Table of content
- One weld, several inspection questions
- Inspection happens at three stages
- Visual Testing (VT): the inspection that happens first
- Surface methods: MT and PT
- Volumetric methods: RT and UT/PAUT
- Weld acceptance codes
- Choosing a method by what you’re looking for
- Key takeaways
- FAQs
- References
- Further reading
One weld, several inspection questions
No single NDT method tells you a weld is sound. Visual testing catches what’s on the surface; surface methods find cracks that visual inspection misses; volumetric methods are the only way to see inside the weld at all — and even then, radiography and ultrasonic testing each have blind spots the other covers. This guide walks through the full inspection stack — visual, surface (MT/PT), and volumetric (RT/UT/PAUT) — and the codes that set acceptance criteria for what each one finds. For symbol notation, see OSNDT’s existing welding symbols guide; for joint geometry including fillet welds, see types of weld joints.
Inspection happens at three stages
Weld inspection isn’t one event at the end — it’s commonly organized into three stages:
- Pre-weld: confirming joint preparation, base-metal condition, and that the welding procedure specification (WPS) and procedure qualification record (PQR) are in order before any arc is struck. Pre-weld UT or RT can also check base metal for existing defects.
- In-process: monitoring seam preparation, torch alignment, arc shape, melt pool and bead formation as welding happens, so an operator can correct spatter, undercut or incomplete fusion while it’s still correctable. On multi-pass welds, this includes visual inspection of interpass and root passes before the next pass is laid down.
- Post-weld: checking the finished weld against acceptance criteria for both surface and internal defects — this is where VT, MT, PT, RT and UT/PAUT do their work.
Visual Testing (VT): the inspection that happens first
Final visual inspection is required once a weld is complete, in its final surface and heat-treated condition, before any other NDT is performed — among the required inspection timings industry guidance sets out, a defensible simplification to “VT comes first,” though not an absolute rule for every case. A visual inspector checks the inspection zone (the weld face plus roughly ½ inch to either side), working within about 24 inches of the surface at a viewing angle of 30° or more, for:
- Undercut — a groove melted into the base metal and left unfilled, checked with an undercut gauge.
- Overlap — weld metal rolling over the toe at less than 90°.
- Porosity (“pinholes”) — open holes from trapped gas, visible at the surface.
- Weld profile and reinforcement — bead height and concavity/convexity, checked with gauges.
- Spatter — expelled metal particles, checked against a reference notch.
- Arc strikes — localized remelted or heat-affected metal from a stray arc; these can create stress risers that lead to later cracking.
- Cracks — the most serious weld defect overall.
Surface methods: MT and PT
VT only sees what’s exposed. Magnetic particle testing (MT) and liquid penetrant testing (PT) extend that to surface-breaking flaws VT might miss — but neither sees anything below the surface.
Magnetic particle testing works only on ferromagnetic weld materials — carbon steel and low-alloy steel. Austenitic stainless steel is non-magnetic and gives invalid MT results; duplex stainless (with its mixed ferrite/austenite structure) gives inconsistent or unreliable MT results, so PT is typically preferred for those grades instead. The part is magnetized, fine ferromagnetic particles are applied, and discontinuities create magnetic flux leakage that attracts the particles into a visible indication. MT reaches only surface and shallow near-surface flaws, with sensitivity falling off quickly with depth, and the part must be demagnetized afterward.
Liquid penetrant testing is the standard surface method where MT can’t be used — non-ferromagnetic materials such as stainless steel, duplex, nickel alloys, aluminum and titanium. A low-viscosity penetrant is drawn into open surface defects by capillary action, excess is removed, a developer pulls the penetrant back to the surface, and the indication is read under white or UV-A light. PT finds cracks, porosity, laps and seams that are open to the surface — but, like MT, nothing below it. On stainless and nickel alloys, chloride contamination from PT consumables can cause stress-corrosion cracking, so low-halogen certified consumables are required; for critical butt welds, PT is often paired with a volumetric method.
Both methods share the same fundamental limitation: surface-breaking (or very near-surface) flaws only. Porosity clusters, slag inclusions and lack of fusion buried inside the weld need a volumetric method to find at all.
Volumetric methods: RT and UT/PAUT
Radiography and ultrasonic testing are the only methods that see inside a weld — and they don’t see the same things equally well.
