UT Test for Welding: The Complete Guide to Ultrasonic Weld Inspection

A structural steel fabricator signs off on a 2-inch CJP groove weld, the project moves forward, and months later a crack traces back to a lack-of-fusion defect that was never seen. Ultrasonic testing exists for exactly that moment: it listens inside the weld with sound waves so a hidden planar flaw is caught before the structure ever carries a load.

What Is a UT Test for Welding, and Why Does It Matter?

A structural steel fabricator signs off on a 2-inch complete-joint-penetration (CJP) groove weld, the erection crew moves on, and months later a fatigue crack traces back to a lack-of-fusion planar defect that no visual check ever saw. That is the exact moment ultrasonic testing exists for. UT lets an inspector listen inside solid steel with sound waves and catch a hidden flaw before the structure ever carries a real load.

So what is a UT test for welding, in plain language? It is a nondestructive, volumetric examination method that uses high-frequency sound waves to detect internal discontinuities in welds. A transducer sends a short pulse of sound into the material; if the sound hits an internal break such as a crack, a slag line, or an unfused sidewall, part of the energy bounces back early. The instrument shows that echo on a screen, and the technician reads its position and height to estimate where the flaw sits and how big it is. Nothing is cut, no radiation is used, and the weld stays exactly as it was.

Why “volumetric” is the important word

Surface methods such as dye penetrant and magnetic particle only look at the skin of the weld. UT interrogates the full cross-section, root to cap, across the thickness. That is why codes treat it as equivalent to radiography for verifying soundness. In fact, AWS D1.1, Clause 2 and Section 6 requires that the soundness of CJP groove welds subject to tension and reversal of stress be established by either radiographic or ultrasonic testing. A visual walk-through cannot satisfy that requirement.

One practical advantage matters on site: unlike radiography, UT requires access to only one surface of the weld in most plate and pipe applications. There is no film to position on the far side, no darkroom, and no radiation work exclusion zone that shuts down the surrounding area while the exposure runs.

Where UT is required or commonly specified

• Structural steel: CJP groove welds in tension and cyclic service, per AWS D1.1, including the detailed UT procedures in AWS D1.1 Clause 6, Part F.
• Pressure vessels and power piping: ASME Boiler and Pressure Vessel Code Section V, Article 4 is the governing method reference for ultrasonic examination of welds in code-stamped construction, while Section VIII and Section III supply the acceptance rules.
• Bridges: AWS D1.5 Annex J sets out an advanced UT approach for bridge welds, and state agencies such as FDOT have studied the method extensively.
• Nuclear components: the NRC has evaluated UT in lieu of radiography for reactor internals and piping.
• Pressure vessels, tanks, marine hulls, and heavy fabrication of the kind covered by resources like Alamo Welding & Boiler Works: Services & Boiler Expertise.

Where UT sits among the five main NDT methods

Method	Type	What it finds
Visual (VT)	Surface	Profile, undercut, visible cracks, overlap
Penetrant (PT)	Surface	Open-to-surface cracks and porosity
Magnetic particle (MT)	Surface + near-surface	Linear indications in ferromagnetic material
Radiographic (RT)	Volumetric	Internal volumetric flaws, projected to film
Ultrasonic (UT)	Volumetric	Internal planar and volumetric flaws through thickness

PT and MT excel at surface-breaking defects; RT shows a planar projection and is strong on porosity and slag; UT is strongest on planar, oriented flaws like lack of fusion and cracks, which is precisely the failure mode that produces the scenario above. No single method covers everything, which is why most specifications stack VT with one volumetric method.

What this guide will cover

The rest of this article walks through the physics behind the beep, the equipment and calibration setup, the code landscape for acceptance criteria, personnel qualification levels, the discontinuities UT actually finds, a fair UT-versus-RT comparison, newer techniques like phased array and TOFD, common troubleshooting mistakes, and a practical preparation checklist you can use on your next weld. If you are earlier in the journey and still building core skills, a structured Aluminium Welding Course: What You Learn, Costs & Careers overview shows how inspection knowledge fits into a broader welding career path.

By the end, you should be able to read a UT callout on a drawing and know exactly what procedure, calibration, and acceptance criteria the inspector is working from.

How Ultrasonic Testing Works: The Physics Behind the Beep

Everything about ultrasonic weld inspection comes down to one trick: make a sound, then listen for the echo. Tap a buried water pipe with a wrench and you can hear where the water stops. UT does the same thing inside a weld, using sound no human ear can detect and timing echoes to fractions of a microsecond.

The Pulse-Echo Cycle

At the heart of every probe sits a thin piezoelectric crystal, usually lead zirconate titanate, bonded inside the housing. Apply a short electrical pulse and the crystal changes shape, snapping against its backing to drive mechanical vibration into the metal — the inverse piezoelectric effect. Returning sound squeezes the same crystal and produces an electrical signal again. One element transmits, then listens, hundreds of times per second.

Almost all weld UT is pulse-echo: one search unit both sends and receives. The instrument measures the time between the outgoing pulse and the returning echo. Because sound travels through steel at a fixed speed — about 0.129 in/µs for shear waves and 0.233 in/µs for longitudinal waves — time converts directly into distance. That relationship is the foundation of AWS D1.1’s ultrasonic examination provisions.

Wave Types and Angle-Beam Wedges

A straight-beam probe sends longitudinal waves — particle motion parallel to the direction of travel — straight down. Handy for laminations, gross porosity, and thickness, but a poor geometry for a vertical groove weld sitting mid-plate.

