Ultrasonic Inspection of Welds: The Complete UT Weld Testing Guide

A pressure vessel passes every visual check and even a radiograph, yet a hidden lack-of-fusion flaw grows quietly inside a butt weld until the joint fails in service. Ultrasonic inspection is the technique that looks into the full thickness of that weld in seconds, without radiation and without damaging the part, and this guide explains exactly how it does it.

What Ultrasonic Inspection of Welds Really Means

Picture a fabrication shop on a Friday afternoon. A 40 mm thick butt weld on a structural column splice has just been ground flush, and it looks textbook perfect: even ripple, no undercut, no spatter, no visible cracks. The welder signs off, the fitter moves on, and the detail is scheduled for erection next week. Then an ultrasonic technician lays a small angled probe on the plate beside the weld, watches a trace jump on a screen, and marks a 60 mm indication sitting right at mid-thickness where the side wall of the joint never truly fused. The weld passed every visual check. It would have failed in service.

That gap between “looks good” and “is sound” is the entire reason ultrasonic inspection exists. So let’s answer the principal question early.

So what is ultrasonic inspection of welds, exactly?

Ultrasonic testing (UT) is a nondestructive method that sends high-frequency sound waves into a welded joint and reads the echoes that bounce back from internal surfaces. Because sound travels through solid steel and reflects off any boundary it meets, a trained technician can locate, depth-size and often characterise flaws hidden in the full thickness of a weld without cutting it, heating it, or damaging it in any way. No radiation, no scrap, no downtime beyond the minutes it takes to scan.

The point to grasp is that a weld is a cast structure made in place, and it can carry defects no one can see. Porosity, slag, lack of fusion and lack of penetration are volumetric or planar features buried inside the metal. Surface-oriented methods cannot reach them. Ultrasonic examination can.

NDT, NDE, NDI: does the acronym matter?

You will see the same discipline called nondestructive testing (NDT), nondestructive examination (NDE) and nondestructive inspection (NDI), often used interchangeably in shop drawings and standards. AWS treats all three as equivalent, but prefers NDE as the umbrella term, and its Guide for the Nondestructive Examination of Welds is the document that frames most of the discussion in this article (AWS B1.10). AWS defines nondestructive examination as the discovery, location, evaluation and interpretation of indications in materials, components or assemblies without impairing their future usefulness, and it is careful to add that visual, liquid penetrant, magnetic particle, radiographic, ultrasonic, electromagnetic (eddy current) and leak methods are all part of that family.

That list of seven comes straight from AWS B1.10, and it matters because no single method does everything. UT is one tool inside a toolkit, not the answer to every weld.

Why welds need inspection through the full volume

Visual inspection only sees the surface. Liquid penetrant and magnetic particle also only interrogate the surface and near-surface region, which is why they are quick and cheap but blind to mid-thickness flaws. Radiography produces a shadow image of the whole thickness, yet AWS B1.10 notes a real limitation: ultrasonic examination can detect cracks oriented in any direction provided the beam is essentially perpendicular to the crack’s major axis, whereas radiography usually cannot detect cracks lying parallel to the plate surface. The classic example is a cold crack running along the fusion line under a heavy root pass. RT may show nothing. UT, angled correctly, shows a sharp echo.

That strength is why UT is standard practice in:

• Structural steel fabrication and bridge girders, governed by alloy steel welding procedures and welder qualification records
• Pressure vessels and boilers, where full-thickness integrity is a safety requirement
• Pipeline girth welds and offshore tubulars, scanned circumferentially in the field
• Ship and marine construction, from aluminum boat hull joints to heavy structural welds
• Power piping, storage tanks, rail and heavy equipment, plus repair verification after grinding or re-welding

Cost is a factor in choosing it. AWS B1.10 observes that radiographic, ultrasonic and eddy current examinations generally cost more than visual, magnetic particle or liquid penetrant examination, with visual usually the least expensive. UT earns that premium on thick sections, on joints where access allows scanning from one side, on projects where radiation would shut down the area, and where flaws are expected to be planar.

What this guide will cover

We will start with the physics: transducers, shear waves, couplant and A-scan displays. Then we will look at which discontinuities UT finds best, the standards that govern it (AWS D1.1, ASME Section V, ISO 17640 and friends), who is qualified to run it, and how conventional UT compares with phased array and TOFD. From there you will get a step-by-step field workflow, the mistakes that cause missed defects, the safety picture, and a decision framework for choosing between methods. By the end you should be able to read a written UT procedure and know exactly what each clause is doing.

How Ultrasonic Weld Testing Works: Sound, Probes and A-Scans

Rest a steel ruler flat on a bench and tap one end with a coin. You hear a clear ring. Now press the ruler firmly against your palm and tap again: the sound nearly vanishes. That tiny experiment holds the entire idea behind ultrasonic weld testing. Sound travels through solid material, bounces off anything that interrupts it, and tells you where that interruption sits. Ultrasonic testing (UT) simply does this at frequencies too high for us to hear, with timing precise enough to locate a flaw a fraction of a millimetre deep.