Radiographic testing (RT) passes X-rays or gamma rays through the weld; porosity, slag inclusions, lack of fusion and cracks absorb radiation differently than sound metal, producing a visible indication on film or a digital image. RT produces a permanent record that becomes part of the weld’s data package, but it comes with real operational cost: controlled exclusion zones, dosimetry, licensed operators, and often night work on live sites to limit exposure.
RT’s key limitation is directional: it’s less effective for planar defects oriented parallel to the beam. A tight planar flaw perpendicular to the beam — faces in close contact, little change in beam-path length — can be effectively invisible on RT, even though the same flaw would show clearly on angle-beam ultrasonic testing. Sidewall lack-of-fusion is a classic example: often faint or absent on RT, often clearly visible on UT. For this reason, codes often pair RT with MT or PT — not specifically to patch RT’s crack-detection blind spot, but because the two cover different defect categories: RT for volumetric flaws, MT/PT for surface-breaking ones.
Ultrasonic testing (UT) sends sound waves through the weld and reads their reflections; it’s generally the stronger method for tight cracks (where angle-beam access is available) and for lack-of-fusion, while RT is generally stronger for volumetric flaws like porosity clusters and gives a permanent visual record. Conventional UT uses a single, manually-angled probe and produces an A-scan an operator reads and logs by hand.
Phased array UT (PAUT) replaces that single element with a multi-element probe (typically 16–128 elements), each pulsed with individually controlled timing delays to electronically steer, focus and sweep the beam without moving the probe. This gives PAUT better multi-angle crack coverage, more consistent porosity detection, and more accurate flaw sizing than conventional UT, plus a full encoded digital record (S-scan/B-scan/C-scan) rather than a hand-logged A-scan. Neither conventional UT nor PAUT involves a radiation hazard — a real practical advantage over RT on live sites.
Weld acceptance codes
The same defect found by the same method can be acceptable under one code and rejectable under another — acceptance criteria are code-specific, not universal.
- AWS D1.1/D1.1M (Structural Welding Code — Steel) — governs welded structures in carbon and low-alloy constructional steels: design, fabrication, welder qualification and inspection for building and structural steel work.
- ASME B31.3 (Process Piping) — the primary standard across the full lifecycle of process piping (design, materials, fabrication, examination, testing) for oil & gas, petrochemical, chemical, pharmaceutical, LNG and power generation; excludes utility piping like HVAC or potable water, which fall under other standards. Its NDT methods (VT, MT, PT, RT) are the same families covered above; the 2024 edition allows PAUT or TOFD to partially replace radiography where technically justified.
- ASME Section VIII (Pressure Vessels) — sets its own acceptance criteria by NDT method, with separate appendices for radiography and for MT/PT, referencing ASME Section V for the underlying NDT methods.
- API 1104 (pipeline welding) — covers gas and arc welding of butt, fillet and socket welds in carbon and low-alloy steel piping for transmitting crude petroleum, petroleum products, fuel gases and related fluids, for both new construction and in-service welding; it sets its own acceptance standards for RT, MT, PT, UT and visual results.
Four different codes, four different acceptance criteria for the same nominal defect types — a weld that passes under one code’s porosity limit may not automatically pass under another’s. Specific numeric acceptance limits (undercut depth, porosity-diameter sums, indication-length thresholds) vary by code and edition; check the applicable code directly — AWS D1.1/D1.1M, ASME BPVC Section VIII, ASME B31.3 or API 1104 — rather than relying on a secondary summary.
Choosing a method by what you’re looking for
What you’re checking for | Best method(s) | Why |
|---|---|---|
Surface profile, undercut, overlap, spatter, arc strikes | VT | First, fastest, catches what’s visible before anything else is applied |
Surface-breaking cracks on ferromagnetic steel | MT | Flux-leakage detection, surface/near-surface only |
Surface-breaking cracks on stainless, nickel alloy, aluminum, titanium | PT | Works where MT can’t (non-ferromagnetic materials) |
Porosity clusters, slag inclusions (volumetric, no tight orientation concern) | RT | Permanent image record, strong for volumetric flaws |
Lack of fusion, lack of penetration, tight planar cracks | UT / PAUT | Stronger than RT where the flaw is tight or unfavorably oriented to a radiation beam |
Precise flaw sizing and a full digital record | PAUT | Electronic beam steering, encoded S-scan/B-scan/C-scan data vs. hand-logged A-scan |
In practice, codes rarely rely on one method alone — RT or UT for volumetric coverage is routinely paired with MT or PT for surface cracking, because no single method covers every defect type a weld can develop.
