Weld Testing Methods: Complete NDT Guide for Structural Steel (2026)

What Is Weld Testing — and What Happens When a Weld Fails

Weld testing is the inspection of completed welds to ensure that they are up to code with respect to the required component structural strength and dimensions prior to the weld being subjected to actual service loads. Weld testing can be broken down into two branches, non destructive testing or NDT, which is further broken down into three subgroups, and destructive testing which destroys and/or exhausts the weld in order to physically measure the mechanical properties.

The cost of a missed weld defect is significant. A report by The Welding Institute (TWI) calculated an average repair cost across the oil and gas and power fabrication sectors of 1-3% with a peak of 25% at joints with restricted access (and isolated occurrences into the 50% region). On structural steel 80-90% of in-service failure is due to fatigue fracture (Research Gate, 2021) and most of this, again, was caused by defects identifiable at the inspection stage.

Looking at the shop floor, however, the numbers are clearer. Industry users on the AWS forum complain of a tolerable shop rejection ratio of 1-2%; the industry norm applied by most fabricators is when the rejected percentage gets to 5%, to take followups for all-Joint-penetration (CJP) welds from spot-UT to 100% UT until the unacceptable percentage drops again. One shop worked at 15% for fifteen years until it tackled the root-cause welder problem. As long as it was, any rework costs labour, gas, consumables, schedule delay: finances drain away.

How do you test a weld?

Begin with 100% visual inspection, which is a minimum requirement of all structural steel Codes, and costs nothing when applied properly. For full penetration groove welds AWS D1.1 states that volumetric inspection: UT, RT, or in some cases MT, is required. VT is for defining surface conditions and checking finish of welds, including profile and geometry. UT and RT are for defining internal conditions: porosity clusters, slag inclusions, lack of fusion which cannot be seen by the naked eye. Destructive tests (bend, tensile) are permitted in the case of trials as qualification of a WPS, but not actual production welds.

8 Weld Defects That Testing Is Designed to Catch

NDT techniques are not interchangeable; they are all specialized in detecting a certain sort of flaw at a certain location. Prior to identifying a suitable inspect method a fabricator must be aware of which types of discontinuity are expected for the particular process, material and joint shape, etcl! The following table details the eight types of fault encountered by the most structural steel Inspectors.

6 NDT Weld Testing Methods Compared (Visual, UT, PAUT, RT, PT, MT)

Different NDT methods use different physical principles that define their capability of detection, penetration depth and costs of operation. All six methods used in structural steel fabrication are compared below.

Is visual inspection of welds enough for structural steel?

For fillet welds under AWS D1.1 (statically loaded structures), simply using visual inpection alone will do – if every weld is within the dimensional and profile limits shown in table 9.1. But for complete joint penetration (CJP) groove welds, D1.1 mandates volumetric inspection – either UT, RT, or MT in specific situations. Why? Geometry. A planar lack-of-fusion or a tight cracking parallel to the weld line is a blind spot for the eye, but easily picked up by UT. An inspector in the AWS forum summarized it as: “100% visual inspection only becomes an 80% visual inspection because things can be missed.” VT is the gold standard, not the gold medal – for any CJP weld in structural steel.

Ultrasonic Testing (UT) — The Standard for Thick Structural Steel

Conventional UT sends high frequency sound pulses (namely 2-5 MHz) into the weld from an angled transducer on the plate surface, with internal flaws producing a return echo. The phase and magnitude of the reflected signal reveal flaw size and depth. AWS D1.1 Part F specifies UT of 5/16 (8mm) to 8 (200mm) inch thickness CJP groove welds, and UT can detect flaws that 0.5mm wide. Because it does not need radiation safety protocols, UT is preferred for thick plate. It is simply very operator skill-dependent – the quality relies on the coder’s ability to interrogate the correct beam angles and respond from the A-scan display.

Phased Array Ultrasonic Testing (PAUT) — Higher Resolution, Permanent Record

PAUT utilises a multi-element array (minimum 16 elements under AWS D1.1 Annex H) that electronically scans the ultrasonic beam at multiple angles simultaneously – no manual rastering in conventional UT required. The procedure creates a 3D weld cross-section image, a permanent, encoded record similar to an X-ray film, and the scan resolution is only 1-2mm (versus 5-10mm in conventional UT). AWS D1.1 added PAUT approval with the 2015 D1.5 Bridge Welding Code, and in the 2020 D1.1 Structural Welding Code -Annex H – for 3/16 to 8 inch material.

“Phased array is widely accepted in industry standards including AWS D1.1 and AWS D1.5 which govern the inspection of structural welds” said Haworth. Focusing techniques – FMC (Full Matrix Capture) and TFM (Total Focusing Method) – are now applied along with PAUT where conventional sectorial scans cannot detect complex or atypically located flaws, however FMC/TFM are yet to be incorporated into D1.1 acceptance standards.