The angle beam fixes that. The transducer is bonded to a plastic wedge cut at a fixed angle, and the sound refracts, or bends, as it changes speed entering the steel, following Snell’s law. Just past the first critical angle the longitudinal component dies out, leaving a clean shear wave. Technicians choose wedges that refract the beam to commonly 45°, 60°, or 70°, picking whichever angle sweeps the whole weld volume from the surface, often skipping off the back wall to reach the far side of the joint.

Couplant: Why Air Is the Enemy

Sound at ultrasonic frequencies cannot cross even a thin film of air; at a steel-to-air boundary nearly all the energy reflects back, so a dry probe reads nothing at all. Couplant — gel, oil, or water — is mandatory for contact UT because it displaces the air and gives the sound a continuous path into the part. Surface prep is the other half: scanning surfaces must be ground free of scale, spatter, and sharp irregularities. The rough surface left by carbon arc gouging scatters sound and eats penetration, so repair areas get a smooth grind before anyone raises a probe. It mirrors the rule for prepping aluminum before welding: nothing transfers through a dirty, uneven interface.

Reading the A-Scan

What the operator watches is the A-scan: time along the horizontal axis, echo amplitude as vertical spikes. The initial pulse sits at the left, the backwall echo at the right, and any spike between them is a reflector somewhere in the sound path.

Depth comes from geometry. After subtracting the wedge delay measured on a calibration block, the operator multiplies the remaining sound path by the cosine of the beam angle to get vertical depth. Amplitude needs correction, because sound weakens with distance through beam spread and attenuation, so codes require a distance-amplitude correction (DAC) curve, or its timed equivalent TCG, plotted from side-drilled holes at varying depths in a reference block. Indications are then judged against the curve at their own depth.

Sizing, though, is estimation rather than measurement. The 6-dB drop method marks where an echo falls to half height as the probe moves, and peak amplitude is compared with reference reflectors. A tight planar crack square to the beam returns a big signal; a rounded pore returns a small one even when it is physically larger. The instrument reports reflector response, not true flaw size — a distinction that matters the moment acceptance criteria apply.

Why 1–5 MHz Rules

Frequency is the master trade-off. Higher frequency means a shorter wavelength, finer resolution, and smaller detectable discontinuities — but also more grain-boundary scattering and less penetration. Lower frequency punches through heavy, coarse material at the cost of detail.

Field weld work usually sits between 1 and 5 MHz. The U.S. Army Corps of Engineers’ UFGS 05 05 23.13 10 requires instruments able to generate, receive, and present pulses across the 1 to 10 MHz range, leaving technicians room to adapt. A 2.25 MHz probe is the workhorse for thick sections and coarse grains; 4–5 MHz handles thin plate where resolution matters. The physics points forward: phased array probes pack 16 to 128 individually pulsed elements to steer the beam electronically, a development FHWA research has been evaluating for bridge welds — a story covered in the advanced methods section.

UT Equipment and Setup: Instruments, Probes, and Calibration

Knowing the physics of ultrasound is only half the job; the other half is a setup that produces trustworthy readings from the first sweep. Most UT results fail on a dry coupling surface, a worn cable, a block that does not match the code, or sensitivity set at the wrong distance, not on exotic theory. What follows covers the hardware, the blocks and curves that define sensitivity, and the surface prep behind a dependable ut test for welding.

The Core Kit: What Sits in the Case

A conventional UT kit is compact. The main components are:

• Flaw detector. A portable pulse-echo unit that pulses the transducer and displays returning echoes on an A-scan screen. Modern units add digital filtering, onboard DAC/TCG curves, data logging, and flaw-depth readouts.
• Search unit. The transducer combined with a plastic wedge that refracts the beam into the plate at a fixed angle, most often 60°, 70°, or 80° for shear-wave weld scanning. Straight-beam probes handle thickness gauging, calibration checks, and laminations.
• Cables and connectors. Coaxial cables take a beating on job sites, and an intermittent connection mimics a false indication, so spares are standard kit.
• Couplant. Gel, water, or water-based solution bridging the air gap between probe and part. Air reflects nearly all of the sound energy, so without couplant almost nothing enters the material.
• Calibration blocks and reference standards. V1/V2 radius blocks, IIW-type blocks, and material-matched standards with side-drilled holes (SDH) at known depths.
• Accessories. Rulers and markers for scan plans, spare wedges, emery paper, and a spray bottle.

All of it sits under a written procedure. AWS D1.1 Clause 6, Part F requires UT to follow a written procedure specifying joint configuration, technique, equipment, personnel, and calibration, and improvising the setup is a code violation even if the readings look perfect. Gear buying deserves the same discipline; this Airgas Welding Equipment Guide shows what to compare.

Calibration Blocks: V1/V2, IIW, and Side-Drilled Holes

Calibration blocks give echoes at known locations. The V1/V2 block uses its 100 mm radius to produce first and second back-wall echoes, which set range and sensitivity quickly. IIW-type blocks add notches and holes for checking beam exit point, refraction angle, and resolution. Reference blocks machined from the same material as the test piece, with side-drilled holes at several depths, show how sound attenuates in that alloy and heat condition.

The block must match the job. NRC Inspection Procedure 57080 requires verification that block type, size, geometry, material, and reflector locations agree with the applicable code and procedure before any data is taken. A block of the wrong thickness or a different alloy quietly shifts every depth and amplitude reading that follows, a point regulatory guidance on ultrasonic examination repeats across reactor vessel and piping inspections.

DAC/TCG Curves and Corrections That Keep Sizing Honest

Sound loses energy as it travels, so the same-sized flaw reads smaller at 4 inches deep than at 1 inch. Examiners plot a distance-amplitude correction (DAC) curve from side-drilled holes at increasing depth. A time-corrected gain (TCG) setup does the same job electronically, boosting gain at longer times of flight so every equivalent reflector produces the same screen height. Either way, flaw size is judged against the curve, not raw screen height.