The Transducer and the Pulse-Echo Principle

The tool that makes this happen is the transducer (often called a probe). Inside its housing sits a piezoelectric crystal. Apply a short electrical pulse and the crystal physically changes shape, pushing against the material and launching a burst of sound. When an echo comes back, the same crystal flexes in reverse and generates a small voltage the instrument can display. One crystal both shouts and listens.

Working method is called pulse-echo: the instrument sends a pulse, then switches to listening. Because sound in steel travels at a fixed, known speed, the time between the outgoing pulse and the returning echo converts directly into distance. Two echoes arriving at different times mean two reflectors at different depths. That is really all the math involved.

Frequency, Wave Type and Beam Angle

Weld UT generally runs between roughly 2 and 5 MHz. Lower frequencies punch deeper through thick, coarse-grained or highly attenuative material but see smaller flaws less clearly; higher frequencies resolve tiny reflectors beautifully but lose energy faster. For structural groove welds, AWS D1.1 specifies angle beam transducers of 2 to 2.5 MHz with elements 15 to 25 mm long, noting that a 5/8 in (16 mm) square element is the most common choice.

Two beam families are used:

Probe type	Wave	Primary job
Straight-beam (0°)	Longitudinal	Screening base metal for laminations; checking thickness
Angle beam (45°, 60°, 70°)	Shear (transverse)	Interrogating the weld volume and fusion faces

ASME Boiler and Pressure Vessel Code Section V shear wave examinations use angle beam search units at 45, 60 or 70 degrees as the procedure, qualified to Section V, directs. Shear waves are used because they reflect strongly off planar flaws and do not pass through liquids, which conveniently keeps the energy where you want it.

Angle selection is not arbitrary. A flat, planar reflector returns the strongest signal when the beam strikes it square-on, so inspectors choose the angle expected to sit nearest perpendicular to the anticipated flaw. Root and fusion-face indications in AWS D1.1 are therefore evaluated with whichever of 70, 60 or 45 degrees is closest to perpendicular to the expected fusion face. A vertical crack in a groove weld, for instance, is best served by a 60° or 70° beam travelling nearly horizontally across it.

Skip Distance, Legs and the A-Scan

Because an angled beam bounces off the back wall, its path is folded. The distance from the entry point to the first back-wall contact is the skip distance; the path before reflection is the first leg, after reflection the second leg, and so on. Inspectors aim the skip so the beam crosses the weld volume at a useful angle, and they must remember which leg the signal came from, since the same A-scan position can represent two different physical points.

The A-scan is the instrument’s display: horizontal position represents depth (time), vertical height represents signal amplitude. The tall, steady pulse at the far left is the initial pulse; the recurring peak that walks across the screen as the probe moves is the back-wall echo. If the back-wall echo drops while the probe is properly coupled, something in the material is intercepting the beam.

Depth becomes trustworthy only after calibration. A typical setup builds a distance-amplitude correction (DAC) line using a 1.6 mm (1.5 mm) side-drilled hole or reference hole in an IIW-type calibration block, checked at several depths so that identical flaws produce comparable heights regardless of where they sit. Before any of that, standard procedures require the entire base metal through which the sound will travel to be scanned first with a straight-beam probe, catching laminar reflectors that could block or masquerade as weld flaws.

Which Weld Discontinuities Ultrasonic Testing Detects Best

Ultrasonic testing has a reputation for being the method that “sees everything” inside a weld. It does not. UT is brilliant at some discontinuities, merely adequate at others, and genuinely awkward with a few. Knowing where the technique is strong and where it struggles is the difference between a report you can trust and one that quietly misses the flaw that later becomes a crack.

What ASTM E273 Treats as the Target List

The quickest way to understand UT’s official job description is to look at ASTM E273, the standard practice for pulse-echo ultrasonic examination of pipe and tube welds. It names the discontinuities the technique is designed to find: lack of fusion, pinholes, lack of penetration, longitudinal cracks, general porosity, and inclusions. That list is deliberately short and practical, and it mixes sharp planar flaws with rounded volumetric ones, which is exactly the tension at the heart of ultrasonic interpretation.

Planar vs Volumetric: Why Reflector Shape Decides the Result

Everything about UT detectability comes down to geometry. A planar flaw, meaning a crack, a cold-lack-of-fusion plane, or an unwelded land, presents a flat face to the sound beam. Hit that face near perpendicular and nearly all the energy bounces straight back to the transducer, producing a tall, sharp echo. A volumetric flaw such as a rounded porosity cavity or a glob of slag presents curved surfaces that scatter the sound in many directions. Only a fraction returns to the probe, so the echo is smaller and behaves differently when the probe moves.

This is why ultrasonic codes and training programs teach reflector characterization as a core skill. The operator swings the probe in an arc centered on the reflector and watches how the amplitude changes. A planar reflector holds a strong, narrow peak and drops away quickly as you swing. A spherical reflector gives a broader, gentler rise and fall, because its curved face keeps catching some energy over a wider range of angles. That single motion separates a real fusion defect from a harmless rounded inclusion faster than any amount of staring at the screen.