Radiographic Testing (RT / X-ray) — Full Cross-Section Visibility

RT – Radiography analyzes the weld by passing a beam of ionizing radiation (X-ray, gamma-ray, etc.) through it and recording a shadow image on a film or digital detector. It reveals defect volumetric information very well – porosity, slag, and lack of penetration. An RT indication shows as any darker area in the film. RT provides a permanent record that can be used as a court-record, and is easily interpreted by non-UT personnel. RT’s limited sensitivity to tight planar defects (cracks running in the plane of the beam), the need for radiation safety procedures (exclusion zone, license), and slower setup time than UT are conmanding factors. Of comparative note – a US DOT study radiographed and ultrasonically tested the same structural welds and determined that UT is most economical for planar defect detection, but RT images are often better for volumetric weld quality improvement.

Liquid Penetrant Testing (PT) — Surface Cracks on Any Material

PT – Penetrant analysis applies a liquid penetrant to the weld surface and then develops the indication through a series of steps including application of a special developer.This method can identify any surface cracking material regardless of the base material – a standard in detecting sharp notches or cracks in stainless steel, aluminium, and other nonmagnetic alloys; it makes PT the de facto standard for austenitic welds that cannot be inspected magnetically. On carbon steel structural welds it was about equal in sensitivity to open surface cracks like MT and faster to perform on ferromagnetic material. Any subsurface indications that do not path to the surface leave an inconclusive indication – a hard limit. When welds are not visually clear, where weld profile limits VT interpretation, PT can be used as an addition.

Magnetic Particle Testing (MT) — Near-Surface Sensitivity on Carbon Steel

MT – Magnetizes the weld zone, then applies iron powder – dry or wet. Discontinuities near the surface along the weld zone disturb the magnetic flux; the particles collect and form a visible indication. MT detects near-surface cracks reliably in about 3 mm of material, effectively making it faster and more sensitive to near-surface cracking than PT. Conditions for use are: ferromagnetic material, and at minimal 3 mm depth of indication from the surface. Unlike PT, function is restricted to ferromagnetic alloys – structural carbon steels and low alloys. MT finds use on fillet welds and welds with restrictively little access to allow UT scanning.

What is the most commonly used method of weld inspection?

The ‘traditional’ volumetric method: ultrasonic testing and phased array ultrasonics are the most popular NDTs on structural steel welds, especially groove welds above 10mm plate thickness. The most-performed method (because of AWS D1.1) is visual testing – but the visual method will only fully evaluate the weld surface. For a complete joint penetration weld the volumetric inspection method will be UT. RT is also often used on piping and thin-section welds where a permanent record is required.

Destructive Weld Testing: When You Need It and What It Proves

The destructive method: destructively tests (crushes or cuts) the welds in order to measure the physical properties that define the welds – tensile test, bend test, fusion, etc. Its use is limited to testing the WPS (Welding Procedure Specification) qualification. Once the procedure has been defined through DT, all welds performed to that WPS are qualified to be inspected nondestructively.

The deliverable of successful DT sequence is a Procedure Qualification Record (PQR) – the proof that a certain set of welding variables results in welds that meet the code mechanical requirements. All production WPS have to have a valid PQR as reference. No qualified WPS, no documented proof, no assurance any NDT result is reproducible against a baseline.

Weld Testing Standards and Acceptance Criteria (AWS D1.1, ASME IX, ISO 3834)

Applicable weld testing code is not a matter of choice — it follows from the structure type, contract specification, and jurisdiction. Using the wrong code, or misapplying the right one, produces an inspection result that is not legally defensible. Four codes most relevant to structural steel fabrication are compared below.

How to Choose a Weld Testing Method: The 5-Factor Weld Test Selection Matrix

No single NDT method fits every structural steel application. Five factors determine the right selection: joint type, plate thickness, access availability, code requirements, and budget. Below, these factors map to method recommendations. Three scenario examples that follow show how the logic applies in practice.

Scenario 1 – H-beam fillet welds, static load, AWS D1.1 specification: If X = fillet weld + plate <20 mm + static load + D1.1 code + cost sensitive budget Use VT + MT. Visual inspection offers a full two dimensional surface scan of the weldment, while MT contributes enhanced near-surface flaw detection at the weld root, which is also the startup site for fatigue on fillet welds. No volumetric testing is specified by the specification for fillet welds in this scenario.

Scenario 2 – CJP groove weld, 40 mm plate, AWS D1.1: If X = CJP groove weld + 40mm plate + D1.1 code + moderate budget Use UT. When the plate thickness is 40 mm, standard UT offers the advantage of full through-thickness coverage and a higher speed of execution (no IR exclusion zone) over radiography, as well as compliance with D1.1 acceptance criteria for groove welds. PAUT becomes a good enhancement if record of scan coverage is a contractual requirement.

