Every API 1104 Acceptance Limit, and the Repair Rules Around Them

API 1104 Section 9 judges pipeline girth welds on imperfection length, not height. Each discontinuity type carries an individual length limit of 1 in. or 2 in., an aggregate limit in any continuous 12 in. (300 mm) of weld, and for several types a cap at 8% of weld length. Cracks are defects at any size unless they are shallow crater or star cracks.

Section 9 is a workmanship standard. It asks whether a weld looks like the weld the qualified procedure was supposed to produce, and it answers with lengths: an individual indication limit, an aggregate limit inside any continuous 12 in. of weld, and for several discontinuity types a percentage of total weld length. Height never enters. That is deliberate, because decades of pipeline service backed those empirical limits, but it means a short deep flaw passes and a long shallow one fails. Appendix A exists for operators willing to buy back that margin with fracture mechanics, CTOD testing and a documented stress analysis. Layered on top are the federal rules: how many welds get tested at all is set by 49 CFR, not by API, and the edition the regulator enforces is not the edition currently in print.

Source: Clause text read from API Standard 1104, 19th edition (1999), Sections 8.4, 9.1 through 9.7, 10.1 through 10.5, 11.1 and Appendix A, published in full via Public.Resource.Org under 49 CFR 195.214(a). Current published edition confirmed as the 22nd, July 2021, via apiwebstore.org and pubs.aws.org listings. Incorporation by reference confirmed at 49 CFR 192.7 and 195.3 as the 21st edition, September 2013, including Errata 1 through 5 and Addenda 1 and 2. Extent-of-testing percentages read from 49 CFR 192.243 and 195.234; acceptance referral from 49 CFR 192.241(c) and 195.228(b).

Technically reviewed by Anoop Rayavarapu — ASNT NDT Level III (UT, RT, MT, PT, VT, ET) · API 653 · ISO 9001:2015 Lead Auditor
API 1104 Section 9.3 radiographic acceptance, pipe 2.375 in. (60.3 mm) outside diameter and larger
CodeDiscontinuityIndividual indication is a defect aboveAggregate in any continuous 12 in. (300 mm)Percent-of-weld limit
IPInadequate penetration without high-low1 in. (25 mm)1 in. (25 mm)8% of weld length in any weld under 12 in. long
IPDInadequate penetration due to high-low2 in. (50 mm)3 in. (75 mm)None stated
ICPInadequate cross penetration2 in. (50 mm)2 in. (50 mm)None stated
IFIncomplete fusion1 in. (25 mm)1 in. (25 mm)8% of weld length in any weld under 12 in. long
IFDIncomplete fusion due to cold lap2 in. (50 mm)2 in. (50 mm)8% of weld length
ESIElongated slag inclusion2 in. (50 mm) length, or 1/16 in. (1.6 mm) width2 in. (50 mm)ESI plus ISI above 8% of weld length
ISIIsolated slag inclusion1/8 in. (3 mm) width, or more than four indications at that width in 12 in.1/2 in. (13 mm)ESI plus ISI above 8% of weld length
CCracksAny crack that is not a shallow crater or star crack; a crater or star crack above 5/32 in. (4 mm)Not applicableRepair permitted only below 8% of weld length
Read from API 1104, 19th edition (1999), Section 9.3. Parallel elongated slag indications separated by approximately the width of the root bead count as a single indication unless either exceeds 1/32 in. (0.8 mm) in width. Pipe under 2.375 in. (60.3 mm) outside diameter is judged by the separate, thickness-proportional limits in 9.3.7.3 and 9.3.8.3. The current published edition is the 22nd (July 2021); 49 CFR 192.7 and 195.3 incorporate the 21st edition (September 2013) with Errata 1 through 5 and Addenda 1 and 2.

