Six Rejects and One Welder
On a 16-in., 0.375-in. wall, X65 mainline, six welders using a semi-automatic process complete 84 girth welds in one day. Radiography reports: three welds with incomplete fusion at the root, two welds with porosity above the acceptance limit, and one weld with a 2-in.-long arc burn on the pipe body about 1 in. from the cap, with visible pitting. Welder R produced four of the six rejected welds and had no rejects the previous week. The contractor has a repair crew grinding the root defect now and wants all six repaired in place to hold a three-day weather window. The crew's RT film is read by the contractor's own technician.
1. Sort the six rejects into repair, cut out, and inspect further, and state the basis for each.
Read the worked answer
I do not sort on the basis of the contractor's urgency; I sort on what the discontinuity is and what the acceptance standard allows.
- Incomplete fusion at the root (3 welds). Root IF is a planar, crack-like defect and is the least forgiving to repair. If the excavation is complete, the defect is fully removed, and after grinding you are still at or above the specified minimum wall with a repairable groove, a repair is permissible under the welding procedure. If the excavation would take you below minimum wall, or the defect runs circumferentially, cut out. I require the excavation to be verified by the same NDE method before welding, not eyeballed by the grinder.
- Excessive porosity (2 welds). Typically repairable after complete removal, but excessive porosity in a semi-automatic root usually signals a shielding-gas or wind problem, not a random event. Repair, but also fix the cause before more welding happens.
- Arc burn with pitting (1 weld). This is the one that is most often mishandled. An arc burn is a rejectable discontinuity in the pipe body, and the affected metal may have a hardened, brittle layer. The burn must be ground out and the area inspected; if the remaining wall is below the specified minimum, or the ground area exceeds the allowable depth, or hardness testing is required by the specification, the piece is cut out - you cannot weld over an arc burn and call it repaired.
I do not authorize any grinding until the excavation limits and the repair procedure are confirmed in writing.
2. What do you do about Welder R, and what is the RT technician conflict?
Read the worked answer
Welder R. A 4-of-6 reject rate in a single day is not bad luck; it is a performance signal. The correct sequence is: pull R from production welding now, quarantine and identify R's welds still in the line from the surrounding period for review, and look at the variables before blaming the welder - root pass technique, heat input, joint fit-up and hi-lo, weather and wind shielding, wire or gas lot changes, and whether R was welding on a different internal lineup. Then requalify R on a test joint per the qualified procedure before R returns to production. If the project specification or API 1104 §6 sets a repair-rate trigger, that trigger has likely been crossed and requalification is not optional. Requalification is a documented test with NDE, not a verbal "he's fine now."
RT technician conflict. The contractor employing the welders cannot be the sole reader of the film on welds those welders produced. I require an independent, qualified NDE technician with current certification - the operator's or a third-party's - and I audit the technique, the film quality, and the weld identification map so that the film on the table is provably the weld in the ditch. I also confirm the acceptance criteria being applied are the project's, not a default the technician chose.
3. The contractor asks how many repairs a weld can take and whether repaired welds need re-shooting. Answer, and describe the records you require.
Read the worked answer
Repair limits. The project specification governs, and the common limit is two repair attempts on a given weld: a weld that fails after the second repair is cut out. I confirm the numeric limit in the specification and in the applicable API 1104 repair clause before authorizing the second attempt, because operators do vary it. The reason for the limit is metallurgical - each thermal cycle coarsens and embrittles the heat-affected zone, so the third attempt is not the same weld as the first repair.
Re-examination. Yes. Every repaired area is re-examined by the same NDE method at 100% coverage over the repaired region plus the excavation, and the repair weld is subject to the same visual and dimensional acceptance as a production weld. A repair that is "probably fine" is a cut-out waiting to happen under a road crossing.
Records I require before the line is accepted:
- Weld log tying each weld number to welder, date, heat number, WPS number, and joint location.
- NDE report with technique, technician certification, and the film or digital image retained.
- Repair record: which method, how many attempts, excavation depth verified, and who authorized it.
- Cut-out record with the removed piece's heat number and disposition, and the replacement joint's traceability.
- Welder requalification test records for R and for anyone else whose rate triggers review.
If any of these is missing, the segment is not ready to be coated, let alone buried.