Cracks, Lack of Fusion and Lack of Penetration

These are UT’s headline performances. Cracks and lack of fusion are flat, tight, and usually oriented in ways that a properly angled shear-wave beam intercepts close to perpendicular. A transverse crack may need a different scanning approach entirely, but once the beam hits it, the echo is strong and sharply responsive to probe movement. Lack of penetration along the root often produces a clean, repeatable signal from the unwelded land.

The consequences are severe: AWS D1.1 Annex O and its acceptance tables treat cracks, lack of fusion, and lack of penetration as not acceptable regardless of measured length. There is no “it’s short enough to pass” clause. A ten-millimeter lack-of-fusion indication is rejected the same way a hundred-millimeter one is, which is why UT operators are trained to chase these signals carefully rather than write them off as marginal.

Porosity, Slag Inclusions and Undercut

Volumetric discontinuities are a different conversation. Scattered, fine porosity often produces a scatter pattern of small, grass-like indications across the screen rather than one dominant peak. Cluster porosity concentrates that scatter into a distinct zone with multiple signals within a defined range. Slag inclusions sit somewhere in between: elongated slag films can behave almost planar and return a respectable echo, while rounded slag pockets behave like porosity. Undercut at the weld toe is detectable but competes with the geometric reflection from the toe itself, so it demands a careful scan angle.

Separating Real Flaws from Geometric Noise

Every weld has reflectors that are not defects: the root face, reinforcement geometry, the “ladder rung” reflections from the bevel land, and counterbore transitions. Learning to recognize these non-relevant indications is as important as finding real ones. The giveaway is usually position and repeatability. A geometric echo sits exactly where the root geometry says it should, appears identically at every point along the weld, and does not respond to the swing test the way a planar flaw does. Operators are taught to calculate where the root reflection should appear before scanning, so they are never surprised by it.

What the Latest Detection Research Shows

A 2025 probability-of-detection study on 9% nickel steel girth welds put numbers on these differences. For planar defects such as lack of fusion, phased array UT achieved an a90 detection size of 6.87 mm, while radiography required 12.72 mm to reach the same confidence. The trade-off appeared in the false-positive rate: PAUT produced 13.59% false positives against just 0.58% for radiography (Journal of Materials Research and Technology, 2025). The same study confirmed the conventional wisdom: ultrasonic methods were more efficient on planar defects, while radiography performed better on volumetric flaws such as incomplete root penetration.

Discontinuity	A-scan Behavior	UT Effectiveness
Crack (longitudinal)	Tall, sharp peak; drops fast when swung	Excellent
Lack of fusion	Strong signal at predictable depth on bevel face	Excellent
Lack of penetration	Repeatable root-level echo, strong response	Very good
Slag inclusion (elongated)	Moderate to strong, moderately directional	Good
Porosity / cluster porosity	Multiple small scatter indications	Fair to good
Undercut	Moderate signal near toe geometry	Fair, needs skill
Geometric root reflector	Stable, position-predictable, swing-insensitive	Must be discounted

The practical takeaway is simple: specify the method around the flaw you actually fear. Planar, safety-critical discontinuities play to ultrasonic testing’s strengths. Heavy volumetric quality concerns may justify pairing UT with radiography or another method for confirmation.

Codes, Standards and Written Procedures Governing Weld UT

A contractor once submitted a weld inspection report that read: “No relevant indications found.” The client’s engineer rejected it within a day. Nothing in the report said which procedure was followed, what sensitivity was used, which code was applied, or what “relevant” meant. The scan itself may have been excellent, but the paperwork was worthless. Ultrasonic inspection of welds does not happen in a technical vacuum. Every credible UT job runs on three things: an approved procedure, a qualified operator, and acceptance criteria agreed before the first pulse was ever sent into the steel.

The Three Categories of Weld UT Standards

It helps to sort the standards into three families, because they answer three different questions:

1. Personnel standards answer “who is allowed to do this?” They define training, examination, vision testing and certification levels. Examples: ASNT SNT-TC-1A and ISO 9712.
2. Examination technique standards answer “how is the test performed and reported?” They define probes, frequencies, calibration, scanning coverage, sensitivity and documentation. Examples: ASME Section V, ISO 17640 and ASTM E273.
3. Acceptance criteria standards answer “is this weld acceptable?” They convert recorded amplitude, length and location into an accept or reject decision. Examples: AWS D1.1 Tables 8.2 and 8.3, and ISO 11666.

A common field error is mixing them up: running a technique from one code and judging results by another. Most contracts explicitly forbid that. The table below maps the standards you are most likely to meet.

