API 1104 Pipeline Welding Acceptance Criteria — RT, AUT, Visual Guide

Long-distance pipelines move energy under thousands of pressure cycles per year, and the girth welds that join their joints are the population most likely to fail. API 1104 — the American Petroleum Institute’s “Welding of Pipelines and Related Facilities” — is the standard that defines how those welds must be made, qualified and inspected. Compared with ASME Section IX, API 1104 is more prescriptive on acceptance criteria and uniquely supports an engineering critical assessment (ECA) route that lets operators replace the workmanship limits with fracture-mechanics-based ones. This guide walks through the inspection methods that API 1104 recognises, the numerical acceptance limits that drive most pass/fail calls in the field, and the practical evidence package an inspector should demand before signing a girth-weld report.

Scope: what API 1104 covers and what it does not

API 1104 applies to onshore and offshore arc-welded pipelines made of carbon steel and low-alloy steel, including girth welds, branch connections, fillet welds and repair welds, when transporting liquid petroleum, gas, slurries, water in oil-and-gas service, and CO₂. It does not cover pressure vessels (ASME Section VIII), in-plant piping under ASME B31.3, or composite/non-metallic pipe. Pipelines of cryogenic service, hydrogen-service applications above the standard’s qualification scope, and sour-service pipelines requiring NACE MR0175 carry additional rules layered on top of API 1104.

Two acceptance regimes co-exist in the standard:

• Workmanship criteria (Section 9) — empirical limits on imperfection length, width and total cumulative length per weld. Used by default and on most cross-country jobs.
• Alternative criteria (Annex A) — fitness-for-service limits derived from CTOD/J fracture mechanics. Permitted only when the project specification, the AUT procedure and the welding procedure together satisfy the prerequisites listed in Annex A.

A girth weld inspector must know which regime applies before opening a film envelope or AUT scan, because the numerical limits are different and the rejection rate can vary by an order of magnitude.

Inspection methods recognised by API 1104

API 1104 recognises visual examination, radiographic testing (RT), magnetic particle (MT), liquid penetrant (PT), ultrasonic testing (UT) and, since the 21st edition, phased-array and time-of-flight diffraction automated ultrasonic testing (AUT). Each method has a defined acceptance scope and qualification path.

Visual examination

Visual is the only method required on every weld. The standard sets workmanship limits on undercut depth, weld reinforcement, root profile and cracks. Visible cracks of any length are rejectable; undercut adjacent to the cap or root must not exceed 0,8 mm or 12,5 % of pipe wall thickness, whichever is less. Misalignment between abutting pipe ends is limited to 1,6 mm internally and externally for common pipe schedules.

Radiographic testing (RT)

RT is the historical workhorse for pipeline girth welds and is still the default acceptance medium on most onshore projects. API 1104 specifies film classes (compatible with ISO 17636 Class A and Class B), sensitivity by wire IQI or hole-type penetrameter, and image quality requirements. Acceptance limits are length- and width-driven: for example, an isolated slag inclusion is rejectable when it exceeds 12 mm in length, or when total length of all slag inclusions in any 300 mm weld length exceeds 50 mm. Porosity is acceptable up to a defined cluster diameter and pore size relative to wall thickness.

Automated ultrasonic testing (AUT)

AUT — usually phased array combined with TOFD — is the dominant method on modern offshore lay-barge and high-volume onshore spreads, and it is the only practical choice when running ECA per Annex A. AUT advantages over RT include height sizing of planar flaws (incomplete fusion at the bevel, sidewall lack of fusion), real-time results that keep up with mainline production rates of 80–120 welds per day, and elimination of radiation safety zones. Disadvantages: AUT requires a calibrated procedure validated on representative reference reflectors of known geometry and depth, and the operator’s training package is far more demanding than a film interpreter’s.

Magnetic particle and penetrant testing

MT and PT are surface methods used as supplements, typically for branch connections, repairs, and the visible portion of root welds before cap deposition. They are not substitutes for volumetric inspection.

