Why Did My Weld Fail UT (or RT)? A Diagnostic Guide from an ASNT Level III with 30+ years experience

 

Why Did My Weld Fail UT (or RT)? A Diagnostic Guide from an ASNT Level III

You just got the call: "Your weld failed UT." Or worse, a radiograph landed on your desk with a red circle around a defect nobody wants to explain. Your first questions are always the same — what exactly failed, why did it happen, and what do I do now?

After 25+ years running NABL ISO/IEC 17025:2017 and NADCAP-accredited NDT operations at Trinity NDT, I can tell you: a failed UT or RT result is rarely a mystery once you know how to read it. It's a diagnostic clue, not a dead end. 

This guide walks through how to interpret a failure, trace it back to its root cause, and decide the right next step — repair, re-qualify, or dispute the call.

Step 1: Confirm What Actually Failed

Before anything else, separate three things that get conflated on the shop floor:

1. The weld failed — an indication was found and evaluated against the applicable acceptance standard (ISO 5817, ISO 10675, ASME Section VIII Div. 1, AWS D1.1, API 1104) and it exceeded the allowed limit.
2. The test failed — the NDT itself was inconclusive (poor image density, incorrect probe angle, insufficient coverage) and needs to be redone, not the weld repaired.
3. The welder failed qualification — this is a WPQ (Welder Performance Qualification) test, not a production weld, and a failure here means the welder — not necessarily the procedure — needs re-testing.

Ask for the actual report before assuming the worst. A UT report showing "amplitude exceeds reference level per ISO 11666 Level 2" is a completely different conversation than "signal lost, re-scan required." 

If you're unsure how your UT results are being graded, our guide to ISO 11666 acceptance levels for UT of welds breaks down how amplitude, length, and defect type combine into a pass/fail call.

Step 2: Match the Indication to a Defect Family

Most Ultrasonic testing(UT) and Radiography testing (RT) failures trace back to one of five defect families. Here's how each shows up and what usually causes it:

Lack of Fusion / Lack of Sidewall Fusion

• UT signature: Sharp, non-planar-looking reflector along the sidewall or between passes, often with poor amplitude response to angle changes.
• RT signature: Dark, linear line running parallel to the weld preparation, often at the fusion face.
• Root cause: Contamination, insufficient heat input, incorrect travel angle, or poor interpass cleaning — very often a welder technique issue rather than a procedure issue.

Slag Inclusions

• UT signature: Rounded or irregular reflectors, generally lower amplitude and more scattered than a crack.
• RT signature: Irregular dark patches, sometimes elongated along the weld axis.
• Root cause: Poor interpass slag removal, incorrect electrode manipulation, or joint geometry that traps slag in the root.

Porosity

• UT signature: Multiple small, low-amplitude reflectors — often dismissed individually but flagged when clustered.
• RT signature: Rounded, well-defined dark spots, sometimes in clusters or aligned ("wormhole" porosity).
• Root cause: Contaminated filler wire, moisture in flux/consumables, drafty site conditions disrupting shielding gas, or excessive arc length.

Cracks

• UT signature: Sharp, high-amplitude, planar reflector with strong angular sensitivity — this is the one that gets flagged and escalated immediately, and rightly so.
• RT signature: Fine, often jagged dark line, harder to see on film than UT makes it seem.
• Root cause: Hydrogen embrittlement, excessive restraint, incorrect preheat/interpass temperature, or rapid cooling — always treat as a metallurgical issue, not just a technique issue.

Root Undercut / Concavity

• UT signature: Geometric reflector near the root, easily confused with lack of fusion by an inexperienced operator — this is a common source of false calls.
• RT signature: Localized thinning visible as a lighter band along the root.
• Root cause: Excessive travel speed, high current, or incorrect torch angle at the root pass.

If you're seeing repeated calls in the same defect family across a project, the fix usually isn't "gouge and reweld one more time" — it's revisiting the WPS parameters or the welder's technique on that specific joint configuration.

Step 3: Ask the Questions That Actually Narrow It Down to facts

When a critical weld fails, work through this list before ordering a blanket repair:

• Is the defect isolated or repeating? One indication on one joint is a technique slip. The same indication type recurring across multiple welders or joints points to a procedure, material, or consumable problem. 
• Was the Welding Procedure Specification WPS actually followed? Pull the welding log — current, voltage, travel speed, preheat, inter pass temperature — and compare against the qualified WPS range. A shockingly large share of "test failures" are actually parameter drift.
• Was the test itself valid? For UT, check probe calibration records, reference block use, and scanning coverage. For RT, check film density, IQI sensitivity, and source-to-film distance. An invalid test isn't a weld failure — it's a retest.
• Who performed the test, and at what level? Under SNT-TC-1A / ISO 9712, evaluation and disposition should be signed off by a qualified Level II or reviewed by a Level III. If there's any ambiguity in interpretation, an independent Level III review resolves it faster than an argument on the shop floor.
• Is the acceptance criteria even the right one? We regularly see disputes evaporate once someone checks whether the applicable code is ASME Section VIII, AWS D1.1, or a project-specific spec — they are not interchangeable, and using the wrong one is a very common (and avoidable) failure.

Step 4: Decide the Right Corrective Path

Once you've isolated the cause, the response depends on what you found:

Root cause identified	Typical corrective action
One-off welder technique slip	Localized repair, re-test the repaired area only
Repeating defect across multiple welds by the same welder	Welder retraining and re-qualification (WPQ)
Defect pattern tied to a specific WPS parameter	Procedure requalification (PQR review), not just repair
Consumable or material issue	Batch investigation before any further welding continues
Invalid or inconclusive test	Retest with corrected technique/equipment — no weld repair needed

Repairing without understanding which of these five you're in is how the same defect quietly reappears three welds later.

When PAUT or TOFD Changes the Conversation

One thing worth knowing if you're still relying on film RT: a lot of the ambiguity in "was that really a crack or just geometry" disappears with Phased Array UT and TOFD, because you get real-time, storable imaging and far more precise through-wall sizing than a single UT A-scan or a radiograph can give you. 

It doesn't replace root-cause thinking, but it does dramatically cut down on inconclusive or disputed calls — which is exactly the "was it the test or the weld" problem this guide is about. We've written a detailed comparison in PAUT vs Conventional UT: Which Ultrasonic Testing Method Do You Need? if you're deciding whether to move a project over.

The Bottom Line

A failed UT or RT result is data, not a verdict. The fastest route back to production isn't grinding out every flagged joint — it's spending fifteen minutes tracing the indication to its defect family, checking whether the WPS was actually followed, and confirming the test itself was valid before you touch a grinder. 

Get that sequence right and you'll cut repeat failures dramatically, not just patch the one in front of you.

If you're getting inconsistent UT or RT calls on a project, or you want a second, independent Level III opinion on a disputed result, Trinity NDT's ultrasonic testing team and radiography inspection services are available for onsite and lab review across India. For welder-specific failures, our welding and welder qualification services team can help you determine whether you're looking at a technique issue or a procedure issue — and get the right corrective path moving fast. Reach out here and we'll typically respond within two hours.