NDE method selection for structural welds under AWS D1.1

Choosing the wrong NDE method doesn't just waste money — it creates false confidence. A radiograph can miss a tight planar crack oriented parallel to the beam. A yoke MT examination won't detect a subsurface porosity cluster at 1.5 in depth. Each method has a physics-defined sensitivity envelope, and that envelope must match the flaw types most likely to occur given your weld process, joint geometry, and base metal.

AWS D1.1:2025 Clause 8 governs testing and inspection of structural steel welds. It specifies visual testing (VT) as the baseline requirement for all welds and allows or requires radiographic (RT) or ultrasonic (UT) testing for complete joint penetration (CJP) groove welds in the categories specified by contract or by code.

The four methods in structural welding practice

Radiographic testing (RT) passes X-ray or gamma-ray radiation through the weld and records the transmitted pattern on film or a digital detector panel. It excels at volumetric discontinuities: porosity clusters, slag inclusions, incomplete fusion with a measurable gap. Its weak point is orientation: a tight planar crack parallel to the beam produces minimal contrast on the image and can be missed. RT also creates a permanent image record, which is valuable for documentation-heavy projects and owner archives.

Ultrasonic testing (UT) sends high-frequency sound pulses into the weld and measures reflections from internal interfaces. A trained UT technician can locate, size, and characterize flaws in three dimensions. UT is the standard method for thick plate (above approximately 3/4 in) where radiographic sensitivity degrades and for CJP welds in primary structural members where planar-flaw sensitivity matters. Phased-array UT (PAUT) and time-of-flight diffraction (TOFD) offer higher resolution and faster scanning speeds; both are accepted under D1.1:2025 when the procedure is qualified per Clause 8.

Magnetic particle testing (MT) magnetizes the part and applies ferromagnetic particles — dry powder or wet fluorescent suspension — that bridge surface and near-surface discontinuities to form visible indications. MT is fast, inexpensive, and highly sensitive to fatigue cracks, surface-breaking lack of fusion, and linear surface indications on carbon steel and low-alloy steel. It does not work on austenitic stainless or aluminum — the material must be ferromagnetic. MT comes in two field configurations: dry-powder with prods or yokes (common for field work and large structural members), and wet-fluorescent (more sensitive, better for complex geometries in shop inspection).

Liquid penetrant testing (PT) uses capillary action to draw a dye — red visible or fluorescent — into surface-breaking discontinuities, then develops the indication with a white developer. PT works on any material, including non-ferromagnetic alloys. Its limitation is that it only detects surface-open defects. On carbon steel, MT is almost always preferred over PT because MT adds near-surface volumetric coverage that PT cannot provide.

Method selection by joint type and application

CJP groove welds in primary structural members — RT or UT is the default. The choice comes down to plate thickness, access, and project preference. Above approximately 2.5 in thickness, UT is more practical and more sensitive than film RT. For thin plate under about 3/8 in, RT film interpretation is easier and cheaper. Digital radiography (computed radiography and direct digital radiography) is closing that gap on setup time, but the orientation-sensitivity limitation remains.

CJP groove welds in secondary members or minor attachments — VT is the code minimum. MT or PT may be added at the engineer's discretion or in response to elevated repair rates on the joint type.

Fillet welds — VT baseline. MT is added for fatigue-critical welds, cyclic-load applications, or when the quality history for that joint configuration shows surface-cracking issues. Fillet welds are rarely radiographed — the geometry creates excessive scatter that makes film interpretation unreliable.

Repair welds — RT or UT on the repair volume, plus MT at the repair toes to check for surface cracking introduced by grinding. Hold the repair weld to at least the same NDE standard as the original.

T-K-Y tubular connections — UT is typically preferred over RT because the geometry makes radiographic setup difficult and beam orientation hard to optimize. MT at the weld toes is standard for fatigue-rated tube connections.

How D1.1:2025 structures NDE requirements

Clause 8 establishes two layers of NDE obligation:

1. Minimum code requirement — VT on all welds; RT or UT on CJP groove welds in the categories specified by Clause 8 (which references structural loading category: statically loaded vs. cyclically loaded, and connection type).
2. Contract requirement — the EOR can expand scope to 100% RT on all CJP welds in primary members, 10% MT sampling on fillet welds, or any other requirement. The contract is additive; it cannot reduce below the Clause 8 minimum.

Acceptance criteria differ by method. RT uses a tabulated approach: elongated indications, piping porosity, and slag are each measured and compared against Clause 8 tables. UT uses a reference-level decibel scheme — indications above a rejection level get sized and reported. MT and PT criteria are based on indication length and type (linear vs. rounded).

Defect-type alignment

The match between method and defect type matters as much as any other selection criterion:

Hydrogen-induced cracking (HIC) deserves special attention. HIC cracks are typically tight, planar, and oriented along the HAZ — the worst case for RT beam orientation. For hydrogen-crack-susceptible joints (high-strength base metals, high restraint, cold temperature), UT plus MT at the weld toes provides better coverage than RT alone.

Scheduling NDE in production

NDE creates hold points. A CJP weld can't be painted, back-filled, or enclosed until NDE is complete and accepted. The scheduling impact:

• RT film — 4–8 hours from shot to interpretation in-shop with a film darkroom; digital panels can cut this to 1–2 hours. Overnighting to a contracted lab adds a day.
• UT — same-shift for most joints; thick plate on large structural members takes longer. PAUT scanning is typically faster than conventional UT on long weld lengths.
• MT — 30–90 minutes per joint depending on size: magnetize, apply particles, interpret, demagnetize, document.

Integrating NDE hold points into your weld map prevents assemblies from being enclosed before inspection. For building the documentation package, see NDE documentation and audit packet assembly. For structuring hold points into the inspection sequence, see weld inspection hold points for CWIs and the CWI WPS review checklist.

What belongs on the WPS vs. the inspection plan

The WPS identifies joint classification (CJP or PJP) and process parameters. It does not specify the NDE method — that belongs in the inspection and testing plan (ITP) or the contract documents. However, CJP vs. PJP classification on the WPS is the trigger for which NDE scope applies, so that field must be accurate.

Some WPS forms include a "NDE required" checkbox or a field for NDE type. If yours does, complete it accurately. A WPS that says "VT only" on a joint that the contract requires to be RT'd creates a documentation deficiency that will surface in any third-party audit. The WPS is the paper record of intent; it should reflect actual project requirements.

Ready to generate audit-ready WPS documentation with accurate joint classification? See plan options at the pricing page.

Rule library based on AWS D1.1:2025; verify against your governing edition.