Two further corrections matter in practice. Transfer (or sensitivity) correction compensates for energy lost on the rough scanning surface compared with the smooth calibration block. Horizontal linearity checks confirm the time base is accurate across the full screen so depth readings do not drift, and resolution checks confirm the probe can separate two closely spaced echoes.

Preparing the Scanning Surface and Planning the Scan

Sound enters through the surface, so coupling surfaces must be clean and reasonably smooth: free of loose scale, rust, paint, and dirt, with any irregularity that could mask a defect indication removed, as required by UFGS 05 05 23.13 10. Temporary weld backing strips must be removed and their sharp edges ground smooth before examination, as must crown reinforcement where the procedure requires it. Anyone prepping aluminum joints knows the habit from aluminum surface prep guides, where residue ruins both coupling and fusion.

Once the surface is ready, the examiner builds a scan plan: which side of the weld, which probe angle, what index point, what overlap between passes, and what maximum scan rate. Coverage must reach the full weld volume, including the far side of the joint and the root, usually meaning mirrored passes from both sides or a second wedge angle. Geometry itself echoes: the weld root, a corner trap, a counterbore, or a backing strip produces structural signals that are not flaws. The scan plan marks them so they are recognized, not chased.

Codes, Standards, and Acceptance Criteria for Weld UT

Ultrasonic testing without a governing document is only a beep and a guess. Before the probe ever touches the steel, someone has to answer two questions in writing: which examination standard applies, and which acceptance table decides pass or fail. Weld UT sits inside a layered web of standards, and confusing the document that describes the method with the document that judges the result is one of the fastest ways to produce a report nobody can act on.

Method Codes vs. Referencing Codes

The code landscape splits cleanly in two. Method codes tell you how to examine: transducer frequencies, calibration blocks, scan plans, sensitivity setting, evaluation of indications, documentation. ASME BPVC Section V, Article 4 works exactly this way for welded joints, and ISO 17640 does the same internationally.

What method codes almost never do is tell you what is acceptable. That duty falls to the referencing code: Section VIII for pressure vessels, Section III for nuclear components, AWS D1.1 for structural steel, AWS D1.5 or state DOT specifications for bridges, or the contract itself. A shop can perform a textbook Section V Article 4 examination and still not know whether the weld passes until it opens the acceptance rules in the referencing code. Extent of examination, which welds get examined, and the rejectable threshold all live there.

AWS D1.1 Clause 6 Part F: Structural Steel

For structural steel, AWS D1.1/D1.1M places ultrasonic requirements in Clause 6, Part F: written-procedure requirements, operator qualification, equipment checks, scanning coverage, and evaluation of indications for complete-joint-penetration groove welds. The clause carries a dedicated Ultrasonic Acceptance-Rejection Criteria table for statically loaded nontubular connections, shown as Table 6.2 in earlier editions of the code.

That table runs on amplitude-plus-length logic. An indication becomes rejectable only when it pushes a signal above the evaluation threshold and runs longer than the allowance granted for that amplitude band, and the allowable length grows as the echo gets weaker. Cyclically loaded and dynamically loaded joints are judged on different, tighter terms, which matters on anything carrying fatigue.

AWS D1.5 Annex J: Bridge Welds

Bridge work follows AWS D1.5, where transportation research on ultrasonic examination of bridge welds has pushed practice toward Annex J, a more advanced UT annex with its own technique requirements and acceptance provisions, including height-based evaluation of crack-like indications on fatigue-critical details rather than echo amplitude alone. FHWA guidance and state DOT special provisions increasingly point to it.

ISO 17640 and ISO 13588: The International Picture

Globally, ISO 17640 sets out ultrasonic testing of fusion-welded joints: techniques, testing levels, and the acceptance framework, sizing indications by measured height and length rather than raw amplitude. ISO 13588:2019 covers phased array UT for fusion-welded joints in metallic materials of at least 6 mm minimum thickness, and a 2024 review confirmed the edition remains current. Critically, ISO 13588 explicitly does not include acceptance levels. It is a method standard, full stop.

The same separation shows up in defense work. DoD UFGS 05 05 23.13 10 requires engineers to establish acceptance criteria before approving advanced techniques such as phased array or TOFD in lieu of conventional UT. Technique approval and pass/fail authority are two different decisions.

Codes also branch by industry and material. Aerospace welding stacks program-specific acceptance on top of base standards, while corrosion-resistant work such as Alloy 20 welding triggers supplementary procedure and testing requirements that plain carbon steel never does.

How the Acceptance Logics Compare

Code / Standard	Primary scope	Acceptance basis
AWS D1.1 Clause 6 Part F	Structural steel CJP groove welds	Amplitude plus maximum indication length for static nontubular connections
AWS D1.5 Annex J	Bridge welds	Length-based, with height evaluation for advanced UT details
ASME Section V Article 4	Vessel, boiler, piping, nuclear welds	Method only; acceptance sits in Section VIII, Section III, or B31
ISO 17640	Fusion welds, international work	Indication height and length against the chosen testing level
ISO 13588:2019	Phased array, 6 mm and above	Method only; no acceptance levels

Put the two documents side by side and UT acceptance stops being mysterious. The method code tells you how to find the indication; the referencing code tells you whether it matters. Checking that pairing at the start of a project means you will never be hunting for the right acceptance table at the end of a shift.

