Ultrasonic testing acceptance criteria are the limits, set by the construction code rather than the examination standard, that decide whether a UT indication is accepted or repaired. AWS D1.1 classes indications A to D from d = a − b − c; ASME Section VIII and B31.3 reject "other" indications only when amplitude exceeds the reference level and length exceeds a thickness limit.
The same indication can therefore pass under one code and fail under another. This guide covers the codes a US buyer usually meets: AWS D1.1 for structural steel, ASME Section VIII and B31.3 for vessels and process piping, API 1104 for pipelines, and 49 CFR 213.113 plus the owner's specification for rail.
Contents: Examination vs acceptance · Which code applies · AWS D1.1 · ASME VIII · B31.3 · API 1104 · Rail · Over the limit · Limits · FAQ
About the numbers: AWS D1.1, ASME, API and ISO standards are paywalled. The limits below come from the publishers' public pages and from secondary sources that agree with one another. Check every value against the edition named in your contract before accepting or rejecting a weld.
What is the difference between an examination standard and ultrasonic testing acceptance criteria?
An examination standard says how to perform UT; acceptance criteria say what result is acceptable. A technician can locate and size an indication perfectly and still need a separate document to decide whether it passes.
The work runs in this order: written procedure → calibration → examination → indication → evaluation → acceptance decision. The documents split along the same line.
| Document | What it does | Sets acceptance limits? |
|---|---|---|
| ASME BPVC Section V, Article 4 | UT method for welds: equipment, calibration, scanning, recording | No. It defers to the referencing code |
| ISO 16810:2024 | General principles of industrial UT | No. Its scope excludes acceptance criteria and the extent of testing |
| ASTM E164-24 | Contact UT of weldments | No. Limits must be agreed between manufacturer and purchaser |
| ASTM E2700-26 | Contact UT of welds with phased arrays | No. It is a practice, not an acceptance standard |
| AWS D1.1/D1.1M:2025, Clause 8 | Structural steel welds | Yes: Classes A to D |
| ASME Section VIII, Division 1 | Pressure vessels | Yes, when it calls for UT of welds |
| ASME B31.3 | Process piping | Yes |
| API Std 1104 | Pipeline girth welds | Yes |
| ISO 11666:2018 | Full-penetration welds in ferritic steel | Yes: acceptance levels 2 and 3 |
| 49 CFR 213.113 | Defective rails in track | Remedial action by defect type and size |
ASTM puts the point plainly in E164-24: when UT is the basis of weld acceptance, "there should be agreement between the manufacturer and the purchaser" on the reference standards and limits (ASTM E164-24).
ISO 11666:2018 is the European-style example of a true acceptance standard. It sets acceptance levels 2 and 3 for welds 8 to 100 mm thick, tied to ISO 5817 quality levels B and C, for testing performed to ISO 17640. US projects rarely invoke it unless the client is working to EN or ISO practice.
Key takeaway: identify the construction code first. A UT method standard will never give you a reject number.
Which code sets the acceptance criteria for your job?
The component and the contract decide the code, and the code decides how an indication is judged. The table below shows the rating logic each code uses.
| Component | Usual code | How an indication is judged | Qualifier you must state |
|---|---|---|---|
| Structural steel building or plant welds | AWS D1.1 | Rating d = a − b − c, then class A–D and a length limit | Static or cyclic loading; weld thickness; probe angle |
| Pressure vessel welds | ASME Section VIII, Div. 1 | Flaw type first, then amplitude above reference and length | Thickness band; whether UT replaces RT |
| Process piping welds | ASME B31.3 | Linear indication: amplitude above reference and length | Tw of the thinner component; fluid service |
| Gas and liquid pipeline girth welds | API 1104 | Indication type (linear surface, linear buried, volumetric) and aggregate length | Edition (21st is the one in 49 CFR) |
| Bridges | AWS D1.5 | D1.5's own UT tables | Fracture-critical members |
| Rail in track | 49 CFR 213.113 plus owner's spec | Defect type, then percent of head area or length | Track class; who is the designated person |
| Railcars and locomotives | AWS D15.1/D15.1M:2025 | D15.1 and the car builder's spec | Not a rail-in-track standard |
If a project specification is stricter than the code, the specification governs. The code is the floor.
How does AWS D1.1 rate and accept a UT indication?
