Sour Service Girth Weld: ECA Versus Cut-Out
A 323.9 mm OD × 9.5 mm wall API 5L X52 gas line is designated sour by the owner (H₂S partial pressure 4.5 kPa), so ISO 15156 applies. Automated PAUT on one girth weld reports a 20 mm long, 3 mm through-wall indication at the fusion line at 6 o'clock, characterised as lack of fusion - 32% of the wall. Workmanship acceptance rejects it. The contractor proposes an engineering critical assessment under API 1104 Appendix A instead of cutting out: the cut-out, re-weld and re-inspection costs three days, the schedule float is two days, and the welder's previous twelve welds were all acceptable. The PAUT procedure has never been validated for sizing accuracy at this wall thickness.
1. What must exist before an ECA is even a legitimate option on this weld? Distinguish what is a precondition from what is merely convenient.
Read the worked answer
Preconditions (an ECA is invalid without all of them):
- Contractual authority. Appendix A is not self-executing: an alternative acceptance criterion must be permitted by the contract and agreed in writing by the company/owner before it is applied. Retro-fitting an ECA after a rejection is an acceptance-criteria decision made by the wrong party.
- Valid flaw sizing. The ECA consumes a height, and a 3 mm height is being quoted by a procedure that has never been validated for sizing. That is the fatal gap: unvalidated sizing makes the 32%-of-wall figure unusable, and the whole assessment rests on it. A qualified sizing procedure with a demonstrated error band on this diameter, wall and weld prep is required first.
- Planar-flaw treatment. Lack of fusion is crack-like and must be assessed as a crack, with the correct orientation and the correct driving stress (pressure plus axial/bending plus residual). Treating it as a volumetric inclusion would be unconservative.
- Real material properties. Toughness of weld metal, fusion line and HAZ at the operating temperature, from the procedure qualification or the actual weld; SMYS and the actual yield; and the applied stress state - not handbook defaults.
- Flaw characterisation confirmed, ideally by a second, independent technique; a single automated PAUT call at 6 o'clock is a starting point, not a proven morphology.
Merely convenient (and irrelevant to the decision): two days of schedule float, three days of cut-out cost, and the welder's twelve good welds. Cost and schedule can inform a decision; they cannot satisfy a precondition.
2. How does the sour-service designation change the assessment, and could an ECA on a purely fracture-mechanics basis accept a flaw that is nonetheless unacceptable here?
Read the worked answer
Yes - a mechanically acceptable flaw can still be unacceptable in sour service, because the ECA and the sour-service standard answer different questions.
- ISO 15156 is a metallurgical and hardness-based standard. For carbon steel it limits hardness in the weld metal and HAZ (typically 250 HV10 maximum) and requires the welding procedure to be qualified for the severity of the environment. An ECA says nothing about whether the joint is metallurgically fit for H₂S.
- Hydrogen changes the failure mode. In sour service, a planar flaw at the fusion line is a hydrogen-assisted crack initiation site. The assessment must account for sulfide stress cracking and hydrogen-assisted growth, not just the dry fracture-toughness value. At 4.5 kPa H₂S the environment is not benign.
- Residual stress plus a sharp fusion-line defect plus hydrogen is the classic SSC configuration. The defect is at 6 o'clock, the highest restraint region of a girth weld, and its size (20 mm long) concentrates hydrogen charging.
- Hardness data must be real. Any hardness survey has to be made with a properly calibrated instrument on a prepared and replicated surface, with conversion uncertainty declared; a single Leeb reading on a ground cap is not a basis for declaring ≤250 HV10.
Therefore: the assessment must be run with the sour-service environmental crack growth allowance, the hardness/metallurgical acceptance must be met independently, and the writer of the ECA must be competent in both - or the flaw must simply be removed. An ECA that ignores the environment is worse than no ECA, because it creates a documented rationale for leaving a defect in a sour line.
3. Give your recommendation to the project, with the reasoning that a Level III would put in writing.
Read the worked answer
Recommendation: cut out and re-weld. The reasoning is not that an ECA is always wrong - it is that this ECA cannot be completed defensibly inside the available time, and the preconditions are not negotiable:
- The sizing basis is unvalidated, so the input height is unknown within a band that could easily be ±1 mm on a 3 mm indication - a 30% error on the governing number.
- Authoring and agreeing a fitness-for-service assessment, obtaining representative sour-service toughness and hardness data, and getting owner and AI acceptance cannot be done in two days.
- Even a successful mechanical assessment would not discharge the ISO 15156 hardness and procedure-qualification requirements, which are separate obligations.
What goes in the letter to the project: the flaw, the two options, the schedule impact of each, the preconditions that the ECA option cannot meet in the available window, and the explicit statement that the schedule float is not a technical input. Record the decision and the rationale - if the joint later fails, the file must show that the choice was made on engineering grounds.
Then treat it as a population signal: review the welder's and the crew's other girth welds with the same automated system, check the internal line-up and the root gap, and look for a systematic cause of lack-of-fusion at 6 o'clock rather than paying for one re-weld.