Key takeaways
- Weld inspection runs in three stages — pre-weld, in-process and post-weld — with final VT required before any other NDT is applied.
- VT catches surface profile defects (undercut, overlap, spatter, arc strikes) and visible cracking, but nothing below the surface.
- MT works only on ferromagnetic steel; PT is the standard alternative for stainless, nickel alloys, aluminum and titanium — both find surface-breaking flaws only.
- RT and UT/PAUT are the only methods that see inside a weld, and they have different blind spots: RT is weaker on tight planar cracks oriented parallel to the beam; UT/PAUT is generally stronger there, while RT gives a permanent image record and is often stronger for volumetric porosity.
- PAUT’s electronic beam steering gives better crack coverage, sizing accuracy and digital record-keeping than conventional single-element UT.
- AWS D1.1, ASME B31.3, ASME Section VIII and API 1104 each set their own acceptance criteria for the same nominal defect types — a weld passing one code’s limits doesn’t automatically pass another’s.
FAQs
What’s the first inspection method applied to a weld? Visual testing. Final VT is required once the weld is complete and in its final condition, before any other NDT method is applied and before the weld becomes inaccessible.
Why would you use PT instead of MT? MT only works on ferromagnetic materials like carbon and low-alloy steel. For stainless steel, nickel alloys, aluminum or titanium — none of which are reliably magnetic — PT is the standard surface method instead.
Can MT or PT find internal weld defects? No. Both are surface methods only — they find cracks, porosity and similar flaws that break the surface, but nothing buried inside the weld. That requires RT or UT.
Is RT or UT better for finding weld defects? Neither is universally better — they have different blind spots. RT is generally stronger for volumetric flaws like porosity and gives a permanent image record; it’s weaker on tight planar cracks oriented parallel to the beam. UT (especially PAUT) is generally stronger for lack-of-fusion and tight cracks where angle-beam access is available. Codes often use both together.
What’s the advantage of PAUT over conventional UT? Electronic beam steering across multiple probe elements gives better multi-angle crack coverage, more accurate flaw sizing, and a full digital record — versus a conventional probe’s single fixed angle and hand-logged readings.
Do all weld codes use the same acceptance criteria? No. AWS D1.1 (structural steel), ASME B31.3 (process piping), ASME Section VIII (pressure vessels) and API 1104 (pipelines) each set their own limits for the same defect types — check the applicable code directly rather than assuming a limit from one code applies under another.
References
- Wikipedia, “Welding inspection” — pre-weld/in-process/post-weld staged framing.
- HII (shipbuilder), Welders Visual Inspection Handbook (2013) — VT defect types, inspection geometry, sequencing relative to other NDT.
- ProjectMaterials, “What Is Magnetic Particle Testing?”
- ProjectMaterials, “What Is Liquid Penetrant Testing?”
- ProjectMaterials, “What Is Radiographic Testing?”
- Open Exam Prep, ASNT-oriented study guide on RT limitations and safety (RT-vs-UT comparison for planar defects).
- ProjectMaterials, “What Is Phased Array Ultrasonic Testing?”
- AWS, D1.1/D1.1M Structural Welding Code — Steel, official scope description (aws.org).
- Inspenet, “ASME B31.3 Standard for Piping Systems.”
- MaterialWelding.com, “Acceptance Criteria for Welding Defects” (ASME Section VIII appendices summary).
- GlobalSpec, API 1104 standards listing.
- National Board of Boiler and Pressure Vessel Inspectors, ASME Section V 50th-anniversary retrospective (nationalboard.org) — confirms Section V’s intended role as the NDT-methods reference for other code sections including Section VIII.
Further reading
- Welding Symbols: Types, Meaning, Welding Codes — symbol notation referenced throughout weld documentation.
- Types of Weld Joints — joint geometry including fillet welds.
- Welding Procedure Specification and Welder Qualification Codes and Certification — the procedural and personnel side of weld quality this hub doesn’t cover.
- CWI Exam Preparation Guide — the personnel-certification path for welding inspectors applying these methods.
- Magnetic Particle and Penetrant Testing: Process Control and Acceptance — process-control detail for the MT/PT surface methods covered above.
- Electromagnetic Testing: ECT, ECA, RFT and ACFM — an alternative surface/near-surface method not covered in this guide.
- ASME Section V Article Pages — one of the acceptance-criteria codes referenced in the code-comparison table above.




There are no comments yet!