If the base of a crack or a notch is the main reflector in a piece of steel, is it possible to accurately plot the through-wall thickness of it using the 3-point 6 dB drop method?
I practiced plotting some notches simulating lack of fusion on a weld today. They were very shallow, and my plotting simply plotted the beam spread. But I could tell they were ID connected and what their widths were. However, when I attempted to plot a .100" deep ID notch in the edge of the block, the plotting continuously came out wrong. It was showing a square notch to be C shaped. The main plotting angle (70 degrees) was at the end of the first leg (so it was obviously ID connected). Moving in and dropping the signal amplitude by half, the next plotting angle (64 degrees) registered a depth of .850", where as it should have read .900" (since it was a .100" deep notch in a 1" block). Pulling back to my third plotting angle (76 degrees), it vastly oversized the notch and put its maximum depth at .750".
Is this simply the various angles registering off the same reflection point, and not truely plotting the defect?
David, it sounds as though you are doing just that. Have you been using a beam profile and plot of the block?, if so, you should be able to simulate and identify what is happening to the sound beam and not guess.
Trying to size and depth an ID connected reflector (as you have found out) can have many problems: mainly from the acoustics of the internal reflecting surfaces. This is due to the sound reflecting back from a corner at half the internal angle ***please see diagram in attached document*** for example, you will find that a 90° notch (or corner) will reflect an acoustically enhanced (amplified) wave at 45° - a.k.a. ``The Corner Trap``. This causes the dB drop methods to oversize the reflector(s).
Probe selection is important.
Because of this `corner trap` NEVER use a 60° on an ID connected reflector. To demonstrate how misleading this is try and depth the bottom corner from the side of an I.I.W. V1. The I.I.W. V1 has a thickness of 1.0`` (or, 25mm) so simple trigonometry would put the maximized bottom reflector at a 2.0`` (50mm) sound path. In reality, the reflector maximizes on the screen appears at approximately 1.6`` (42mm) sound path. ***please see diagram in attached document***
(Though the older style Panametrics probes did not have this problem because of their unique design)
This phenomenon is less evident (though still there)with a 70°, but using the trailing edge at 64° (as you were) has picked up the corner trap.
Using a 45° will work with a corner trap, but in the real world cracks are not all at 90° (and the material thickness may not allow for a 45° probe), so there will always be a margin of error.
I would prefer to use Max. Amp. for the cross for both sectional and depth sizing.
For further interest, please note that the attached diagram also demonstrates how a 90° corner trap responds to a 30° probe. This will introduce you to another school of thought in the creeping wave debate. It is my belief that if one takes a little time to understand how a corner trap amplify reflections (simple acoustics) he (or she) will be in a much better position to understand the true properties of the so-called creeping wave.
Please find 2 attachments that I am putting together for this topic.
The Max. Amp., attachment is Not a Final draft and I encourage your feedback. Please respect all Copyrights.
I experimented with my plotting again. I used a 70 degree, .25" 5mhz probe to plot lack of penetration in a 1 inch weld. The 70 degree beam and 76 degree beam plotted correctly, but the 66 degree beam did not. Though it should have plotted in the first leg, there was too much sound path. So I disregarded that angle and only plotted the 70 and 76 degree beams. It plotted vertically, on the appropriate side of the weld. It did oversize the throughwall depth of the flaw, but not by much. This can be attributed to beam spread. Length sizing of the flaw was accurate.
I didnt have time, but I would like to try this two-point plotting method on a ID connected flaw that is larger than my beam spread, such as the 10% notch in my 1" pipe. Hypothetically it should work, as it avoids the creeping-wave error.