What Section 9 measures, and what it deliberately ignores

The acceptance standards in Section 9 apply to imperfections located by radiographic, magnetic particle, liquid penetrant and ultrasonic methods, and may also be applied to visual inspection. They are empirical workmanship criteria that place primary importance on imperfection length, and Appendix A says so explicitly before offering an alternative. That record of reliability in pipeline service is the justification for ignoring height, and it is also the reason a deep, short lack of fusion can pass a radiographic acceptance that a long, shallow one fails.

Section 9.2 keeps a door open that inspectors forget exists. All nondestructive test methods are limited in the information their indications carry, so the company may reject any weld that appears to meet the acceptance standards if, in its opinion, the depth of an imperfection may be detrimental to the weld. That is a contractual right held by the operator, not by the inspection contractor, and it is exercised through the company representative rather than through the film interpreter's report.

Radiographic densities referred to throughout 9.3 are based on negative images, which matters for burn-through and internal concavity where the decision turns on whether an image is denser than the thinnest adjacent parent material. Getting that convention wrong inverts two acceptance decisions. The comparison of what radiography and ultrasonics each see well is set out on RT versus UT for weld inspection.

The radiographic limits, discontinuity by discontinuity

Inadequate penetration without high-low, coded IP, is incomplete filling of the weld root. It is a defect when an individual indication exceeds 1 in. (25 mm), when the aggregate in any continuous 12 in. (300 mm) of weld exceeds 1 in. (25 mm), or when the aggregate exceeds 8% of the weld length in any weld shorter than 12 in. Inadequate penetration due to high-low, coded IPD, arises when one edge of the root is exposed because adjacent joints are misaligned. Its individual limit doubles to 2 in. (50 mm) and its aggregate limit is 3 in. (75 mm), the most generous aggregate in Section 9.

Inadequate cross penetration, coded ICP, is a subsurface condition between the first inside pass and the first outside pass caused by inadequately penetrating the vertical land faces. It is a defect above 2 in. (50 mm) individually and above 2 in. (50 mm) aggregate in any 12 in. Incomplete fusion, coded IF, is surface-connected between weld metal and base material, and carries the tightest common limits: 1 in. (25 mm) individually, 1 in. (25 mm) aggregate, and 8% of the weld length in any weld under 12 in. long.

Incomplete fusion due to cold lap, coded IFD, sits between adjacent weld beads or between weld metal and base metal without breaking the surface. It is a defect above 2 in. (50 mm) individually, above 2 in. (50 mm) aggregate in any 12 in., or above 8% of the weld length. The distinction between IF and IFD is surface connection, and it is the distinction most often argued over on film. Codifying how your interpreters make that call, consistently across crews, is procedure work rather than interpretation work, and belongs in the written technical procedure.

Slag has two codes and five separate triggers

Elongated slag inclusions, coded ESI, are continuous or broken slag lines, the wagon tracks found at the fusion zone. Isolated slag inclusions, coded ISI, are irregularly shaped and may sit anywhere in the weld. When the size of a radiographic slag indication is measured, the maximum dimension is treated as its length, which prevents an interpreter from taking a favourable axis.

For pipe 2.375 in. (60.3 mm) outside diameter and larger, slag is a defect when: an ESI indication exceeds 2 in. (50 mm) in length; the aggregate ESI length in any continuous 12 in. (300 mm) exceeds 2 in. (50 mm); an ESI indication exceeds 1/16 in. (1.6 mm) in width; the aggregate ISI length in any continuous 12 in. exceeds 1/2 in. (13 mm); an ISI indication exceeds 1/8 in. (3 mm) in width; more than four ISI indications at the maximum 1/8 in. width appear in any continuous 12 in.; or the aggregate of ESI and ISI exceeds 8% of the weld length.

One note governs how wagon tracks are counted. Parallel ESI indications separated by approximately the width of the root bead are considered a single indication, unless the width of either of them exceeds 1/32 in. (0.8 mm), at which point they become separate indications. That single sentence decides whether a pair of tracks is measured once or twice against the 2 in. limit. On pipe under 2.375 in. outside diameter, the limits change shape entirely and become proportional to wall thickness rather than fixed.