AWS D1.1 rates each indication with d = a − b − c and looks the rating up in a static or cyclic table by weld thickness and probe angle. The result is Class A, B, C or D, and each class has its own length rule. This is a summary; our full guide to AWS D1.1 acceptance criteria covers the visual limits, table numbering and personnel rules in depth.
| Term | Meaning |
|---|---|
| a | Indication level: dB needed to bring the flaw echo to the reference line |
| b | Reference level: dB needed to bring the 0.060 in. (1.5 mm) side-drilled hole echo to the same line |
| c | Attenuation factor: 2 × (sound path in inches − 1) dB |
| d | Indication rating: a − b − c |
The sign convention trips people up. A lower, more negative rating means a stronger reflector, because the flaw needed less gain than the reference hole. Class A, the most severe class, therefore sits at the low end of each table.
| Class | Statically loaded | Cyclically loaded |
|---|---|---|
| A (large) | Reject at any length | Reject at any length |
| B (medium) | Reject if longer than 3/4 in. | Reject if longer than 3/4 in. |
| C (small) | Reject if longer than 2 in. | Reject if longer than 2 in. in the middle half of the thickness, or 3/4 in. in the top or bottom quarter |
| D (minor) | Accept at any length | Accept at any length |
Thickness scope: the D1.1 Part F procedure covers groove welds from 5/16 in. to 8 in. (8 to 200 mm). Thinner or thicker welds need a procedure qualified under the code's alternative provisions. D1.1 also doesn't require UT on its own; the contract, AISC 360 or AISC 341 does.
Worked example: one rating, two loading cases
Setup. A 3/4 in. CJP butt weld is scanned with a 70° probe. The reference level from the side-drilled hole is b = 65 dB. An indication peaks at the reference line at a = 79 dB, with a sound path of 3.0 in.
- Attenuation: c = 2 × (3.0 − 1) = 4 dB
- Rating: d = 79 − 65 − 4 = +10
- Location check: 3.0 in. × cos 70° = 1.03 in. of depth travel, which is past the 0.75 in. back wall, so the reflector is on the second leg at about 2 × 0.75 − 1.03 = 0.47 in. below the scanning surface
- Surface distance: 3.0 in. × sin 70° = 2.82 in. from the probe exit point
Statically loaded. In the 5/16 to 3/4 in. row at 70°, Class A is +5 and lower and Class B is +6, and published summaries put Class C at +7 and Class D at +8 and higher. So +10 is Class D: accept at any length.
Cyclically loaded. Secondary reproductions of the cyclic table place Class A at +10 and lower in the same row. The same +10 is Class A: reject at any length.
Nothing about the flaw changed; only the loading column did. A 2 dB error in c would also move this indication across a class boundary, which is why the report must record a, b, c and d for every indication. A crack identified by any means is rejected whatever its rating.
Calibration rules that make the rating valid
A rating is only as good as the calibration behind it. Secondary summaries of D1.1 give these requirements:
| Requirement | Summary value (confirm in your edition) |
|---|---|
| Instrument | Pulse-echo, for search units of 1 to 6 MHz, A-scan display |
| Gain control | Calibrated in 1 or 2 dB steps over at least 60 dB |
| Recalibration | On a change of operator, at 30 minutes maximum, and whenever the circuitry is disturbed (transducer, cable, battery or power change) |
| Reference reflector | 0.060 in. (1.5 mm) side-drilled hole in an IIW-type block |
A "recheck every two hours" is not the D1.1 interval. See our guides to UT formulas and DAC curves for the sizing and amplitude basics.
Key takeaway: always say "static or cyclic" and the thickness before quoting a D1.1 class.
What are the ASME Section VIII ultrasonic acceptance criteria?
Under ASME Section VIII, Division 1, Mandatory Appendix 12, indications above 20% of the reference level are investigated; cracks, lack of fusion and incomplete penetration are rejected at any length; other imperfections are rejected only if their amplitude exceeds the reference level and their length exceeds a thickness-based limit. Section V, Article 4 sets how the examination is done; Appendix 12 sets acceptance.
The "and" is the part most summaries drop. A long, low-amplitude indication that is not a crack, lack of fusion or incomplete penetration is not rejectable on length alone.
| Weld thickness t | Maximum length of an "other" imperfection above reference level |
|---|---|
| Up to 3/4 in. (19 mm) | 1/4 in. (6 mm) |
| 3/4 in. to 2¼ in. (19 to 57 mm) | t/3 |
| Over 2¼ in. (57 mm) | 3/4 in. (19 mm) |
Source: Appendix 12 wording as reproduced in UT procedures and NDT references. Confirm against the 2025 edition of Section VIII, Division 1, or the edition on your contract.