Scenario 3 – Box column, CJP, bridge application, high cycle: If X = box column CJP + >40 mm plate + D1.5 code + bridge (cyclically loaded) + permanent record required Use PAUT. D1.5, because of the tension welds, demands a higher level of inspection scrutiny; PAUT provides 3D record coverage, 1-2 mm lateral resolution and greater speed of execution than conventional UT and is specifically approved in D1.5 Annex H.

Building a Weld QC System for Structural Steel Fabricators

Weld testing does not begin after the weld is complete. Effective Quality Control systems identify defect states before the first arc ever hits the weld joint. Studies over the past forty years show that the only way to reasonably eliminate NDT rejection of complex fabrication is to affect inspection through bake-in quality control: a 30-50% reduction in NDT rejection is possible throughFit-Up inspection alone. The three-phase roadmap below visualizes what the systems check and when.

How Robotic Welding Systems Affect NDT Pass Rates

The connection between weld consistency and NDT pass rate is direct. Manual welding introduces arc-on time variability. A skilled welder averages around 25-35% arc-on time in a shift (regardless of arc idling, repositioning, retooling, rest periods, or fit up adjustments). Weld travel speed oscillates, interpass temperature management relies on individual judgment, and welder fatigue affects the quality of late-shift passes. Each of those factors is a defect introduction pathway that UT and RT field discovery seek to prevent.

Robotic welding removes those factors. Automated robotic systems reproduce identical voltage, amperage, travel speed, and torch incline from pass 1 through pass 12,040. Such control is clearly recordable with NDT data. At a bridge fabrication shop utilizing Zhouxiang’s intelligent steel structure welding system, sole-developed NDT pass rate on groove welds jumped from 82% to 97%. Arc-on duty cycles more than doubled, and the equivalent number of productive weld hours per shift increased with it.

For structural steel fabricators benchmarking the high-production NDT challenge of high-volume overhead gantries, H-beams, box girders, and portal frames, the ROI question is not simply speed. It is rework cost. At 1-2% NDT rejection, most shops make money. At 10%, rework cost outweighs the labor efficiency of a manually weld shop entirely. A robotic structural steel welding system tackles the consistency variable NDT measures to compensate for.

Related Zhouxiang resources: structural steel welding robot solutions | robotic vs manual weld quality comparison | how does robotic welding technology work | gantry welding robot workstation | how does a gantry welding robot work | cantilever welding robot for steel fabrication

Weld Testing Trends: What’s Changing Through 2030

The NDT and inspection market is poised for 8.3% CAGR growth, up from $14.99 billion in 2025 to a forecasted $22.34 billion by 2030 (MarketsandMarkets). Widespread growth in that US market will manifest in five patterns affecting structural steel weld inspection.

• PAUT replacing conventional UT on thick plate. Implementation of PAUT was driven by AWS D1.1:2020 codification (Annex H) [a]. The permanent encoded scan record, faster coverage, and 3D flaw imaging help make conventional UT progressively difficult to justify for CJP groove welds above 20 mm.
• Digital radiography (DR) replacing film RT. Film RT, slow and a source of negative chemical and metal waste, can be replaced with digital detectors delivering the same volumetric image in real time, with no film archiving cost and immediate electronic delivery of records to the design team.
• AI-assisted defect detection. Machine learning models trained on RT and PAUT scan libraries are deployed to mark likely indications for human review – reducing the risk of false-negative interpretation on high-volume inspection programs. Early results show detection consistent with experienced Level II inspectors on normal defect types.
• 100% traceability requirements expanding. Increasingly strict bridge codes and seismic zone structural specifications require digital weld records linking every joint to its inspection result, inspector ID, and welding parameters – a feature only robotic systems with integrated data logging can provide without manual data entry overhead.
• FMC/TFM as upgrades to PAUT. Full Matrix Capture and Total Focusing Method allow inspector-level flaw characterization beyond what standard PAUT sectorial scans can resolve. Although still not acceptable under AWS D1.1, they are used on complex weld geometries where sectorial scans are blind to off-axis or transverse flaws.

For fabricators deploying new NDT equipment and methodology today, the bottom line is clear: PAUT and digital radiography are future-proof in structural steel. Film RT and conventional UT still meet code, but efficiency, record quality, and future specification compliance will continue diverging each code cycle. For overview, see: welding robot maintenance and robotic welding ROI analysis in the larger automation investment.

FAQ — Weld Testing Questions from Steel Fabricators

How do you test a weld?

What is the most commonly used method of weld inspection?

What are the five standard NDT methods for welding?

What is the difference between destructive and non-destructive weld testing?

Which code governs weld testing for structural steel in the US?

Can visual inspection alone pass welds for structural steel projects?

How does robotic welding affect weld testing requirements?

What is a Welding Procedure Specification (WPS) and why does it matter?