Porosity has its own arithmetic

Individual or scattered porosity, coded P, is a defect when the size of an individual pore exceeds 1/8 in. (3 mm), when it exceeds 25% of the thinner of the nominal wall thicknesses joined, or when the distribution of scattered porosity exceeds the concentration permitted by the two distribution figures in Section 9. The percentage-of-wall trigger is the one that bites on thin wall: on 0.250 in. pipe it drops the allowable pore to 0.0625 in., half the absolute limit.

Cluster porosity, coded CP, occurring in any pass except the finish pass is judged against the scattered porosity criteria. Cluster porosity in the finish pass is a defect when the diameter of the cluster exceeds 1/2 in. (13 mm), when the aggregate of CP in any continuous 12 in. (300 mm) exceeds 1/2 in. (13 mm), or when an individual pore within a cluster exceeds 1/16 in. in size.

Hollow-bead porosity, coded HB, is elongated linear porosity in the root pass and carries four triggers. It is a defect when an individual indication exceeds 1/2 in. (13 mm); when the aggregate in any continuous 12 in. (300 mm) exceeds 2 in. (50 mm); when individual indications each greater than 1/4 in. (6 mm) in length are separated by less than 2 in. (50 mm); or when the aggregate of all HB exceeds 8% of the weld length. The separation rule catches strings of short hollow bead that would otherwise pass on individual length alone.

Burn-through and internal concavity turn on film density

Internal concavity, coded IC, is the one condition where any length is acceptable. Any length passes provided the density of the radiographic image of the internal concavity does not exceed that of the thinnest adjacent parent material. Where areas do exceed that density, the burn-through criteria apply instead. So the same root profile is either unlimited or tightly capped depending on a densitometer reading.

A burn-through, coded BT, is a portion of the root bead where excessive penetration blew the weld puddle into the pipe. For pipe 2.375 in. (60.3 mm) outside diameter and larger, it is a defect when the maximum dimension exceeds 1/4 in. (6 mm) and the density of the burn-through image exceeds that of the thinnest adjacent parent material; when the maximum dimension exceeds the thinner of the nominal wall thicknesses joined with that same density condition; or when the sum of the maximum dimensions of separate burn-throughs meeting that density condition exceeds 1/2 in. (13 mm) in any continuous 12 in. (300 mm) of weld or in the total weld length, whichever is less.

On pipe under 2.375 in. outside diameter the third trigger changes character: more than one burn-through of any size is a defect when the density of more than one of the images exceeds that of the thinnest adjacent parent material. Small-diameter pipe is therefore judged on count rather than on summed dimension. Every one of these decisions rests on a density comparison, which makes densitometer calibration records part of the acceptance evidence, not an optional attachment.

Undercutting and accumulation: two rules that overlap

Undercutting is a groove melted into the parent material adjacent to the toe or root of the weld and left unfilled. Adjacent to the cover pass it is coded EU, adjacent to the root pass IU. Radiographically, the pair is a defect when their aggregate length in any combination exceeds 2 in. (50 mm) in any continuous 12 in. (300 mm) of weld, or when it exceeds one-sixth of the weld length.

Section 9.7 provides a second rulebook for when visual and mechanical means measure depth, and it supplements rather than replaces the visual inspection requirements elsewhere in the standard. Undercut deeper than 1/32 in. (0.8 mm), or deeper than 12.5% of pipe wall thickness, whichever is smaller, is not acceptable at any length. Undercut between 1/64 in. (0.4 mm) and 1/32 in., or between 6% and 12.5% of wall, is limited to 2 in. (50 mm) in a continuous 12 in. weld length or one-sixth of the weld length, whichever is smaller. Undercut at or below 1/64 in. (0.4 mm) or 6% of wall is acceptable regardless of length. Where both mechanical and radiographic measurements are available, the mechanical measurement governs.

