger than 52 degrees. Usually therefore we use standard shear wave probes of 45 degrees.
If we look at the amplitude dynamics at corner effect, we see the following:
Of course for a flaw perpendicular to the surface opposite to the probe the corner effect amplitude signal is zero, if the flaw depth is zero. If the flaw depth is increasing, the signal will increase, with a maximum signal at that probe position, where the probe is insonificating with its nominal beam angle the corner point. The signal source in that situation is the effective probe output point (transmit/receive), about in the middle of the probe aperture, and the parallel transmit/receive sound pathes are identic (just one line).
But there is (in theory, but also in practice) a saturation of the signal, which is given by the aperture of the probe. Assume, that we transmit (or receive) at an probe aperture point, which is at the far (to the reflector) border of the aperture, and that we receive (or transmit) the parallel signal path at an probe aperture point, which is at the near (to the reflector) border of the aperture, when we scan perpendicular to the reflector. If you will make a sketch on this, regarding the parallel lines, you will see, that the maximum flaw depth, where we will have signal saturation, is given by the projection of the probe aperture under the nominal probe angle. Flaw depth, which is larger than the projection of the probe aperture, can't be measured by corner effect signal. The maximum signal therefore of course is identically also to the through wall flaw (specimen border).
Just another remark to 'creeping waves', which are 90 degree compression waves parallel to the probe surface or to the opposite surface: As far as I can remember, this expression (in German: Kriechwelle) was used by WUESTENBERG from BAM, Berlin, Germany, and later by KROENING from KWU Erlangen /IZFP Saarbruecken, Germany, in the 1970'ties resp. 1980'ties.