Standard	What it governs	Typical use
AWS D1.1 (Annex O, Tables 8.2/8.3)	UT technique and acceptance for structural steel	Buildings, general steel fabrication
AWS D1.5	Bridge welding, including NDT requirements	Highway bridges
ASME Section V, Articles 4 and 14	General UT and phased array UT methods	Boilers, pressure vessels, piping
ISO 17640:2018	Weld UT technique, test levels and assessment	International and European work
ASTM E273	UT of the weld zone of welded pipe and tubing	Pipe and tubing production
API 1104	Welding and NDT of pipeline girth welds	Transmission pipelines
ASNT SNT-TC-1A	Personnel qualification and certification	Employer-based certification schemes

AWS D1.1 and AWS D1.5: The Structural and Bridge Codes

Alamo Welding & Boiler Works-style fabrication shops living under AWS rules work mainly with D1.1. Its ultrasonic requirements sit in Annex O, while Tables 8.2 and 8.3 turn the recorded data into decisions: one table sets limits for linear indications such as cracks, lack of fusion and elongated porosity, and the other handles rounded indications, sizing them against a drilled-hole reference. Welds that will never be radiographed still have to meet the same rigor, so the acceptance tables are always specified before work starts, not after.

AWS D1.5, the bridge code, historically required complete joint penetration welds subjected to tensile stress or stress reversal to be examined by radiography rather than UT. That single clause slowed phased array adoption across bridge fabrication for years, because fabricators were reluctant to invest in PAUT equipment the governing code would not let them use in place of film.

ASME Section V: Articles 4 and 14

The ASME Boiler and Pressure Vessel Code separates methods from acceptance. Section V describes how to examine; the construction code (Section VIII, Section I, B31.3 and similar) says whether the result passes. Article 4 covers general ultrasonic examination requirements for welds, including search units, calibration blocks, beam search and evaluation. Article 14 covers phased array ultrasonic examination, and its Mandatory Appendix VIII sets out calibration and validation requirements for advanced ultrasonic systems, including demonstration that the equipment, software and scanning arrangement produce repeatable results before real work begins.

ISO 17640 and Its Family of Standards

Internationally, ISO 17640 (last updated in 2018) is the common reference for weld ultrasonic testing, covering techniques, testing levels and assessment. It carries an interesting lineage: it replaced EN 1714, which itself had replaced older national standards such as DIN 54125 and BS 3923. ISO 17640 does not stand alone. It references ISO 5577 for vocabulary, ISO 9712 for personnel certification, ISO 11666 for acceptance levels, ISO 16810 for general principles, ISO 16811 for sensitivity and range setting, ISO 16826 for the examination for discontinuities, and ISO 23279 for the related examination practice. Because the family is fully cross-referenced, an examiner working to ISO 17640 inherits a complete, internally consistent system rather than a single isolated document.

Why a Written Procedure Is Mandatory

AWS B1.10 requires that all nondestructive examination be performed under an approved written procedure covering test preparation, performance and interpretation, precisely so results are reliable and reproducible between operators and between days. Without that document, two technicians can scan the same weld and reach different conclusions, and neither report carries authority. The same logic explains why acceptance criteria must be fixed before examination starts: judging a weld against rules chosen after seeing the data is not inspection, it is negotiation. For anyone working across borders, ASTM E273 fills a niche the others do not, focusing specifically on the weld zone of welded pipe and tubing, while aerospace work layers additional program-specific specifications on top of all of it.

With the standards mapped, the next question is obvious: who is legally and technically qualified to execute them? That is a matter of certification levels, written practices and continuing vision checks, which is where we go next.

Who Can Perform Weld UT: Personnel Qualification and Certification

With the amount riding on a single weld, ultrasonic inspection cannot be left up to guesswork. From pressure boundaries to heavy structural connections, the examination, interpretation, and evaluation of weld UT must be carried out by personnel certified to UT Level II or higher under schemes such as ASNT SNT-TC-1A, ISO 9712, or an equivalent national standard. On top of this, every certified individual must hold a current, valid vision test to ensure they can accurately read the displays in front of them.

Understanding the Three Certification Levels

Not every level carries the same weight of responsibility. The breakdown below is the industry standard used across the vast majority of code work:

• Level I: Primarily trained to set up equipment, perform calibrations under supervision, and carry out scanning to a written procedure. A Level I is not authorized to independently interpret or evaluate indications against acceptance criteria.
• Level II: Able to set up, calibrate, scan, interpret, and evaluate indications to the applicable code. For example, a Level II working on a multi-pass butt weld may spot a cluster of reflections within the root area, measure their amplitude against a DAC curve, apply the correct sizing method, and determine whether they are relevant defects or harmless geometric reflectors.
• Level III: The highest technical authority. A Level III is responsible for writing, reviewing, and approving UT procedures, verifying calibration adequacy, and signing examination reports for critical work. This authority is put to use on high-stakes contracts, such as the Brookhaven National Laboratory weld inspection contract, where every procedure must be formally approved before scanning begins. Uniquely, Level III candidates that perform examinations must pass a UT Level III demonstration consisting of a UT Level II practical examination to prove their hands-on competence.

Written Practices, Vision Testing and Major Schemes

Before any certification is issued, most employers are required to maintain a written practice. This document dictates the minimum training hours, on-the-job experience, general and specific examinations, vision requirements, and recertification intervals for their personnel. These requirements are enforced across the four most utilized schemes within the trade: ASNT SNT-TC-1A (the most common employer-based system), ASNT CP-189 (a more rigid, centrally controlled program), ISO 9712 (widely used for international projects), and CAN/CGSB 48.9712 (Canada’s national certification scheme).