Who Can Perform a UT Test: Personnel Qualification and Certification

Here is the uncomfortable truth about ultrasonic weld inspection: the instrument does not decide anything. It draws a waveform on a screen, and a human being decides whether that spike is a rejectable lack-of-fusion flaw or a harmless geometric echo from the weld root. Two technicians can stand at the same weld with the same calibration and reach opposite conclusions. That is why the most important piece of UT equipment is not the flaw detector or the probe. It is the qualified, certified person holding it, working to a written procedure.

Because the whole quality decision rests on human interpretation, every serious welding code does the same thing: it controls who is allowed to look at that screen. In the United States, the framework most industry follows comes from ASNT SNT-TC-1A, the American Society for Nondestructive Testing’s recommended practice for personnel qualification and certification.

The Level I, Level II, and Level III Hierarchy

SNT-TC-1A defines a three-level structure you will find in shipyards, fabrication shops, refineries, and nuclear plants across U.S. industry. The levels are not seniority titles or pay grades. They are strict boundaries on what a person is authorized to do with the data.

Level	Authorized to do	Not authorized to do
Level I	Set up equipment, calibrate the instrument, perform scans exactly as instructed by a written procedure, mark and record the location of indications, maintain records	Interpret results, accept or reject a weld, modify the scanning technique or procedure
Level II	Interpret and evaluate indications, accept or reject against the code’s acceptance criteria, calibrate, choose the technique for a specific job, train and directly supervise Level I	Write or approve procedures, establish the technique for a new application, serve as the technical authority
Level III	Write and approve procedures, establish techniques and acceptance criteria, train Level I and Level II, audit the program, resolve code questions	Bypass the referencing code or the written procedure on a specific job

The practical consequence is simple: a Level I technician may put a probe on your 2-inch CJP groove weld, calibrate the box, and walk the scan path all day. The moment an indication appears, the Level I must call a Level II. There is also no such thing as a “general” UT certification. Every certificate is method-specific and often product-specific, so an MT or PT certificate confers nothing on a UT job.

Training and Experience Requirements

SNT-TC-1A is deliberately an employer-driven document, so its hour tables are recommendations rather than mandates. The typical starting point is roughly 40 hours of formal classroom training per level, layered on documented on-the-job experience: usually around six months at Level I before testing for Level II, and two years of Level II-level work before testing for Level III (with a longer, degree-dependent route for those without an engineering background). Employers routinely demand more than the minimum, and a welder who already understands codes and joint geometry from trade training, say through a hands-on aluminium welding course that covered inspection basics, usually picks up UT interpretation far faster than someone starting cold.

AWS D1.1 is considerably more specific. For structural steel UT, AWS D1.1 Clause 6 requires the operator to complete classroom instruction plus eight months of on-the-job training and experience under the direct supervision of a certified Level II or Level III before sitting the qualification exam. Federal and military specifications that invoke D1.1, such as the Unified Facilities Guide Specifications, inherit those same personnel requirements.

Employer-Based Certification vs. Third-Party Schemes

Here is a point that surprises newcomers: an “ASNT certified” technician does not hold a professional license the way a professional engineer does. Under SNT-TC-1A and its American National Standard counterpart ASNT CP-189, certification is granted and revoked by the employer, and it stays with that employer. If a certified Level II moves to another company, the new employer must review the prior documentation and decide whether to grant credit, typically through a transfer examination or an audited equivalence review.

The international alternative is ISO 9712, structured as a third-party scheme administered by accredited, independent certification bodies rather than by the individual employer. Whichever system applies, most programs tie the credential to a renewal cycle, commonly every five years, with documented recent work in that method or a re-examination as the price of keeping it current.

Why Codes Demand a Practical Examination

A written exam tests whether someone understands ultrasonic theory, equipment, and code clauses. It proves nothing about whether they can hold a probe steady on a greasy plate at 2 a.m. and spot a 3/16-inch notched reflector hidden under a weld crown. That is why AWS D1.1 requires UT operator qualification to include a practical examination: the operator must detect and evaluate discontinuities on test specimens with known reference reflectors, and the results are scored against those known answers. If you cannot find the seeded flaws the code says you must find, you do not operate on production welds. Regulators apply the same logic: NRC Inspection Procedure 57080 verifies that UT examinations are conducted by properly qualified personnel working to qualified procedures approved under the licensee’s quality assurance program.

For welders and inspectors weighing career moves, this is why UT certification is one of the better investments in the trade. It is the bridge between standing at a booth and working in zero-tolerance sectors like aerospace welding, nuclear, or pressure vessel fabrication, where the pay reflects the responsibility behind every signature. The qualification is demanding by design, because the weld you do not inspect properly is the weld that fails later, when it is too late to do anything about it.

What a UT Test Finds: Common Weld Discontinuities and How They Read

Ask ten welders what ultrasonic testing is for and you will get one answer: cracks. The truth is broader. UT detects internal defects including cracks, porosity, slag inclusions, lack of fusion, lack of penetration, and voids within welded joints. What separates them is not whether the sound can reach them, but how each one bounces the sound back. Understanding that difference is what turns a row of blips on an A-scan into a real judgement call.

Planar vs. Volumetric: The Acoustic Split

Every discontinuity UT encounters falls into one of two acoustic families:

• Planar flaws — flat, oriented features such as cracks, lack of fusion, lack of penetration, and toe cracks. They act like mirrors. When the sound beam hits them square, a large burst of energy returns, giving a sharp, tall signal that is easy to detect and, with practice, reasonably easy to size in height.
• Volumetric flaws — rounded or scattered features such as porosity, slag inclusions, and voids. They scatter sound in many directions rather than reflecting it cleanly, so the returned echo is lower and can look deceptively similar to a small but acceptable indication.