Worked example: four indications in a 1½ in. vessel seam
For t = 1.5 in., the length limit is t/3 = 0.50 in. Each indication below exceeded 20% of the reference level, so each was investigated.
| Indication | Characterized as | Amplitude vs reference | Length | Result |
|---|---|---|---|---|
| 1 | Slag | 140% | 0.40 in. | Accept: above reference, but under 0.50 in. |
| 2 | Slag line | 60% | 0.70 in. | Accept: over 0.50 in., but not above reference |
| 3 | Slag line | 130% | 0.70 in. | Reject: above reference and over 0.50 in. |
| 4 | Lack of fusion | 50% | 0.20 in. | Reject: lack of fusion at any length |
Indication 2 is the one a careless evaluation rejects. Indication 4 is the one an amplitude-only evaluation accepts.
Watch the trigger for UT. Appendix 12 applies where Division 1 calls for UT of welds under that appendix. When UT is used in place of radiography, Division 1 points to different, more detailed acceptance rules (workmanship or fracture-mechanics based) under UW-51 in recent editions. Our guide to ASME weld ultrasonic inspection covers the vessel side in more depth.
How does ASME B31.3 accept UT indications?
ASME B31.3 paragraph 344.6.2 rejects a linear-type discontinuity when its amplitude exceeds the reference level and its length exceeds a limit based on Tw, the nominal wall thickness of the thinner component in the butt weld. The bands match Section VIII.
| Tw | Length limit |
|---|---|
| Up to 3/4 in. (19 mm) | 1/4 in. (6 mm) |
| Over 3/4 in. to 2¼ in. (19 to 57 mm) | Tw/3 |
| Over 2¼ in. (57 mm) | 3/4 in. (19 mm) |
The examination itself follows Section V, Article 4, with B31.3's own allowances for calibration blocks. B31.3 also sets how much to examine, which differs by fluid service. Published summaries put normal fluid service at no less than 5% random radiographic or ultrasonic examination of circumferential butt welds, with heavier examination for severe cyclic conditions; read paragraph 341.4 in your edition for the exact extent.
Two points cause confusion. First, requirements for examining pipe and tubing as product forms are material requirements, not weld acceptance criteria, so keep them separate. Second, ASME B31.1 (power piping) has its own UT acceptance paragraph. Some summaries say it mirrors Section VIII, but read B31.1 itself rather than assume it.
Key takeaway: in both Section VIII and B31.3, "other" indications need to fail on amplitude and length together.
What are the API 1104 UT acceptance standards?
API 1104 judges pipeline girth weld UT indications by type and aggregate length, not by a single amplitude-and-length formula, and it rejects cracks outright. Secondary summaries of the UT acceptance standards give these limits for linear surface, linear buried and volumetric root indications:
| Indication type | Rejected when |
|---|---|
| Crack | Any crack |
| Linear surface (LS), open to the OD or ID | Aggregate length over 1 in. (25 mm) in any continuous 12 in. (300 mm) of weld, or over 8% of the weld length |
| Linear buried (LB) | Aggregate length over 2 in. (50 mm) in any continuous 12 in. of weld, or over 8% of the weld length |
| Volumetric root (VR) | Maximum dimension over 1/4 in. (6 mm), or aggregate over 1/2 in. (13 mm) in any continuous 12 in. |
| Volumetric cluster and individual | Size limits in the edition's UT acceptance section |
Edition matters here more than anywhere. API's current edition is the 22nd, but 49 CFR 192.7 incorporates the 21st edition (September 2013) with its errata and addenda, for welder and procedure qualification and for 192.241(c) weld acceptance (49 CFR 192.7 on eCFR). On a PHMSA-regulated gas line, the regulatory floor is the 21st edition unless the operator's procedures adopt something stricter.
API 1104 also has an annex that allows alternative acceptance standards based on fracture mechanics, used with automated UT on some mainline projects. That route needs a qualified procedure and the operator's approval; it is not a field choice.
What are the acceptance criteria for rail ultrasonic testing?
For rail in track, there is no accept/reject class table. FRA's 49 CFR 213.113 names the defect types and sets a remedial action by defect size, and the track owner's specification and AREMA recommended practice add the rest. The regulation puts the decision with a qualified person:
"When an owner of track learns that a rail in the track contains any of the defects listed in the table contained in paragraph (c) of this section, a person designated under § 213.7 shall determine whether the track may continue in use." (49 CFR 213.113(a))
The remedial action table sizes transverse defects as a percentage of the rail head cross-section weakened, and longitudinal defects by length in inches.
| Defect (transverse group) | Head area weakened | Remedial action note |
|---|---|---|
| Transverse fissure, detail fracture, engine burn fracture, defective weld | 5% to under 25% | C |
| Same | 25% to under 60% | D |
| Same | 60% to under 100% | A2, or E and H |
| Same | 100% | A, or E and H |
| Compound fissure | 5% to under 70% | B |
| Compound fissure | 70% to under 100% | A2 |
| Compound fissure | 100% | A |
Source: the remedial action table as revised by FRA's 2014 Improving Rail Integrity final rule (79 FR 4234, January 24, 2014). Check the current table on eCFR before relying on a cell.
