API 1104, ISO 13847 or GB 50369? How Pipeline Welding Acceptance Criteria Reshape Automatic Welding Parameters

Same machine, same pipe, different repair rate. How API 1104, ISO 13847 and GB 50369 acceptance limits rewrite automatic welding parameters.

Two spreads weld the same pipe, with the same machine model, to broadly the same procedure, and finish with very different repair rates. On one project the cap toe passes AUT without comment; on the other, every eleventh joint returns for grinding at the same clock position. Nothing in the welding set has changed. What changed is the document the inspector is holding.

Acceptance criteria behave like a lens. They do not alter the arc, but they decide which consequences of the arc become visible, which become chargeable, and therefore which parameters a crew will defend and which it will quietly adjust. For an owner writing a technical specification, or a contractor buying automatic pipeline welding equipment, understanding that lens is as important as understanding the machine itself. This article compares the three acceptance frameworks most often named in international pipeline tenders – API 1104, ISO 13847, and the Chinese construction acceptance series GB 50369 / GB/T 31032 – and traces how each one pushes down into the parameter sheet and into equipment selection.

One machine, two verdicts: what an acceptance document really controls

Every acceptance framework does four jobs, and only the first one is usually discussed:

  • It defines which geometric and volumetric imperfections count as defects, and at what size.
  • It sets how much of the work is examined, and by which method – radiography, manual ultrasonic, or automated ultrasonic testing (AUT).
  • It defines the evidence package: which records must exist, in what form, and for how long.
  • It allocates signature authority, and therefore the practical power to reject a joint.

The philosophical split matters more than any single number. Workmanship-based acceptance compares the weld against fixed geometric limits that were written for manual welding: cap height, cap width, undercut depth and length, hi-lo, porosity counts. Fitness-for-service acceptance, in the form of engineering critical assessment (ECA), asks a different question: given the actual flaw sizes and the actual loading, will this joint survive the design life? API 1104 keeps workmanship as its default and offers alternative acceptance criteria for girth welds as an option; ISO 13847 is built around procedure qualification and project-agreed imperfection levels; GB 50369 and GB/T 31032 set construction acceptance with examination proportions tied to pipeline grade and class. The practical consequence is that the same weld, produced by the same machine, can be compliant on one contract and non-compliant on the next.

Two identical machines can therefore acquire different reputations without any hardware difference at all. When a supplier is blamed for a repair rate, the first question should be which acceptance basis produced that rate.

Single-torch pipeline automatic external welder on a large-diameter girth weld

Three frameworks side by side

The table below is an orientation map for commercial and engineering discussions. It is deliberately not a clause-by-clause citation: editions change, national annexes exist, and the contract always governs. Its purpose is to show where the three systems put their emphasis.

DimensionAPI 1104ISO 13847GB 50369 / GB/T 31032
Typical useNorth American origin; widely invoked by owner specifications far beyond North AmericaEuropean and international projects; replaced the older EN 14163 lineageConstruction acceptance for long-distance oil and gas pipelines in China
Acceptance logicWorkmanship limits, with optional alternative acceptance criteria for girth welds based on fracture mechanicsProcedure-qualification driven; imperfection acceptance levels agreed in the project specificationConstruction acceptance with examination proportion and acceptance level tied to pipeline grade and class
PersonnelWelder and welding operator qualification within the codeWelder per ISO 9606-1; welding operator per ISO 14732Operator qualification per domestic rules plus project specification
Examination emphasisRadiography in the classic form; AUT increasingly demanded by ownersRT, UT or AUT as agreed; AUT common where automatic girth welding is usedRT and UT, with AUT increasingly specified on automatic welding spreads
Documentation weightComparatively lean: weld records and examination reportsHeavier: qualification traceability, WPS/PQR envelope, record retentionFull construction archive integrated into project completion documents
Where it bites automatic weldingCap profile and undercut at fixed clock positionsEvidence that the procedure envelope was actually heldJoint-to-joint consistency plus completeness of the archive

Four limits that quietly rewrite the parameter sheet

Most parameter arguments on an automatic welding spread are really acceptance arguments in disguise. The following four items account for the majority of them.

Acceptance itemWhat the crew ends up changingWhere the machine carries the burden
Cap reinforcement height and widthPass count, travel speed, oscillation width and dwell, sometimes the final pass strategyDual-torch oscillation with arc tracking; single-torch position-synchronised parameters
Undercut depth and length, especially at 4 to 8 o’clockTorch angle, wire position relative to the side wall, heat input at the transition, oscillation dwell at the toeAutomatic arc tracking; angle-oscillation mode; stable travel at low speed
Hi-lo, root concavity and root pass profileFit-up force, internal support, purge discipline, root travel speedHydraulic expansion and centring of internal welders; clamping and fit-up of external stations
Porosity and start-stop defectsPre-flow, post-flow, gas coverage at the joint, inter-pass cleaningOnboard gas supply, torch geometry, sequence control

None of these items can be solved by parameter tuning alone, because each one is a joint property: geometry, procedure and machine behaviour interacting at a specific clock position. That is why acceptance-driven projects tend to buy automation that can hold a position, not just automation that can weld fast.

