A Rounded-Looking Toe Indication and a Ground-Out Indication
During final inspection of a pressure-retaining nozzle-to-shell weld you find a 6 mm long, 3 mm wide indication at the toe, oriented transverse to the weld. The project specification states: any indication whose length exceeds three times its width is a linear, rejectable indication; rounded indications are acceptable up to 5 mm individual length, provided no more than three occur in any 150 mm of weld. A second, 2 mm indication disappears after 0.3 mm of local grinding; the wall still has 3 mm of corrosion allowance remaining above minimum thickness. The fabricator wants the disposition, and the purchaser wants the report.
1. Evaluate the 6 mm by 3 mm indication. Is it acceptable as a rounded indication?
Read the worked answer
On the numbers alone it is a 2:1 indication, so the specification's 3:1 test classifies it as rounded. But the 5 mm individual length cap still rejects it, because 6 mm exceeds 5 mm. It is rejectable on that basis alone, before any cluster count.
Do not stop there. The 3:1 rule is applied to the size of the indication as it appears, and particle build-up widens indications: a tight transverse toe crack can bleed sideways and look rounded at 2:1. Measuring a particle indication with a steel rule also overstates width and understates the sharpness of the metal break.
Recommended evaluation: light local grinding or a PT of the area to establish the true length and width of the metal discontinuity, then re-measure. Also count rounded indications within the 150 mm window, including the 6 mm one if it is retained as rounded, since the three-in-150 mm rule can reject the area independently. The transverse orientation at a toe in a pressure-retaining weld is a fatigue-suspect location, so the default evaluation should be conservative: treat it as a linear indication and repair unless the metal break clearly disproves that.
2. What must happen with the 2 mm indication that was ground away, and what could go wrong with that disposition?
Read the worked answer
Grinding away an indication is a repair and must be controlled, not a casual cleanup. Required steps:
- Re-magnetize the ground area in two directions and confirm by MT that no indication remains; run PT as a cross-check because grinding can smear metal over a short crack and hide it from MT.
- Verify the remaining wall against the minimum plus the 3 mm corrosion allowance; document the actual thickness measurement and location.
- If the indication was a crack, the root may extend below the ground surface; a 0.3 mm grind that removes the indication may have removed only the visible part. Use the geometry (toe, weld, transverse or longitudinal) and the depth of grind to decide whether further excavation or UT is needed.
- Record the grind depth, location, method, and the post-grind MT result.
What goes wrong most often: the area is ground, the indication vanishes, and the part is accepted without a post-grind MT and PT, without thickness verification, and without any record. That converts a detected defect into a documentation gap and leaves a possible crack root in service.
3. What must the inspection report contain so that both dispositions are defensible?
Read the worked answer
Build the report so a third party can reconstruct both the detection and the decision.
Identification and method: part and weld numbers, drawing and specification revision, technique/procedure number and revision, acceptance criteria quoted verbatim, date, and the certified Level II or III who performed and evaluated.
Process parameters actually used: magnetization method and current or yoke lifting power verified at the spacing used, particle type and bath concentration, black light intensity and ambient light reading, and field directions with the number of magnetizations.
Indications: location referenced to a datum (weld toe, clock position, distance from a nozzle centreline), size in length and width, orientation, and classification as relevant, nonrelevant, or false, with the reason.
Disposition: the 6 mm indication rejected against the 5 mm rounded-indication cap with the measurement method stated; the 2 mm indication removed by 0.3 mm grinding, with post-grind MT and PT results, thickness reading, and acceptance.
Close with the disposition authority and any repair/re-inspection record. An entry that says only 'no relevant indications' is indefensible unless the coverage and parameters that support it are recorded.