This split is why UT is famous for finding fusion problems that radiography routinely misses, and why fine porosity is the one thing UT handles less gracefully. An NRC/PNNL study found phased array UT much more capable than standard radiography of detecting planar flaws in carbon steel welds, while noting that PAUT detects all but very small volumetric flaws in carbon steel (NRC/PNNL comparative study).

The Main Discontinuities, One by One

Discontinuity	Type	How it reads
Lack of fusion / incomplete sidewall fusion	Planar	Strong, sharp echo from the fusion plane; orientation makes it UT’s specialty
Lack of penetration	Planar	Signal from the root at expected depth; often caught with a low-angle or root-scan pass
Cracks (linear, toe, root)	Planar	Sharp, often spiky, highly directional echo that can shift as the probe moves
Slag inclusions	Mixed	Irregular, ragged echo; amplitude varies with probe angle
Porosity / gas pockets	Volumetric	Clustered low-to-medium blips; individual fine pores may not register
Undercut	Surface-breaking	Read at the weld toe; usually evaluated visually or by MT/PT as well
Root toe cracks	Planar	Tricky geometry-affected signal near the root; corner-trap amplification can mask it

Orientation is everything. A flaw that lies perpendicular to the beam returns energy beautifully; the same flaw lying parallel to it can vanish. That is why scan plans call for multiple probe positions and why a single pass from one direction is never considered complete coverage. The beam must be swung at the flaw, not along it.

Why Sizing Depends on Which Method You Trust

Detection and sizing are separate skills. Radiography reliably gives two-dimensional flaw size — length and width — but cannot reliably determine the third dimension, flaw height or depth in the weld (review of radiographic vs. ultrasonic sizing). UT flips that: with calibrated distance-amplitude curves and a proper 6 dB drop or endpoint technique, an inspector can estimate how far a crack runs through the thickness, which is exactly the dimension that governs fracture mechanics and remaining life.

Volumetric flaws complicate the picture. A cluster of fine porosity may raise a scatter of low signals that still passes the acceptance table, while a single elongated slag line may not. In aluminum welds, where gas porosity is a recurring concern, this matters more than most people expect, and the practical prevention happens at the torch rather than the inspection station — as covered in Aluminum MIG Welding: The Complete Practical Guide.

How the Inspector Decides

The instrument never makes the accept/reject call. The examiner measures amplitude against the DAC curve, measures length along the scan, checks height where the code permits it, then compares those numbers to the acceptance criteria in the referencing code — for structural work, the tables in AWS D1.1 Clause 6 Part F. Geometry echoes from the root, backing strip, and corner trap are separated from real flaws by moving the probe; if a signal does not move predictably, it is not yet an indication worth reporting.

The practical takeaway: UT is strongest where failure is most dangerous — planar, crack-like flaws oriented where a crack would actually grow. It is weakest at tiny rounded voids, and honest reports say so. A well-run examination reports what it found, states the limits it worked within, and leaves the disposition to the code.

UT vs Radiographic Testing: Which Weld Inspection Method Wins?

Ask a room of inspectors which volumetric method is better and you will start an argument that lasts all shift. Both are workhorses, and both live in the same standards ecosystem: AWS D1.1 carries both methods, and ISO 17640 sits alongside the ISO radiography standards for weld examination. But they do not see flaws the same way, they fail differently, and they cost differently. The honest answer to “which wins” is: whichever one reliably detects the specific discontinuity your welding procedure is most likely to produce.

Factor	Conventional UT	Radiographic Testing
Portability	Instrument, probe, wedge and couplant fit in a small case	Source or X-ray unit, panels/film, cabling, barricades
Access needed	One surface only	Both sides (source behind, film or detector in front)
Safety	Non-hazardous; no work exclusion zone	Radiation hazard, exclusion zones, safety officer
View produced	Depth-resolved A-scan along the beam path	Plan-view shadow image of the weld volume
Best flaw type	Planar: lack of fusion, cracks, lack of penetration	Volumetric: porosity, slag inclusions
Depth of flaw	Measured directly from time-of-flight	Cannot be determined through thickness
Operator effect	High; interpretation is experience-dependent	Lower; film or image can be reviewed by anyone
Record quality	Conventionally transient unless captured on screen	Permanent, reproducible film or digital image
Material limits	Austenitic and coarse-grained metals scatter sound	Dense/thick sections need higher energy sources

Portability, Access, and Jobsite Disruption

This is where UT wins on almost every job that is not in a darkened bay. The Florida DOT research summary of conventional UT versus RT states it plainly: UT equipment is more portable, penetrates larger thicknesses, is non-hazardous, requires access to only one surface, and determines the depth location of a flaw. RT requires radiation safety measures and exclusion of workers from the test area, which imposes added cost and disruption on top of the inspection itself. On a bridge, a pressure vessel in an operating plant, or a jobsite with no access to the back side of a joint, that difference decides the method before anyone argues about sensitivity. The underlying data is in the FDOT BDK-84-977-26 final report.

What Each Method Actually Sees

RT produces a shadow photograph: density changes on film or a digital panel reveal absorption differences, so rounded volumetric flaws such as porosity clusters and slag inclusions show up beautifully. What RT handles poorly is a planar flaw lying parallel to the film plane, because it presents almost no absorption contrast. A tight lack-of-fusion plane running along the weld axis can be nearly invisible on film.

UT is the mirror image. A shear wave sweeping through the weld reflects strongly off flat, perpendicular discontinuities, which is why UT has long been favored for CJP groove welds in demanding service and why crack-like indications appear as sharp, tall responses. The trade-off is geometry sensitivity: UT reflects off the weld root, the counterbore, the backing strip corner, and any sharp irregularity, and the operator must separate real flaws from geometry. RT has no such ambiguity.