With the rise of advanced methods, base certification alone is no longer enough for every job. Public procurement documents, such as DOE/BNL statements of work, now require documented PAUT-specific training on top of a base UT Level II certification before an inspector can run array-based scans.

Competence Over Convenience

Certification is not a formality to be rushed. From maintaining a vision test, to having a procedure approved by a Level III, to respecting code-mandated waiting periods, each requirement exists to protect the integrity of the joint. The goal is simple: ensure that the person rejecting, or accepting, a weld has both the knowledge of the standard, and the proven ability to tell a real flaw from a trick of sound.

Conventional UT vs Phased Array vs TOFD: A Head-to-Head Comparison

Ask three inspectors how they would volumetrically examine the same butt weld and you will likely get three answers: a conventional angle-beam probe, a phased array scanner, or a TOFD pair. All three are ultrasonic. None of them behaves quite like the others. Understanding how they differ is what lets you match the method to the joint, the code and the budget instead of defaulting to whatever instrument happens to be in the case.

Conventional UT: one beam, one A-scan

Conventional ultrasonic testing fires a single focused pulse from one crystal through a wedge at a fixed angle, most often 45°, 60° or 70°, and displays the returning echoes as amplitude against time on an A-scan. The operator physically moves and angles the probe to interrogate the weld volume, judges which peaks are relevant, and sizes indications using techniques such as the 6 dB drop. It is inexpensive, rugged, field-proven and covered by almost every code in existence. Its weakness is equally simple: the result lives in the operator’s head and on a paper report. As FHWA research on bridge weld examination notes, conventional UT data cannot be stored as a permanent record, which makes re-evaluation or dispute resolution genuinely difficult.

Phased array: many elements, steered and focused sound

A phased array probe contains multiple elements, usually between 32 and 128, each acting as a single transducer. By applying focal laws that control the amplitude and time delay of each element’s firing, the beam can be steered electronically and focused at depth without changing the probe. That produces linear scans along the weld, sectorial (S-scan) sweeps that fan through a range of angles in one pass, and dynamic focusing that keeps the beam tight at the depth of interest. The payoff is coverage and repeatability: one pass gathers far more data than a hand-scanned single probe, and the file can be archived. Pair the array with a linear encoder and the scan becomes a permanent, plottable record, which is exactly the advantage FHWA highlighted.

TOFD: timing diffraction instead of reading amplitude

Time-of-flight diffraction uses a transmitter-receiver pair straddling the weld. Instead of measuring how loudly a flaw reflects sound, it measures the time-of-flight of waves diffracted from crack tips and flaw edges. Because timing is the measurement, TOFD is far less sensitive to couplant condition, probe pressure and flaw orientation than amplitude-based methods, and it sizes defects by geometry rather than by signal height. It does have limits: near the scanning surface the strong lateral wave can mask flaws, reducing reliability in the top few millimetres, and scattering in coarse-grained austenitic stainless steel makes it a poor fit there.

Method	How it works	Strengths	Weaknesses	Data & cost
Conventional UT	Single probe, fixed-angle pulse-echo A-scan	Cheap, portable, code-universal, fast to deploy	Operator-dependent, no stored record, hard on transverse flaws	Paper report; lowest equipment cost, modest training
Phased array (PAUT)	32–128 elements, focal-law steering and focusing	Electronic sectorial and dynamic focusing, fast coverage, archived data with encoder	Costlier instruments, procedure-heavy, needs trained analysts	Digital volume of data; higher capital and training cost
TOFD	Dual-probe diffraction, time-of-flight measurement	Orientation-independent, accurate sizing, insensitive to couplant	Near-surface dead zone under the lateral wave, poor in coarse-grained material	Digital A-scan strips; moderate cost, steep concept curve
Radiography (RT)	X-ray or gamma exposure of film or detector	Intuitive permanent image, good for porosity and slag	Radiation controls, one-sided access, planar flaws can hide edge-on	Archived film/digital image; source handling cost and licensing

What comparative research shows

The worry with any newer method is that it will reject sound welds. A 2014 comparative study of RT, conventional UT and PAUT on steel bridge welds under AWS D1.5 found rejection rates using PAUT similar to those of RT and UT, allaying concerns about unnecessary rejections. Separately, an ICEM study reported TOFD detecting 100% of the specimen defects found by radiography and pulse-echo UT, with superior detection of planar cracks and lack of fusion that radiography never showed. That pattern keeps repeating: RT sees volumetric flaws beautifully, TOFD sees planar flaws reliably, and PAUT bridges both while keeping the data.

In practice, the combination works because the methods fail differently. PAUT gives full coverage, sectorial angle sweeps and an archived, encoded record; TOFD gives geometry-based sizing and orientation-independent detection through the wall. Used together, each method’s blind spot is covered by the other’s strength, and the fabricator ends up with defensible data instead of a single trace on a screen.

Step-by-Step: How a Weld UT Inspection Is Actually Performed

Ultrasonic testing of a weld is not a quick swipe with a probe. It is a controlled sequence that must be followed by a qualified technician to produce a traceable and defensible result.