Workmanship acceptance criteria — the table that drives the field

Section 9 of API 1104 contains the canonical acceptance table. The numerical limits below summarise the most frequent rejection drivers on girth welds; refer to the standard for the complete list, including limits for hollow bead, internal undercut and burn-through.

These limits assume RT or visual; AUT limits are expressed in terms of indication height and length, with thresholds that depend on the AUT procedure’s validated probability of detection (POD).

Annex A and the engineering critical assessment route

Annex A allows a project to replace workmanship limits with limits derived from fracture mechanics and a representative AUT inspection. The prerequisites are non-trivial:

1. Welding procedure CTOD performance. The procedure must be qualified to a minimum CTOD value (typically 0,15–0,25 mm at minimum design temperature), demonstrated on welded samples representing the production heat input, position and consumable.
2. AUT procedure qualification. The AUT procedure must be validated against reference reflectors representative of the flaws Annex A allows, with a quantified POD and sizing accuracy traceable to API 1104 Annex A and ISO 13588.
3. Stress and toughness inputs. The applied stress envelope (axial, bending, residual) and minimum design temperature drive the allowable flaw height table. ECA uses BS 7910 or API 579-1/ASME FFS-1 procedures.

Annex A is most cost-effective on projects where the workmanship reject rate is high (typically thick-wall, sour or strain-based design pipelines) and AUT is already mobilised. It is rarely worthwhile on small-diameter onshore distribution work.

Welder and procedure qualification

API 1104 separates welding procedure specification (WPS) qualification from welder qualification. A WPS is qualified by welding test coupons in the qualification range and subjecting them to mechanical tests (tensile, bend, nick-break, Charpy V-notch when required) and notch-toughness testing (CTOD) when ECA applies. Welder qualification is performed by destructive bend tests on coupons made with the qualified WPS, in the position and direction of progression representative of production work.

Modern operators tie WPS, PQR, welder qualification certificates, NDT reports and consumable lot traceability into a single digital record per girth weld. Therness’s welding QMS software and digital welding quality records workflows show how an audit-ready package looks for an API 1104 spread.

Repairs and rework

Repairs are tightly controlled. The standard requires a qualified repair WPS, often distinct from the production WPS because preheat and run-out length may need adjustment. Repair-of-a-repair on the same area is allowed once unless specifically authorised by the company. Cracks must be excavated to sound metal and verified by MT or PT before reweld. Cap-only repairs may be carried out without removing the entire fill, provided the indication geometry meets the conditions in Section 10.

What an audit-ready acceptance package looks like

Beyond the welds themselves, an API 1104 audit asks for an evidence trail per girth weld. A clean package contains:

• WPS and PQR covering the joint geometry, position, consumable and heat input range used.
• Welder qualification record valid for the position and direction of progression.
• Consumable certificates with lot traceability to the actual rod or wire used on the weld.
• Preheat and interpass temperature record (see ISO 13916 monitoring for the equivalent ISO method).
• NDT report (RT or AUT) with operator qualification certificates, calibration record and the acceptance regime applied (Section 9 vs Annex A).
• For Annex A: the ECA report, the AUT procedure qualification record and the project specification clause authorising Annex A use.

Organisations running large pipeline programs typically replace paper folders with a digital NCR and traceability layer. See the welding NCR management flow for how an NCR raised on a girth weld feeds the repair workflow without losing the inspector–welder–procedure linkage.

How real-time monitoring complements API 1104

API 1104 is an acceptance standard, not a process-control standard. It tells the inspector what an acceptable weld looks like after it has been made; it does not, by itself, prevent a girth weld from being deposited outside its qualified parameter window. Real-time monitoring of arc current, voltage, travel speed, heat input and interpass temperature closes that loop: girth welds outside the qualified envelope can be flagged before NDT, root cause is captured at the time of deviation, and AUT scan time is concentrated on the welds that need it. For pipeline operators carrying out 80–120 welds per shift, even a 5 % reduction in repair rate justifies the cost of inline monitoring.