Contact Shear-Wave Screening of a Nozzle Weld with a Curved Surface
A Level III is asked to qualify a contact shear-wave technique for a 25 mm thick carbon steel nozzle-to-shell weld. The nozzle OD is 168 mm and the accessible surface is the curved shell side, not the nozzle bore. The draft procedure specifies a 5 MHz, 10 x 10 mm, 45-degree wedge transducer, a 2 mm root gap, and a scanning index set to give a single half-skip through the weld. The technician reports that root signals from the inner surface appear at 18 mm and a persistent geometric signal at 41 mm that cannot be resolved from a possible lack of fusion. Couplant is water-based gel. Surface is painted with 200 micrometre epoxy. The Level III must decide whether the technique is fit for purpose.
1. Is the specified transducer and index combination adequate for full volumetric coverage of the weld? Justify with a coverage argument.
Read the worked answer
Coverage must be checked both in the through-wall direction and across the weld width.
- Through-wall: with a 45-degree shear wave in steel, the half-skip path corresponds to one leg of 25/cos45 = 35.4 mm. A single half-skip from a curved entry surface means the beam is refracted and the actual leg length varies with curvature and with the local surface normal. The reported root signal at 18 mm is consistent with a shorter-than-nominal leg, indicating the index or the surface geometry is not what the procedure assumed.
- Across-weld: a single fixed index gives one beam path. For a weld with a 2 mm root gap and typical cap width, one index cannot cover both fusion faces unless the beam width plus index sweep spans the weld. With a 10 x 10 mm crystal at 5 MHz, the beam is narrow enough that a single index is not credible coverage.
Conclusion: the combination is not adequate. The index must be swept across the weld (at least two indices per fusion face) and the through-wall coverage verified by a plotted beam path or by a reference reflector at the far fusion face.
2. The 41 mm signal is ambiguous. What steps should the Level III require before the technique is accepted?
Read the worked answer
The ambiguity must be resolved by characterization, not by assumption.
- Map the signal: vary the index and the probe angle to see whether the signal moves with the weld geometry (geometric) or stays fixed to a weld feature (indication).
- Compare against a known geometric reflector: a reference block with the same thickness and a machined notch or side-drilled hole at the corresponding depth confirms whether 41 mm is a legitimate geometric response.
- Check the paint: 200 micrometre epoxy is a couplant and surface condition variable. If the signal amplitude changes with paint removal or couplant change, the surface is contributing.
- Use a second angle (e.g., 60 or 70 degrees) to interrogate the same volume. A true lack of fusion will respond at the expected depth for the new angle; a geometric signal will not.
Only after the signal is classified should the technique be accepted or the procedure revised.
3. What procedure changes would you mandate, and what acceptance criterion applies?
Read the worked answer
Mandated changes:
- Swept index with documented positions covering both fusion faces.
- Beam-path or coverage plot for the curved entry surface, accounting for refraction at the curved interface.
- Surface preparation requirement: remove or document the paint thickness, since it affects coupling and the effective entry point.
- Second-angle confirmation for any signal that cannot be classified.
Acceptance: the applicable code governs. For a typical fabrication code, lack of fusion is unacceptable regardless of length in many classes, while porosity and slag are assessed against length and amplitude thresholds. The Level III should state the code clause explicitly and require that any recordable indication be evaluated against that clause, not against an internal rule of thumb.