Step 1: Review the Written Procedure

Before any scanning, the technician reviews a written examination procedure based off of a recognized standard such as ASTM E273 or AWS D1.1. This document defines the probe angles, frequencies, calibration reflectors, sensitivity, and acceptance criteria for the job. The examination scope is verified as well, which typically covers at least 25 mm (1 inch) past the weld toe on each side and the entire heat affected zone.

Step 2: Prep the Weld and Control Conditions

Sound cannot travel through air, so contact is the priority. Weld spatter, rust, dirt, and loose scale must be removed through grinding or wire brushing. The part must also be cool enough for inspection, as surface temperatures are commonly limited to about 40 °C (104 °F). At elevated temperatures, couplant thins and wedge performance shifts. This level of cleanliness is no different from the care taken to prep aluminum before welding, as any contamination will break acoustic contact.

Step 3: Screen With a Straight-Beam Probe

A zero-degree probe is scanned over the base metal on both sides of the weld first. This screening helps identify laminations, gross porosity, or inclusions that could block or distort the angle-beam sound path. Any back-wall loss or strong mid-wall indication found here is not necessarily a weld defect, however it will warn the technician to be cautious for shadowing throughout the angle-beam scan.

Step 4: Calibrate Time Base, Index Point, and Beam Angle

Using an IIW (V1) or V2 calibration block, the sound path range, zero point, probe index point, and actual beam angle are all verified. These checks are crucial, as a chipped wedge or shifted index point will misplace any indication found on the weld.

Step 5: Set Sensitivity With a DAC Curve

A distance-amplitude correction (DAC) curve is plotted using known reflectors at varying depths so that echoes of equal flaw size read at a comparable height regardless of their sound path. From here, two sensitivity levels are applied:

• Search sensitivity: Raised 6 to 12 dB above the level established for the maximum sound path to catch marginal indications during scanning.
• Evaluation sensitivity: The level used to measure and judge any indications found.

Calibration reflectors are then adjusted so that none read below 40% of full screen height.

Step 6: Apply Transfer Correction and Scan

Before full scanning, a transfer correction is performed to account for the difference in surface roughness and material between the calibration block and the actual part. With this applied, scanning can begin using the standard movement pattern, commonly known as movements A through E. Scans are completed from both sides of the weld with at least 15% overlap between passes, as a flaw oriented away from the beam on one side may be much more reflective from the other.

Step 7: Evaluate, Size, and Reverify

Any indication exceeding 50% of the DAC line must be evaluated. For those that require sizing, the length is measured using the standard 6 dB drop method. Findings are marked directly on the weld surface, plotted onto a weld map, and documented with their location and length.

Calibration must be re-checked at the end of the exam, every 4 hours of examination, or anytime a system change occurs. If repairs are made, both the gouged surface and the finished repair are re-examined to the original standard. Finally, the results are compiled into a full report listing the procedure, equipment, calibration, coverage, indications, and acceptance verdict in line with AWS D1.1 acceptance tables. For additional field guidance, the Sonatest ultrasonic weld inspection guide is commonly referenced.

Common Mistakes, False Indications and UT Limitations

Ultrasonic testing has a habit of producing confident readings that turn out to mean nothing. An A-scan shows reflectors, not defects. The instrument cannot tell a fatigue crack from a counterbore edge, or a slag stringer from the corner trap at the root of a joint. That is why AWS D1.1 and ASME Section V require every relevant indication to be classified by identity, amplitude, location and extent before any accept-or-reject decision is made. Most UT problems in weld work come from a short, predictable list of causes.

Geometric and Root Reflections: When the Weld Talks Back

The root of a butt weld is a natural corner reflector. A square root opening, a backing bar, a counterbore or a wall-thickness step produces a sharp echo that looks identical to lack of penetration: same depth, same amplitude, similar shape. Root geometry is the single biggest source of false calls in pipe and plate work.

The defence is motion study: slide the probe back and forth and watch how the signal behaves. A real planar flaw rises and falls symmetrically over a short travel, while geometric echoes persist, split into the classic double “devil’s echo,” or appear from both scan directions at the expected skip distance. Comparing amplitude from the left and right scan positions helps too: a genuine planar flaw almost always has one strong side.

Missed Transverse Cracks and Oriented Flaws

Angle-beam shear waves work best on flaws sitting roughly perpendicular to the beam, and many cracking mechanisms, including fatigue, restraint cracking and hydrogen-induced cold cracking, run transverse to the weld axis. A single longitudinal pass can walk straight past them.

This is why ASTM E273 and AWS procedures require every butt weld to be examined from each side of the weld axis, and why most procedures add a transverse scanning pass wherever cracking is credible. On certain high-strength steels, AWS D1.1 also imposes a waiting period, commonly 48 hours, before UT, because hydrogen cracks need time to open far enough to reflect sound.