What the notes require, in short:
| Note | Requirement (summary) |
|---|---|
| A | A designated person visually supervises each movement over the rail |
| A2 | A designated person inspects visually; trains may then run at 10 mph for up to 24 hours before the next inspection or repair |
| B | Speed set by a designated person with at least one year of track supervisory experience, no more than 30 mph |
| C | Joint bars through the outermost holes within 10 days; 10 mph if not applied by then |
| D | Joint bars through the outermost holes within 7 days; 10 mph if not applied by then |
| E and H | Joint bars bolted per 213.121(d) and (e), then 50 mph or the class limit, whichever is lower |
Horizontal and vertical split heads, split webs, piped rail, head-web separations and bolt hole cracks are sized by length, with tighter actions as they grow past 1 in., 1½ in. or 4 in. depending on the type. A defective weld under 213.113 is a field or plant weld with discontinuities over 5% of the rail head area individually or 10% in aggregate, oriented in or near the transverse plane.
Where the owner's spec and AREMA fit
FRA sets the minimum; the owner sets the acceptance standard for new work. A transit agency's thermite or flash-butt weld specification typically names the UT procedure, the reference reflector and the reject level for new welds before they go into service, often citing AREMA's Manual for Railway Engineering Chapter 4 (Rail). AREMA publishes recommended practices, not regulations, so its material governs only when the owner's specification adopts it. AWS D15.1 is a different document: it covers welding of railcars and locomotives, not rail in track.
For the field side of rail weld UT, see our guides to aluminothermic weld inspection, flash-butt weld inspection and FRA defect codes.
Key takeaway: in rail, detection is the UT operator's job; disposition under 213.113 belongs to the owner's designated person.
What happens when an indication exceeds the acceptance limit?
An over-limit indication is confirmed before it is rejected, and the disposition is documented. A practical sequence:
- Rescan to confirm the indication from more than one direction or angle.
- Verify its location, depth and length, and rule out geometry such as root reinforcement or a backing bar.
- Recheck calibration if the procedure or code requires it.
- Confirm the governing code, edition and loading or service category.
- Compare the result with the applicable limit and record a, b, c and d (D1.1) or amplitude and length (ASME, API).
- Report the indication and the disposition to the inspector or engineer responsible for acceptance.
- Repair, or apply an engineering disposition where the code allows one.
- Re-examine the repaired area to the same procedure.
This reduces both unnecessary repairs and missed defects. Under D1.1, the CWI or engineer decides acceptance; under 213.113, the owner's designated person decides whether track stays in service.
How do PAUT and TOFD change the acceptance criteria?
They change how indications are found and sized, not the acceptance limits. ASTM E2700-26 covers phased array UT of welds and notes that sizing needs suitable procedures. D1.1 has a PAUT annex with its own procedure requirements. API 1104's fracture-mechanics annex is the main case where a different technique comes with different acceptance criteria, and only because the annex says so. See our guides to phased array ultrasonic testing and PAUT vs TOFD for rail welds.
What ultrasonic testing acceptance criteria don't cover
Acceptance tables assume the procedure fits the part. When it doesn't, the numbers stop meaning what they say.
| Situation | Why the tables may not apply or may not be enough |
|---|---|
| Thin welds below 5/16 in. (D1.1) | Outside the Part F scope; near-surface resolution suffers and a qualified alternative procedure is needed |
| Complex tubular T-, Y- and K-connections | Need procedures written for the geometry |
| Austenitic stainless and coarse-grained welds | Beam skew and scatter make amplitude-based rules unreliable |
| Smooth planar flaws at a poor angle to the beam | Low amplitude; can rate as Class D or fall below the reference level |
| Thick D1.1 welds | The 2 dB/in. attenuation model can be about 6 dB off at 60° and 70° near 6 in. of sound path (Holloway, CINDE Journal, 2017) |
| Rail with shelling or head checks | Surface damage can mask internal defects from the probe |
| Thermite welds in track | Test cars cover little of the rail base, so field welds need hand UT |
| A project spec stricter than the code | The spec governs |
The common mistakes follow the same pattern:
| Mistake | Correct practice |
|---|---|
| One reject number for every job | Identify the code, edition and qualifier first |
| Treating ASME Section V or ISO 16810 as acceptance criteria | They are method standards |
| Rejecting an ASME "other" indication on length alone | Amplitude above reference and length over the limit |
| Reading a D1.1 positive rating as "worse" | Lower ratings are stronger reflectors |
| Quoting a D1.1 class without the loading column | State static or cyclic |
| Using the 22nd edition of API 1104 for a regulated gas line by default | 49 CFR 192.7 incorporates the 21st |
| Calling the UT operator's call on rail a "reject" | 213.113 disposition belongs to the designated person |
How EMA Industries applies this
EMA Industries performs AWS D1.1 structural weld UT, MT and PT, ultrasonic rail flaw testing and rail weld inspection, and NDT Level III services, including procedure writing and WPS/PQR support. We're based in Southern California and mobilize nationwide. Before scanning, the governing code, edition and loading or service category should be confirmed from the contract, and the report should record the reference level, rating, class or amplitude, and length of every indication evaluated, so the person who decides acceptance can see the basis for the call. See our ultrasonic examination services or contact EMA Industries to scope an inspection.