Machine by machine: where the compliance burden actually sits

Single-torch DW-EW-I: one cap profile around the whole circumference

The DW-EW-I is built around position automation. Welding parameters change automatically as the torch travels, so the cap at 6 o’clock is produced under the same logic as the cap at 12, on a rack-and-hole steel track with a single rack-and-pinion drive and a split ring that installs quickly. For acceptance purposes this is the key property: a workmanship criterion is a limit on the worst point of the circumference, and position synchronisation removes most of the ordinary causes of the worst point. The visual touch screen supports parameter upload and download, which is also how a qualified procedure is transferred between machines without transcription errors. The platform covers Φ219 to Φ1422 mm, so the same operating logic follows the project from small-diameter laterals to mainline.

Dual-torch pipeline automatic external welder with arc tracking and multi-channel data

Dual-torch DW-EW-II: two arcs, one acceptance

With two torches running simultaneously, the characteristic failure mode changes. It is rarely a uniformly bad weld; it is an asymmetric one – a high side and a low side, undercut on one toe and a heavy cap on the other. That asymmetry is exactly what a workmanship criterion catches, and exactly what a fracture-mechanics-based criterion tolerates least predictably, because the flaw population is not the one the ECA was calibrated for. The DW-EW-II addresses this structurally: straight and angle oscillation modes, automatic arc tracking that corrects both torches continuously, digital communication between power sources and control cabinet, and multi-channel storage, recall, import and export of process data. On a contract that demands procedure-envelope evidence, that data channel is not a convenience – it is the deliverable. Shared consumables across the two torches also matter commercially on high-volume spreads.

Battery-powered pipeline internal welding machine DW-EH48/56

Internal welders: the root is the examined surface

On most acceptance regimes the root pass is the single most consequential pass, because it is the pass that cannot be repaired from outside without cutting the joint. Both internal welders in this series work from inside the pipe, with eight welding units, and both are self-powered: the DW-EH48/56 lithium-battery unit (Φ1219 to Φ1422 mm, ≤90 s per pass, 5 to 7 MPa hydraulic expansion, 30 L onboard shielding gas, wall thickness 10 to 35 mm, ≤4000 m altitude, −40 to 65 °C, 2950 kg) and the DW48/56-8H-D flexible unit (≤120 s per pass, fit-up plus root pass in about five minutes, DC24V with three 230 Ah battery groups, 0 to 60 m/min travel, ≤30° gradient, 5/6 D bends, 2400 kg, −50 to 65 °C, ≤3500 m, wire Φ0.9 to 1.2 mm). The hydraulic expansion force is what converts fit-up quality into a controlled hi-lo, and eight-unit symmetry is what makes the internal bead uniform around the circumference. Where an ECA has been calibrated on a specific root profile, this repeatability is the difference between a valid assessment and an argument.

Flexible internal welding machine DW48/56-8H-D on a mountain pipeline

Automatic TIG: the acceptance is usually about cleanliness

Automatic TIG occupies a different niche: root passes on small diameter, on heavy wall where dilution matters, and on tie-ins where access is constrained. Its acceptance arguments tend to be about a clean, low-heat-input root and about oxidation control rather than about cap shape. This family has no independent published data sheet in the current series; its parameters must come from the approved WPS of the project, and its acceptance from the governing standard and specification.

The three-tier specification problem

Acceptance rarely arrives as one document. It usually arrives as three, and they do not always agree.

TierTypical documentsWhat it fixesTypical conflict
Design codeASME B31.4 or B31.8; GB 50251 or GB 50253Design factor, wall thickness, class location, examination classDesign class implies 100% AUT while the construction spec still assumes RT sampling
Construction acceptanceAPI 1104, ISO 13847, GB 50369 / GB/T 31032Imperfection limits, examination proportion, repair and cut-out rulesAcceptance level written for manual welding applied unchanged to automatic welding
Owner supplementProject technical specification, addenda, ECA reportAdditional limits: hardness caps, 100% AUT, tighter hi-lo, data handover formatSupplement stricter than every code, discovered only at mobilisation

Four rules keep this tiering from becoming a dispute:

  1. The strictest applicable requirement governs, unless a written waiver exists. Silence in one document never overrides an explicit limit in another.
  2. Clarify in writing before mobilisation, not after the first repair. A one-page technical clarification letter is cheaper than a re-work campaign.
  3. Record the interpretation. When two clauses can be read two ways, state which reading the project will apply, and file the answer.
  4. Never average. Averaging two acceptance limits produces a third limit that no document authorises and no inspector will sign.

A worked comparison: same pipe, two regimes

The scenario below is illustrative, written to show how the same equipment package is judged differently. It is not a real project and not a recommendation.