Couplant, Wedge Wear and Calibration Drift

Sound will not cross an air gap. A dry patch, a contaminated surface or a dried-out probe face silently removes sensitivity, and the report still reads clean. Surface condition matters as much as couplant choice; see our aluminum surface prep guide for the degreasing discipline couplant needs. Wedges wear too: as the plastic face erodes the beam angle shifts and the index point creeps, so periodic verification of angle, beam path and sensitivity is mandatory, typically every four hours or after anything that might disturb the instrument.

Material and Geometry Limitations

AISI Type 304 austenitic stainless steel welds are widely considered difficult for conventional UT: coarse, anisotropic grains scatter and attenuate the beam, filling the A-scan with grass-like noise that buries real reflectors. Austenitic, duplex and nickel-alloy welds often need lower frequencies, specialized probes, or a different method altogether.

Laminations are a separate trap. Rolled plate can carry a horizontal planar flaw that rings loudly in straight-beam scans and plants ghost indications at misleading depths. AWS D1.1 answers this with a required straight-beam scan of the base metal for laminar reflectors before weld UT, plus alternate scanning procedures when the back-wall echo is lost. Thin-wall and small-diameter work, like the thin sections in alloy wheel repair, squeezes the exam into a zone where the entry dead zone and reverberation hide flaws.

Sizing Accuracy and Human Factors

Finding the flaw is only half the job. FHWA research on weld inspection documents the familiar pulse-echo bias: large defects tend to be undersized while small defects tend to be oversized, a direct consequence of beam spread. The same research found hand-held phased array results shifting with probe position, with paired-probe configurations reading a reflection as high as 6 mm depending on setup.

Human factors stack on top: fatigue, time pressure, noisy sites, unfamiliar weld geometry, and operators working beyond their certification level.

When to Bring in a Second Method

UT is not always enough on its own. Pair it with magnetic particle or dye penetrant testing for surface-breaking flaws, which the near-surface dead zone tends to hide. Add radiography, or a combined PAUT and TOFD examination, when thickness, material or the governing code demands volumetric confirmation; porosity, such as the gas porosity familiar from aluminum MIG welds, is volumetric and reads nothing like a planar flaw. For critical welds or an A-scan nobody can interpret confidently, a second method is not an admission of weakness; it is what the standards and plain common sense expect.

Safety, Environment and Working Conditions During Weld UT

Ask most inspectors why they prefer ultrasonic testing over radiography and the first answer is rarely resolution or speed. It is usually freedom. Ultrasonic examination uses no ionizing radiation, so there are no lead-lined bunkers, no evacuating the bay, and no waiting for a licensed radiography crew’s exclusion zone to clear. The adjacent welding station keeps running and the schedule keeps moving. That is a genuine safety and productivity advantage, and a large part of why UT spread so quickly through structural and pressure-vessel fabrication shops.

But “no radiation” is not the same as “no hazard.” Weld UT has its own exposure profile, and it is one of the most underestimated things about the job. The real risks are noise you cannot quite hear, couplant you get on your skin all day, awkward postures held for hours, and confined spaces where every one of those problems gets worse.

The Radiation Advantage, Stated Honestly

Compared with radiography, UT removes the entire category of ionizing-radiation risk: no dose tracking, no film badges, no area surveys, no interlocked barricades, and no accidental exposure incident reports. Radiographic work also typically requires a certified radiographer, posting of controlled areas, and coordination with every other trade nearby. Ultrasonic work requires none of that. This is why a UT procedure can be run on a live construction floor while a carbon arc gouging repair happens two bays over, whereas a shot of radiography would shut the area down.

The Real UT Hazard: Noise You Cannot Quite Hear

Here is where welders and inspectors get caught off guard. Ultrasound itself is above human hearing, but ultrasonic equipment does not produce a single clean frequency. OSHA’s Technical Manual notes that most of the audible noise generated by ultrasonic sources consists of subharmonics. An ultrasonic welder operating at 20 kHz can therefore produce audible noise at 10 kHz, right inside the upper edge of human hearing, in a band most people rarely think to protect.

Regulators take this seriously enough to publish numbers. In 2003, OSHA adopted the ACGIH airborne ultrasound threshold limit values, roughly 30 dB higher than the earlier, much stricter levels, with the important exception that limits tighten sharply when solid or liquid coupling gives acoustic energy a direct path into the worker’s body. Canada’s Health Canada Safety Code 24 is more explicit: it caps one-third-octave bands at 16 and 20 kHz at 75 dB, rising to 110 dB at 25 kHz and above, and states that total linear sound pressure level delivered to the body must never exceed 137 dB to avoid tissue heating.

The ordinary OSHA noise standard (29 CFR 1910.95) still applies on top of the ultrasound limits: when eight-hour average exposure exceeds 85 dBA, the employer must offer hearing protection or audiometric testing. A 2017 occupational study of 90 workers exposed to ultrasonic equipment noise found measurably worse hearing thresholds at 8 to 12.5 kHz compared with a matched group exposed only to audible noise of a similar overall level (PMC9605123). That study is a warning, not a verdict, but it is exactly why ultrasonic equipment deserves its own risk assessment rather than being waved through as “quiet work.”