Subjectivity: UT’s Chief Weakness

The PNNL/NRC evaluation of UT in lieu of RT met the criticism head-on: flaw characterization by UT can be highly subjective based on operator experience, an area where radiographic imaging retains an advantage. A radiograph can be handed to a second reviewer who sees the same image; a conventional A-scan interpretation lives in the operator’s head unless the screen data is captured. That is the single most common argument against conventional UT in code hearings, and the industry’s answer has been recordable methods such as phased array and TOFD, which store replayable data sets rather than a technician’s notes.

What the FDOT/USF Comparative Study Found

The fear that UT-based methods would reject sound welds never got traction in the data. In the FDOT/USF comparative testing program, 35 pieces were examined with 92 phased-array UT (PAUT) tests, 54 conventional UT tests and 108 RT tests, and the reject rates came out at 8.7% for PAUT, 7.4% for conventional UT and 9.3% for RT. Those numbers cluster closely enough to allay fears that advanced UT would trigger excessive rejections. If anything, RT produced the highest reject rate, which undercuts the assumption that the “objective” film image is the truth and UT is the wild card. The study set is documented in the USDOT research records.

Speed, Cost, and Material Limits

Cost is less about hourly rates and more about disruption. RT adds film or imaging plates, processing, a radiation safety officer, barricading, permits, and lost labor while the exclusion zone stands. UT adds couplant, surface grinding, and calibration time but stops no other work. Speed favors UT on long production runs, while RT can be faster per foot on short, repetitive joints with fixed setups and no surface prep.

Material limits split the field the other way. UT struggles with austenitic stainless, duplex and coarse-grained weld metals that scatter and attenuate the beam, which is why ultrasonic testing of austenitic welds is governed by its own dedicated standard. Similar scattering behavior shows up in many aluminum MIG welding applications, where grain structure and attenuation behavior demand different search units and frequencies than ferritic steel. RT, meanwhile, struggles with very thick or dense sections that require higher-energy sources.

In modern practice the question is increasingly “which UT method” rather than “UT or RT,” because phased array and TOFD deliver the recordability conventional UT lacked, and codes such as AWS D1.5 Annex J now recognize that path. The final choice belongs to the engineer of record, written into the project specification, not to whichever inspector shows up with the gear.

Advanced Weld UT: Phased Array, TOFD, and Automated Systems

Conventional ultrasonic testing does impressive work with a single crystal: one probe, one wedge, one beam angle, one A-scan trace only the technician can see. The newer generation of weld UT keeps the same physics but changes how sound is sent, how much comes back, and what happens to the data afterward. The result is faster scanning, more consistent sizing, and a permanent record a third party can replay years later.

Phased Array Ultrasonic Testing (PAUT)

A phased array probe packs many small elements, often 16, 32, or 64, into one housing. Instead of firing them together, the instrument fires each element after a precisely calculated time delay. Those overlapping wavefronts combine in the material to form a beam that can be steered, swept, and focused electronically without moving the probe or swapping wedges.

Two benefits follow. First, one pass covers a range of angles: a sectorial scan, or S-scan, sweeps a fan of refracted angles and paints a color cross-section of the weld on screen, so the technician sees the whole weld volume instead of one slice. Second, because the angles are repeatable and software-controlled, two operators running the same program on the same weld produce closely matching results. In the Federal Highway Administration’s bridge-weld evaluation, PAUT probes with 64 elements at 5 MHz using a nominal 55° shear wave inspected gusset plate and splice detail welds, the same thickness range where conventional angle-beam UT struggles with coverage (FHWA-HRT-24-010).

Time-of-Flight Diffraction (TOFD)

TOFD uses a different trick. Rather than measuring energy reflected off a flaw face, it listens for faint sound diffracted off the flaw’s top and bottom tips. A pair of probes sits on opposite sides of the weld, one transmitting and one receiving, and the instrument times the diffracted arrivals. Because tip diffraction is far less sensitive to flaw orientation than specular reflection, TOFD sizes through-wall height with impressive accuracy, which is why it is usually paired with pulse-echo PAUT rather than replacing it.

A/B/C/S-Scan Data and Automated Scanners

The biggest cultural shift is recordability. Modern instruments capture several views at once:

• A-scan: the classic radio-frequency trace showing amplitude versus time.
• B-scan: a side view showing depth along the scan path.
• C-scan: a top-down map of the scanned region.
• S-scan: the sectorial fan view unique to phased array.

Encoded mechanical scanners go further, pulling the probe along the weld at controlled speed with positional feedback, holding coupling pressure steady and keeping the scan plan on track. Data is written to the file, then reviewed offline by a Level II or III, a workflow that mirrors how aerospace welding programs have treated inspection records for decades.

Feature	Conventional UT	PAUT / TOFD
Beam coverage	One angle per wedge	Steered range in one pass
Record	Written operator notes	Full multi-scan data file
Sizing basis	Amplitude + operator judgment	Geometry-based, replayable
Radiation needed	No	No
Equipment cost	Lower	Higher

Adoption in Bridges, Nuclear Work, and International Codes

Bridge welding moved first through AWS D1.5 Annex J, which codified advanced ultrasonic examination for bridge components beyond the limits of basic procedures. The FHWA study then put data behind the push, reporting a strong correlation between PAUT, conventional UT and RT results when measuring discontinuity dimensions, which matters because RT has long been the benchmark fabricators trust. The nuclear side asked a parallel question: the NRC had Pacific Northwest National Laboratory evaluate whether UT could replace RT for volumetric examination of carbon steel piping welds in repair and replacement activities, published as NUREG/CR-7204 in 2015. Internationally, ISO standards for weld ultrasonic testing now include ISO 13588, covering automated phased array testing of welds, alongside ISO 17640 for technique and assessment levels.

