There is one setting in guided wave testing that quietly decides whether you can find a defect sitting next to a weld.
It is the number of cycles in the transmitted pulse.
More cycles gives a bigger signal. That is tempting, especially on an attenuating line.
But the number of cycles also sets the width of your signal in space, and signal width is spatial resolution.
Here is the arithmetic.
At 64 kHz the wavelength of the torsional mode in steel is about 2 inches. With a 1-cycle pulse and the 2-inch MsS strip design,
the signal width is about 4 inches. Compare that with 5 MHz conventional UT, where the signal width is around 0.05 inch.
The guided wave signal is roughly 78 times wider.
Now put a defect 2 inches from a weld. Because the wave travels out and back,
those two reflections appear 4 inches apart in the recorded data.
So a 1-cycle signal, 4 inches wide, just barely separates them. Borderline, not comfortable.
Go to 6 cycles and the signal becomes several times wider, and wider than a foot. Thus, the weld reflection signal simply swallows the defect signal beside it. Nothing in the data looks wrong. The signal is not weak, and it is not below threshold. Just the defect signal is not visible at all. Only the Weld signal is there.
And consider where corrosion actually occurs. At supports. At clamps. At welds. Where the geometric features are is exactly where the pipe gets wet, holds debris, and corrodes. So the defects most likely to be present are the ones a wide pulse cannot separate from the feature beside them.
The recommendation from this module:
Use 1 cycle by default. Two cycles is a fair trade when attenuation is high
and you need the amplitude. More than two, and you are trading away the resolution you need most.
This is also why a guided wave report has to carry per-indication and per-section detail plots. A signal-to-noise ratio tells you an indication stands above the noise. It tells you nothing about whether a second indication is hiding inside the first one. Only a plot at working scale shows that, and only a reader who can see it can check it.
The module also covers something worth knowing before you choose an operating mode. Check the dispersion curve.
There are thousands of wave modes in a 4.5-inch pipe between 0 and 100 kHz operating frequency range.
T(0,1) is the only one that is non-dispersive across that whole range.
Every other mode you generate travels at a different velocity and arrives as background noise.
That is the whole reason for insisting on mode purity. It is not elegance. It is that the alternative is noise you created yourself.