AspectRegime A: workmanship plus 100% radiographyRegime B: AUT plus engineering critical assessment
Acceptance basisFixed geometric limits, every joint examinedFlaw-size based assessment with agreed detection thresholds
Root pass priorityRoot profile within workmanship limits at all positionsRoot profile matching the flaw population assumed by the assessment
Fill and cap priorityCap height, width and toe blend within fixed limitsCap geometry adequate for AUT scanning and coating; less sensitive to cosmetic variation
Repair policyRepair anything outside the fixed limitsRepair only flaws that exceed assessed acceptance sizes; more joints pass, more scrutiny on detection reliability
Equipment emphasisPosition-synchronised parameters and arc trackingProcedure-envelope evidence, data export, and repeatability proof
Documentation emphasisExamination reports per jointData packages that can defend the detection threshold in an audit

What the dossier must contain, whichever code governs

  • WPS and PQR, with the qualified envelope stated in the same units the machine actually displays.
  • Machine configuration record per spread: model, serial number, torch configuration, firmware or programme version.
  • Per-joint parameter data, exported rather than photographed, linked to the joint number.
  • Examination reports and, where applicable, AUT data files in the agreed format.
  • Repair and cut-out log with the acceptance basis cited for each decision.
  • Consumable records: heat or lot numbers, wire and gas, mapped to joint ranges.
  • Personnel qualification records mapped to the joints each operator produced.
  • A reconciliation step that proves the four data sets above describe the same joints.

Trial welds under a new acceptance regime: six steps

  1. Read the acceptance basis first, before setting a single parameter. Identify whether the governing logic is workmanship, procedure envelope, or flaw assessment.
  2. Translate the limits into machine settings and name the pass and clock position each limit constrains.
  3. Weld a trial batch that deliberately brackets the envelope – not just the nominal point – so the margin is visible.
  4. Examine the trial with the method that will be used in production, not a substitute. An RT-qualified trial proves little about an AUT regime.
  5. Record the resulting data package and have the inspector accept the format before production starts.
  6. Freeze the configuration and document the change control that applies if any parameter must move afterwards.

Ten-point clarification list before you mobilise

  • Which edition and which addenda of the acceptance standard apply, and from which date?
  • Is examination 100% or sampled, and by which method, on root, fill and cap?
  • Is there an ECA, and does it constrain root profile, hardness, or flaw orientation?
  • Are hardness limits specified, and at which points are they measured?
  • What hi-lo limit applies, and is it measured before or after internal expansion?
  • Which supplementary requirements exist beyond the code – for example data handover, AUT qualification, or operator certification?
  • Who signs acceptance of the trial welds, and what evidence do they want to see?
  • What is the repair versus cut-out rule, and who decides?
  • In what format must per-joint data be delivered, and for how long retained?
  • If two documents conflict, which one has precedence, in writing?

Frequently asked questions

Does a stricter standard require a more expensive machine?

Not necessarily. It requires a machine whose behaviour is documented well enough to prove compliance. Position-synchronised parameters, arc tracking and exportable process data do more for a strict acceptance regime than raw travel speed. What a strict regime does change is the tolerance for undocumented manual intervention.

Can the same procedure be used under API 1104 and ISO 13847?

Usually not without requalification. The two systems define the qualified envelope, the essential variables and the acceptance levels differently, and a procedure qualified under one is normally treated as a new procedure under the other. Budget the requalification as a project cost rather than discovering it at the first joint.

Why does our repair rate change when the inspector changes?

Because a different acceptance basis makes a different population of imperfections chargeable. Before adjusting parameters, confirm that both inspectors are applying the same edition, the same examination proportion and the same measurement method. In practice a large share of apparent quality swings are measurement and interpretation differences, not welding differences.

Where do internal welders fit under a fitness-for-service regime?

They fit well, provided the assessment was calibrated on a root profile the machine actually produces. The eight-unit symmetry and hydraulic expansion of the DW-EH48/56 and DW48/56-8H-D give a repeatable internal geometry, which is exactly what an ECA needs. If the machine or its settings change, the assessment basis should be reviewed with it.

Conclusion

Acceptance criteria are not paperwork that arrives after welding; they are the specification of what good welding means on a given contract, and they reach all the way down into pass strategy, torch behaviour and data output. Teams that read the acceptance basis before setting parameters spend their effort on compliance; teams that read it afterwards spend the same effort on repair. The three frameworks compared here do not rank as better or worse – they rank as different lenses, and the practical skill is knowing which lens is in front of the inspector.

Beijing Anjie Weida Technology Co., Ltd. (DENOH Group) supplies the full pipeline automatic welding series – single-torch and dual-torch external welders, lithium-battery and flexible internal welders, and automatic TIG solutions – with engineering support for procedure transfer, data export and acceptance documentation. Tel: 010-86468776. Email: sales@denohgroup.com. WhatsApp/WeChat: 13521628344.