What Still Holds Advanced UT Back

Three limitations keep PAUT from being a drop-in upgrade. Equipment cost is real: a phased array instrument, multiple probes, wedges, cables, and scanner hardware cost several times a conventional flaw detector. Training is the second wall; interpreting an S-scan or TOFD trace demands more theory and screen time than reading an A-scan, and most employers still require documented practical hours before signing off a technician. The third limitation cannot be bought: acceptance criteria must be written by the engineer of record or spelled out in the referencing code, such as AWS D1.1 Clause 6 for structural work. Beautiful data is worthless if no code tells you whether a 4 mm indication passes.

Advanced UT also inherits the physics limits of conventional UT. Geometry reflections from the weld root, counterbores, and corner traps still show up in a C-scan, coarse-grained materials still scatter sound, and a recorded scan still needs a trained human to call it. The technology reduces subjectivity; it does not eliminate judgment.

Common UT Mistakes, Limitations, and How to Troubleshoot Them

Ultrasonic testing is only as good as the setup behind it. Most disputed examinations are not caused by a broken flaw detector. They are caused by a decision made in the first ten minutes: a dry patch of plate, a wedge that does not match the thickness, a scan plan that never reaches the far side of the weld. These failure modes are predictable, and almost all of them show up on the screen before they turn into a missed defect.

The Nine Mistakes That Undermine an Examination

• Poor couplant contact and dry scanning. Air is the enemy of UT. The moment the couplant film breaks up or the probe runs over a dusty patch, sound stops entering the part and the backwall collapses. The technician then mistakes lost coupling for a flaw, or scans right past a real one. Keep couplant flowing, wipe the surface, and re-check any signal that disappears without explanation.
• Leaving weld reinforcement or a backing strip in place. A proud crown refracts the beam at an uncontrolled angle before it reaches the weld, and a backing strip creates corner reflections that clutter the far root. AWS D1.1 expects scanning surfaces ground smooth and free of sharp irregularities, with reinforcement removed where the procedure requires it.
• Wrong wedge or beam angle for the thickness. A 60° probe on heavy plate leaves an unswept dead zone near the root, while an angle that is too shallow bounces the beam before it covers full penetration. Match the angle to the thickness and confirm the skip reaches every part of the weld volume.
• Skipping transfer and distance-amplitude correction. A reference block is not the same surface as the test piece. If ground condition or path length differs, the DAC curve will lie to you. Ignoring transfer correction is one of the fastest ways to under-report flaw height.
• Ignoring geometry. The root corner, counterbores, stiffeners and change-of-section features all reflect sound as convincingly as a crack. Geometry echoes are the most common false calls in the trade, and they can also mask a real root toe crack sitting beside them.
• Weak calibration discipline. Calibration is not a start-of-shift ritual. Performance checks belong periodically through the examination and again at the end. Nuclear oversight makes this explicit: NRC audits verify beam angles, scanning rate, scan directions, calibration frequency and sensitivity against the written procedure.
• Relying on one scan direction. Conventional UT uses a single-element probe with a fixed-direction beam, so coverage depends entirely on the technician’s manual raster. A lazy forward pass can miss a flaw oriented away from the beam; the FDOT comparative study repeatedly tied variability in results to scanning technique.
• Never reaching the far side. Access limitations silently reduce performance. FHWA-HRT-24-010 reported that PAUT performance was comparable to or greater than radiography when examinations were performed from both sides of the weld and the weld crown was removed. When only one face is reachable, coverage drops and so does confidence.
• Interpretation error. Subjectivity remains UT’s chief weakness. Two technicians can look at the same A-scan and call it differently, which is why acceptance decisions must tie to a written code table rather than personal judgment.

Material Limits: Where Sound Waves Give Up

No amount of technique fixes a metallurgy problem. ISO 13588:2019 covers automated phased array testing on low-alloy or fine-grained steels and points users to ISO 22825 for coarse-grained and austenitic materials, because those microstructures scatter the beam, widen grain noise and eat attenuation. Austenitic and duplex weld metals, including the nickel alloys covered in Alloy 20 Welding: Procedures, Filler Metals, and Best Practices, are the classic example, and coarse-grained alloy steels behave similarly. In those materials expect lower frequencies, smaller probes and a willingness to accept reduced sensitivity.

Thickness creates its own headaches. Thin sections push the flaw signal on top of surface noise and the initial pulse, leaving little room to separate a root toe crack from the bang. Thick sections attenuate energy, demand lower frequencies and multiple skips, and make transfer correction harder to justify accurately.

Troubleshooting Checklist

• Signal vanished mid-pass? Re-apply couplant, wipe the surface, re-check the cable and confirm ground condition.
• Backwall inconsistent? Verify wedge wear, re-zero the delay, and confirm you are on the correct skip.
• Unexpected echoes near the root? Rock the probe and check for corner trap, counterbore or backing reflections before calling a flaw.
• Results differ between technicians? Re-verify DAC/TCG, transfer correction, beam angle and scan rate against the written procedure.
• Indication found from one direction only? Add a reverse raster, a second beam angle, and access from the opposite face if the joint allows.
• Excess grain noise? Drop the frequency, reduce gain, and confirm the material is not one that ISO 22825 treats as a special case.

Run that checklist before you argue with a result. Nine times out of ten, the flaw was never in the weld; the examination was in the setup.

Preparing for a UT Test: Practical Steps for Welders and Inspectors

Most UT failures in the field are not physics failures. The sound wave behaves exactly as the textbook says; the examination goes wrong because something upstream was skipped, assumed, or left off a drawing. Whether you are the welder getting a joint ready or the technician who will sign the report, the same disciplined workflow separates a defensible examination from a costly re-shoot.

