Planar UT indication in a thick nozzle-to-shell weld
A 60 mm thick low-alloy steel nozzle-to-shell weld (set-on, full penetration) is examined by manual pulse-echo UT after PWHT. The scan uses 45° and 60° shear-wave probes, 2 MHz, and a DAC curve from a 3 mm side-drilled hole. An indication is found 20 mm below the outer surface, at the fusion line on the nozzle side, with a maximum amplitude of 80% DAC, length 18 mm, and it is planar (no significant width). Radiography is not possible because of the geometry. The acceptance criteria are ISO 5817 level B, with additional requirement that all planar indications exceeding 10 mm are unacceptable. The welder claims the indication is 'slag' and asks for a retest with a different probe angle. The client wants to avoid a repair because the vessel is already hydrotested and painted.
1. Evaluate whether the indication is likely to be slag, lack of fusion or a crack, and whether it is acceptable.
Read the worked answer
The indication is planar, located at the fusion line, 18 mm long, and gives a strong echo at 80% DAC. Lack of fusion is more likely than slag. Slag is usually volumetric and rounded, with a broader, less directional echo; a planar indication at the fusion line is more consistent with lack of fusion or a crack. A crack must also be considered, particularly if the weld had high restraint or hydrogen exposure, although the post-weld heat treatment may have changed the stress state. The exact type cannot be settled from amplitude alone, but the acceptance decision does not depend on calling it slag. The criteria say that planar indications exceeding 10 mm are unacceptable, and this indication is 18 mm long and planar. Therefore it is unacceptable regardless of whether it is lack of fusion, slag stringer or crack. A retest at a different angle may improve characterisation, but it cannot make the indication acceptable unless the original indication is proved false.
2. What additional NDT or characterisation would you perform, and what are the limitations of radiography here?
Read the worked answer
I would perform phased array ultrasonic testing or time-of-flight diffraction to size the indication through-wall and to confirm its planar nature. Use multiple angles and a 0° probe to check for delamination or laminar tears. Map the indication and compare it with the geometry. Radiography is limited here because the nozzle geometry prevents good access, and radiography is insensitive to planar lack of fusion oriented parallel to the beam. A planar defect can be missed or underestimated by radiography even when access is good, so a clear radiograph would not overrule a confirmed planar UT indication. Surface methods such as magnetic particle or penetrant testing are useful only for surface-breaking defects and would not characterise this embedded indication. If the indication is to be removed, gouge or grind and verify removal with magnetic particle testing and UT before re-welding.
3. Decide whether to repair or accept, considering the hydrotest and painting status and the contractual position.
Read the worked answer
The indication should be repaired. The acceptance criteria are explicit: a planar indication longer than 10 mm is unacceptable. A hydrotest does not prove the absence of planar defects; it may not open a tight lack of fusion, and it does not address fatigue or fracture behaviour. The fact that the vessel is painted is not a technical reason to accept. Contractually, the client should be notified of the nonconformance, and the repair should be made under an approved repair procedure. For a 60 mm low-alloy steel weld, the repair may require preheat and post-weld heat treatment, and the vessel may need re-hydrotesting after repair. The inspector should not certify the weld as acceptable without repair, even under delivery pressure. If the client insists on acceptance, the refusal and the technical basis should be documented in writing.