Related reading: AWS D1.1 acceptance criteria · Weld defect types · Lack of fusion in welds · SNT-TC-1A requirements · 49 CFR 213.237 rail flaw inspection
Frequently asked questions
What are the ASME acceptance criteria for ultrasonic testing?
They come from the construction code, not Section V. In Section VIII, Division 1, Appendix 12, indications above 20% of the reference level are investigated, cracks, lack of fusion and incomplete penetration are rejected at any length, and other imperfections are rejected only when amplitude exceeds the reference level and length exceeds 1/4 in., t/3 or 3/4 in. depending on thickness. B31.3 uses the same length bands for linear indications.
What is an example of acceptance criteria?
AWS D1.1's Class B rule is a typical one: an indication whose rating falls in the Class B band for that thickness and probe angle is rejected if it is longer than 3/4 in. In rail, the equivalent is the 213.113 rule that a transverse fissure weakening 25% to under 60% of the head area requires joint bars within 7 days.
Who prepares acceptance criteria?
Code committees write them (AWS, ASME, API), regulators set them for regulated assets (FRA for rail in track, PHMSA through its incorporation of API 1104), and the engineer or owner adds project-specific limits in the contract. Where no code applies, an NDT Level III typically drafts criteria in a written procedure for the owner to approve.
What is the minimum thickness for ultrasonic weld testing under AWS D1.1?
The D1.1 Part F procedure starts at 5/16 in. (8 mm) and runs to 8 in. (200 mm). Thinner welds can still be tested ultrasonically, but only with a procedure qualified for that thickness, not by applying the Part F tables directly.
What is the difference between accuracy and acceptance criteria?
Accuracy describes how closely a measurement matches a flaw's real position and size, and it depends on calibration against known reflectors. Acceptance criteria are the limits the measured result is judged against. An accurate measurement compared with the wrong code still produces the wrong decision.
Sources
- 49 CFR 213.113, Defective rails, eCFR (text read on Cornell LII, October 2026)
- Track Safety Standards; Improving Rail Integrity, final rule, 79 FR 4234, FRA, January 24, 2014 (remedial action table and notes)
- 49 CFR 192.7, Incorporation by reference, eCFR (text read on Cornell LII): API 1104 21st edition
- AWS announces release of D1.1/D1.1M:2025, American Welding Society, April 17, 2025
- AWS D15.1/D15.1M:2025, Railroad Welding Specification for Cars and Locomotives, AWS
- ASTM E164-24, Contact Ultrasonic Testing of Weldments, ASTM International
- ASTM E2700-26, Contact Ultrasonic Testing of Welds Using Phased Arrays, ASTM International
- ASNT Recommended Practice No. SNT-TC-1A (2024), American Society for Nondestructive Testing
- AREMA publications catalog, American Railway Engineering and Maintenance-of-Way Association (Manual for Railway Engineering, Chapter 4)
- Ultrasonic flaw detection tutorial: AWS weld inspection, Evident Scientific (rating formula and sign convention)
- Structural UT: Variables Affecting Attenuation and Review of the 2 dB per Inch Model, P. Holloway, CINDE Journal, 2017
- ISO 16810:2024, Non-destructive testing – Ultrasonic testing – General principles, and ISO 11666:2018, Non-destructive testing of welds – Ultrasonic testing – Acceptance levels, ISO (catalog scope statements)
- ASME BPVC Section V (Article 4) and Section VIII, Division 1 (Mandatory Appendix 12), 2025 edition; ASME B31.3, para. 344.6; API Std 1104, 21st and 22nd editions (paywalled; values above from agreeing secondary sources, to be confirmed against the licensed text)