Accumulation of imperfections, coded AI, catches welds that pass every individual criterion and still contain too much. Excluding inadequate penetration due to high-low and excluding undercutting, any accumulation is a defect when the aggregate length of indications in any continuous 12 in. (300 mm) exceeds 2 in. (50 mm), or when the aggregate exceeds 8% of the weld length. Interpreters who evaluate discontinuity by discontinuity and never total the film miss this one routinely.

Ultrasonic acceptance uses a different vocabulary entirely

Section 9.6 abandons the radiographic codes and classifies by geometry instead. Linear indications have their greatest dimension in the weld length direction and may be caused by IP, IPD, ICP, IF, IFD, ESI, cracks, undercutting or hollow bead. Transverse indications have their greatest dimension across the weld and may be caused by cracks, isolated slag or cold lap at start-stops. Volumetric indications are three-dimensional and may be caused by internal concavity, burn-through, isolated slag, porosity or cluster porosity. Relevant indications are evaluated at the evaluation level given in the ultrasonic test methods section.

The acceptance limits then run by classification. Indications determined to be cracks are defects. Linear surface indications open to the inside or outside surface are defects when their aggregate in any continuous 12 in. (300 mm) exceeds 1 in. (25 mm), or exceeds 8% of the weld length. Linear buried indications, subsurface and not surface-connected, are defects above 2 in. (50 mm) aggregate in any 12 in. or above 8% of weld length. Transverse indications other than cracks are treated as volumetric and coded T.

Volumetric cluster indications are defects when the maximum dimension exceeds 1/2 in. (13 mm). Volumetric individual indications are defects when the maximum dimension exceeds 1/4 in. (6 mm) in both width and length. Volumetric root indications interpreted as open to the inside surface are defects when the maximum dimension exceeds 1/4 in. (6 mm) or when total length exceeds 1/2 in. (13 mm) in any continuous 12 in. Any accumulation of relevant indications above evaluation level is a defect above 2 in. (50 mm) in any 12 in. or above 8% of the weld length. Geometric indications from alignment offset, reinforcement profile changes, internal chamfering and mode conversion are explicitly called out as not relevant to acceptability.

Surface methods, and the certification rule behind them

Magnetic particle and liquid penetrant acceptance run on parallel logic with one numerical difference. For magnetic particle, an indication with a maximum dimension of 1/16 in. (1.6 mm) or less is nonrelevant. For liquid penetrant, the nonrelevant threshold is stated as 1/16 in. (2 mm). Any larger indication believed to be nonrelevant is regarded as relevant until re-examined by the same or another method, and the surface may be ground or otherwise conditioned before re-examination.

Relevant indications are defects when linear indications are evaluated as crater or star cracks exceeding 5/32 in. (4 mm) in length, when linear indications are evaluated as cracks other than crater or star cracks, or when linear indications are evaluated as incomplete fusion exceeding 1 in. (25 mm) in total length in a continuous 12 in. (300 mm) of weld or 8% of the weld length. Rounded indications are evaluated against the scattered and cluster porosity criteria, with the maximum dimension of a rounded indication taken as its size. Linear means length more than three times width; rounded means three times or less.

Section 8.4 sets the personnel rule that the whole standard rests on. Nondestructive testing personnel shall be certified to Level I, II or III in accordance with ASNT Recommended Practice No. SNT-TC-1A, ACCP, or any other recognised national certification program acceptable to the company for the test method used, and only Level II or III personnel may interpret test results. Level I and II are recertified at least every three years; Level III at least every five. Those recertification intervals are shorter than several other codes use, which is covered on the SNT-TC-1A page and against the alternative scheme on ANSI/ASNT CP-189.

Workmanship versus fitness-for-purpose, and the repair limits

Appendix A replaces length-based workmanship with fracture mechanics. It gives more generous allowable imperfection sizes, and it charges for them with additional procedure qualification tests, stress analyses and inspections. The scope is narrow: only circumferential welds between pipes of equal nominal wall thickness, with pump and compressor station welds excluded, mainline fittings and valves excluded, repair welds excluded, and welds subject to applied axial strain above 0.5% excluded. It applies only to pipeline sections where nondestructive inspection is performed on essentially all girth welds.

