NDT for Heavy Weldments: When UT, MT and PT Are Worth Specifying

“100% UT on all welds” is one of the most expensive sentences a buyer can put on a heavy frame drawing, and it is usually written by someone who is not certain what UT detects. Specifying NDT well means knowing what each method finds, which defects actually threaten your structure, and where those defects occur.

The four methods

Visual (VT)

Everything, every job. A qualified inspector examines every weld for undercut, underfill, overlap, crater cracks, profile, arc strikes and spatter, against the numerical acceptance criteria in the applicable code.

• Finds: surface-breaking and profile defects
• Cost: built into the job
• Do it: always, on 100% of welds

VT catches a surprising share of real problems. A shop that cannot produce a signed visual inspection record is not somewhere to send a load-bearing weldment.

Magnetic Particle (MT)

Magnetise the area, apply iron particles, and surface-breaking or near-surface discontinuities distort the field and gather particles into a visible indication.

• Finds: surface and very slightly subsurface cracks, lack of fusion at the surface, porosity breaking the surface
• Limits: ferromagnetic materials only, no stainless or aluminium; orientation matters, so the field has to be applied in two directions
• Cost: low
• Use it on: fillet welds in fatigue-loaded joints, weld toes, repairs, and anywhere a crack would start at the surface

MT is the best value on this list. On a heavy frame, most cracking that matters starts at a weld toe, and MT finds it cheaply.

Dye Penetrant (PT)

Apply a penetrating dye, dwell, remove excess, apply developer, and the dye bleeds out of surface-breaking defects.

• Finds: surface-breaking defects only
• Limits: truly surface-breaking only; messy; needs clean surfaces
• Cost: low
• Use it on: stainless and non-ferrous work where MT will not function

For carbon steel, MT is usually the better choice. PT is what you use when the material rules MT out.

Ultrasonic (UT)

A transducer injects sound into the material and reads reflections. Modern shops increasingly use phased array (PAUT), which steers the beam electronically and produces an imaged result.

• Finds: internal volumetric and planar defects, lack of fusion, incomplete penetration, internal cracks, slag
• Limits: needs access and a competent operator; geometry can complicate interpretation; results depend heavily on technician skill
• Cost: significant
• Use it on: full-penetration butt welds carrying tension or fatigue

UT is the method that finds the defects that fail structures. It is also the one whose results vary most with the person holding the probe, which is why the technician’s qualification matters as much as the method.

A word on radiography (RT)

Film or digital radiography gives a permanent image and is excellent on butt welds. It also requires an exclusion zone, licensed sources and a shop set up for it. On heavy structural work in most shops, UT has largely displaced it. Specify RT where your code or client requires it, not by default.

Where the defects actually are

This is how you build a rational scope. Defects concentrate:

• At weld toes on fillet welds in fatigue service → MT
• At the root of full-penetration joints, especially single-sided welds without backing → UT
• In heavy multi-pass welds: where interpass cleaning may have left slag → UT
• At repairs: which are statistically worse than original welds → MT, plus UT if the original joint was volumetrically tested
• At joint intersections and copes: where restraint is highest → MT
• In the heat-affected zone of thick, high-restraint joints where preheat was inadequate → UT, and check the preheat logs

Writing the scope

A good NDT specification is a marked-up drawing, not a sentence. Three elements:

1. Which joints. Identify them by weld number on the weld map.
2. Which method. Chosen from the logic above.
3. What extent. 100%, or a percentage with a defined escalation rule.

A workable default for a heavy structural frame:

Joint category	Method	Extent
All welds	VT	100%
Full-pen butt welds, tension or fatigue	UT	100%
Full-pen welds, compression only	UT	10%, escalating
Fillet welds in fatigue-loaded joints	MT	100% of toes
Fillet welds, static	VT	100%
All repair welds	MT + original method	100%

The escalation rule matters. Write something like: “If any tested joint is rejected, testing extends to 100% of that welder’s work on that joint category.” This makes partial sampling statistically meaningful instead of a lottery, and it gives the shop a strong incentive to get it right.

Acceptance criteria and who signs

Reference a code, AWS D1.1, or ISO 5817 quality level B/C/D, rather than describing defects yourself. The codes carry tables keyed to material thickness and loading condition, and their acceptance criteria differ between statically and cyclically loaded structures. Background in AWS D1.1 in plain language.

Specify technician qualification: ISO 9712 or ASNT SNT-TC-1A Level II as a minimum for the method being used, Level III for procedure approval. Ask for the certificates with the reports.

Timing

Two points worth stating on the drawing:

• NDT after post-weld heat treatment: not before. Stress relief can open up a defect that was tight when cold.
• Allow a delay before testing high-strength or thick joints: commonly 48 hours, so that hydrogen-induced cracking has time to appear. On thick, high-restraint joints this is the difference between finding a crack in the shop and finding it in service.

What you should receive

• Signed VT records referencing the weld map
• NDT reports per joint, with technician name, certification number and date
• The NDT procedure used
• Repair records, showing what was found, how it was excavated and the re-test result

That last one is the interesting document. Every shop has repairs. A shop that documents them openly is a shop that is telling you the truth about the rest.

The cost question

Volumetric NDT is a real line item and it is where blanket specifications waste money. The right scope is usually a small fraction of a blanket one, and finds more, because the effort is concentrated where defects live.

On the other hand: under-specifying NDT on a fatigue-loaded frame is a false economy of a different kind. The rule of thumb is to test where a defect would propagate, and skip where it would not.

Have a drawing with a blanket NDT call-out and no weld map to hang it on? Send it over, we will propose a scope within 24 business hours, and the matched shops price the difference.

See also: how much NDT a weld needs by execution class and NDT personnel certification under ISO 9712 and SNT-TC-1A.