The Nine-Step Field Workflow

1. Confirm the scope from the drawings. Before anything is ground or scanned, verify which joints are actually called for ultrasonic examination. The drawings must clearly identify those joints, and the designer should indicate the examination area, the contact surface, and whether one or both sides are inspected, as set out in UFGS 05 05 23.13 10. If the print is vague, resolve it before the scanner rolls, not after.
2. Write or review the written UT procedure. AWS D1.1 requires the written procedure to contain, at minimum, the types of weld joint configurations to be examined and the technique details. Confirm the procedure matches the actual thickness range, joint geometry, and material on site.
3. Prepare the scanning surface. Grind the contact surface smooth, remove mill scale, spatter, and sharp irregularities, and remove backing strips or weld crown reinforcement where the procedure calls for it. Sound couples through a clean, flat surface, not through paint and grinding beads. Preparing aluminum joints brings its own cleaning chemistry, covered in Aluminum Cleaner for Welding: How to Prep Aluminum Right.
4. Calibrate, then verify again. Set sensitivity and timing on a reference block such as a V1/V2, IIW block, or a block with side-drilled holes, and establish your distance-amplitude curve. Recheck calibration at the start of the shift, at intervals during the examination, and once more at the end. A drifted gain setting invalidates every indication you recorded in between.
5. Plan the scan coverage. Decide the index points, the scan direction, and the number of beam angles needed to cover the full weld volume from one side or both. Write the scan plan down rather than improvising at the joint.
6. Scan at a controlled rate with overlap. Move the probe steadily, overlap each pass, and keep couplant film continuous. Fast, dry scanning is the single most common reason defects are missed.
7. Evaluate indications against the right acceptance table. Acceptance criteria live in the referencing code, not in the method standard, so use the table that the drawing or contract actually points to.
8. Document the report. Record the procedure used, operator, calibration data, scan plan, every significant indication with its location and amplitude or size, and the final disposition.
9. Coordinate repair and re-examination. After repair, the repaired weld must be re-examined using the same procedure and acceptance criteria as the original examination, with the repair and re-test documented together.

Conclusion: What This Guide Covered, and Why It Matters

Across nine sections we followed a single hidden flaw from the moment it could have been caught: what a UT test for welding is and where codes require it, the piezoelectric and pulse-echo physics behind the beep, the instruments and calibration blocks that make measurements trustworthy, the code landscape from AWS D1.1 and D1.5 to ASME Section V and the ISO standards, the Level I/II/III qualification structure, how different discontinuities read on the screen, how UT stacks up against radiography, the advanced world of phased array and TOFD, and the mistakes that quietly defeat an otherwise sound examination. The through-line is simple: UT rewards preparation. A verified calibration, a clean surface, a written scan plan, and an honest evaluation against the correct acceptance table are what turn sound waves into a defensible engineering decision.

For welders, supervisors, and inspectors alike, understanding UT changes how you think about the joints you make. You stop seeing inspection as someone else’s problem and start designing and welding for examineability. If you want to build that skill formally, an Aluminium Welding Course: What You Learn, Costs & Careers is a useful starting point for the training pathways and costs involved. Whether you end up running the probe or signing off the drawings, fluency in ultrasonic testing is one of the most reliable ways to advance a welding career.

Frequently Asked Questions

What does a UT test for welding actually check?

A UT test uses high-frequency sound waves to look inside the weld for internal discontinuities such as cracks, lack of fusion, lack of penetration, slag inclusions and porosity. The technician reads reflected echoes on the instrument screen to judge the flaw’s location, depth and size, then compares it against the governing code’s acceptance criteria.

Is ultrasonic testing better than radiography for welds?

Neither is universally better. UT excels at planar flaws like cracks and lack of fusion, needs no radiation safety zone, is portable and gives depth information. Radiography gives a permanent image that is easier for non-specialists to interpret and is stronger at characterizing small rounded volumetric flaws.

Which code governs UT of welds?

In structural steel, AWS D1.1 Clause 6 Part F governs procedure, technique, personnel and acceptance. For pressure equipment, ASME BPVC Section V Article 4 defines the method while the referencing section sets acceptance. Internationally, ISO 17640 and ISO 13588 (phased array) are widely used.

What NDT levels can perform and interpret UT?

Under the ASNT SNT-TC-1A scheme, Level I can operate and calibrate equipment under supervision but cannot interpret results. Level II may calibrate, perform and interpret examinations and accept or reject to code. Level III writes procedures, establishes techniques and resolves interpretation problems.

Does the weld need preparation before UT?

Yes. The scanning surface must be free of loose scale, rust, paint and dirt, weld reinforcement or non-permanent backing may need to be ground flush, and sharp edges removed so they cannot mask or be mistaken for defect indications. Good preparation directly improves coupling and reliability.

Can UT detect all weld defects?

No. UT is strongest on planar, beam-reflecting flaws, while very small volumetric flaws such as fine porosity may not produce a detectable echo. Austenitic and coarse-grained materials scatter sound and reduce reliability, and geometry or access limits can hide regions from the beam.

What is phased array UT and when is it used?

Phased array UT uses a probe with many small elements (commonly 16-128) pulsed with computer-calculated timing to steer and sweep the beam electronically without moving the probe. It improves coverage and repeatability, records replayable data, and is covered by standards such as ISO 13588 and AWS D1.5 Annex J.

How long does a weld UT take and what does it cost?

Time depends on weld length, thickness, access and scan plan, but UT is generally faster and cheaper than radiography because it needs no film, no radiation barricades and no work-area evacuation. The main cost driver is personnel: a qualified Level II interpreting results to a written procedure.