The entry price is CTOD testing. For each welding procedure, both the weld metal and the heat-affected zone are tested, and each must satisfy the fracture toughness requirement. Each test consists of at least three valid specimens at or below the lowest anticipated service temperature, taken from the nominal twelve, three and six o'clock positions. A procedure may qualify to a minimum of either 0.005 in. or 0.010 in. CTOD. Length and height and depth below surface must all be established by appropriate techniques whose accuracy has been demonstrated, with any potential inaccuracy included conservatively in the measurement. Buried volumetric imperfections may take a simplified route through the appendix table rather than the planar curves.

Section 10 governs what happens next. Cracked welds are removed from the line unless the crack length is less than 8% of the weld length and a qualified repair welding procedure is used. Defects in the root and filler beads may be repaired with prior company authorisation; defects in the cover pass may be repaired without it. A qualified repair procedure is required whenever the repair uses a different process from the original weld, or whenever a previously repaired area is repaired again. Repaired areas are inspected by the same means previously used, and repairs must meet the standards of acceptability of Section 9. Federal rules bolt onto that: 49 CFR 192.241(c) and 195.228(b) determine acceptability by Section 9 or Appendix A, but state that Appendix A may not be used to accept cracks. Where an operator needs those two layers reconciled into one auditable acceptance basis, that is the work of ASNT Level III consulting and disciplined report validation.

Does API 1104 measure imperfection height?

Section 9 does not. The workmanship criteria turn on individual length, aggregate length in any continuous 12 in. (300 mm) of weld, and a percentage of total weld length. Height enters only through the alternative acceptance standards in Appendix A, which require the length, the height and the depth below surface of an imperfection to be established before any accept or reject decision is made.

What is the API 1104 crack rule?

A crack of any size or location in the weld is a defect unless it is a shallow crater crack or star crack. A shallow crater or star crack becomes a defect once its length exceeds 5/32 in. (4 mm). Section 10.1.1 permits repair rather than cut-out only when the crack length is less than 8% of the weld length and a qualified repair welding procedure is used.

How much of a pipeline's girth welds must be tested?

API 1104 does not set the quantity; the federal rules do. For hazardous liquid pipelines, 49 CFR 195.234 requires at least 10 percent of each welder's girth welds each welding day, tested over the entire circumference, and 100 percent at road and railroad rights-of-way, water crossings, populated areas, used pipe and tie-ins. For gas, 49 CFR 192.243 sets 10, 15 and 100 percent by location class.

Which edition of API 1104 does PHMSA enforce?

49 CFR 192.7 and 195.3 both incorporate API Standard 1104, 21st edition, September 2013, including Errata 1 through 5 and Addenda 1 and 2. The current published edition is the 22nd, July 2021. A federally regulated pipeline is judged against the incorporated edition unless the operator's own specification calls up a later one, which creates two acceptance baselines on the same spread.

When can Appendix A be used instead of Section 9?

Appendix A covers circumferential welds between pipes of equal nominal wall thickness where nondestructive inspection is performed on essentially all girth welds. It excludes pump and compressor station welds, mainline fittings and valves, repair welds, and any weld subject to applied axial strain above 0.5%. It requires a documented stress analysis and CTOD procedure qualification. Under 49 CFR, Appendix A may not be used to accept cracks.

How are undercut limits measured under API 1104?

Radiographically, external and internal undercut in any combination is a defect when the aggregate exceeds 2 in. (50 mm) in any continuous 12 in. (300 mm) of weld, or exceeds one-sixth of the weld length. When visual and mechanical means determine depth, the Section 9.7 depth-and-length table governs instead, and where both mechanical and radiographic measurements exist, the mechanical measurement governs.

Request a consultation