Couplant, Skin and Housekeeping

Couplant is the second daily exposure. Water-soluble gels are generally mild, but eight hours of repeated application means constant skin contact, and some technicians develop irritation or dermatitis, especially when hands stay damp. Nitrile gloves, a thin barrier cream, and washing before breaks reduce it. Alcohol-based cleaners used to degrease the surface beforehand can dry skin badly, so plan that sequence deliberately. Puddles of couplant on grating are a slip hazard, runoff needs containment in food, marine, or potable-water work, and in winter, frozen couplant on a scaffold deck is its own emergency. If you are prepping non-ferrous surfaces, our aluminum cleaning guide covers the degreasing step that both weld quality and UT coupling depend on.

Ergonomics, Hot Steel and Confined Spaces

Holding a probe at consistent pressure for a full shift is genuinely hard on the wrist, elbow and shoulder, and most scanning happens crouched, overhead, or hanging off staging. Rotate hands, use probe holders or automated scanners where available, and treat scaffolding and fall protection as part of the UT procedure, not an afterthought. Freshly completed welds are obvious burn and steam hazards: couplant flashed off hot steel produces a scalding mist and a ruined coupling path, so surface temperature limits belong in the written procedure. Inside tanks, vessels and pipe, everything compounds. Confined-space entry permits, gas testing, an attendant, a rescue plan and forced ventilation all apply before the first pulse is fired, and reverberant metal shells amplify both subharmonic noise and fatigue.

Reporting and Stop-Work Obligations

Because UT leaves no permanent record unless the technician makes one, safety reporting carries extra weight. Noise exposures, couplant incidents, near misses on staging, and any equipment fault that could invalidate calibration all need to go in the inspection log alongside the actual flaw data. Inspectors also hold real stop-work authority here: an unsafe work area or an unreadable calibration check is grounds to halt the examination, and documenting that decision protects both the inspector and the fabrication record.

Choosing the Right Method: Costs, Practical Tips and What Comes Next

By this point you know how ultrasonic inspection works, what it finds, and who is qualified to run it. The question that still keeps fabricators and inspectors awake at night is a more practical one: which method do I actually pick for this weld, on this job, at this budget? The honest answer is that no single technique wins every time, and the good news is that the industry has already written down how to think about it.

A Framework Instead of a Guess

AWS B1.10, the recommended practice for ultrasonic testing of butt welds, tells you to choose an examination method by weighing four things together: the advantages and limitations of each technique, the type and size of discontinuity you anticipate, the acceptance standard the job must meet, and cost. Run through those four filters and most decisions collapse into one or two sensible options.

Method	Best suited to	Key strengths	Main limitations
VT (visual)	Every weld, every time	Instant, free, catches surface defects and workmanship issues	Sees nothing below the surface
PT / MT	Surface-breaking flaws, nonporous materials	Cheap, fast, simple, sensitive to tight surface cracks	No volumetric data; MT only works on ferromagnetic metals
Conventional UT	Thick butt welds, full-volume screening	Deep penetration, no radiation, instant results, portable	Operator dependent, limited permanent record
RT	Where a permanent film record is required	Excellent for porosity and slag, easy to review by others	Ionizing radiation, access both sides, time and licensing
PAUT	Complex geometry, repetitive production welds	Sectorial and dynamic focusing, fast scanning, fully recordable	Higher equipment and training cost
TOFD	Through-wall sizing of planar flaws	Very accurate height sizing, relatively unaffected by orientation	Near-surface dead zone, poor at corner and root flaws

Material matters too. Highly alloyed, coarse-grained, or anisotropic metals such as austenitic stainless and duplex alloy steel weldments scatter and refract the beam, which pushes many shops back toward radiography or lower-frequency phased array setups. Thin aluminium fabrications, common in aluminium MIG welding work, behave differently again, since near-surface dead zones make conventional UT on very thin walls genuinely difficult.

Where Combinations Win

A DTIC/BTS report concluded that phased array UT could substitute for both radiography and conventional UT in bridge weld inspection, provided an appropriate procedure is followed, and that considerable cost savings could be realized in the process. The full report is available here, with related FHWA research on ultrasonic inspection of bridge welds supporting the direction of travel.

Nuclear-industry experience is more cautious. Studies summarized in a comparative review of bridge and pressure-boundary welding found that neither pulse-echo UT alone nor TOFD alone can fully replace the combination of manual UT plus radiography. The most reliable path to equivalence was combined pulse-echo and TOFD — the pairing that has since become the default on pipeline girth welds.

Practical Tips That Save Jobs

• Buy your own calibration blocks. IIW or V1 blocks and your code-specific reference blocks mean calibration never becomes the reason a job stalls.
• Plan scan paths before you couplant anything. Sketch the movements, the required overlap, and which side of the weld each probe must run from.
• Document obsessively. On critical work the end-of-inspection deliverables include procedure identification, calibration data, indication data, code evaluation, disposition, and a Level III signature, typically with a five-business-day report turnaround.
• Verify with a second method. When a UT result decides an expensive repair, confirm it with MT or PT on the surface, or a targeted radiograph.
• Watch transfer loss and surface prep. Rough surfaces eat sensitivity quietly; correct for it or regrind.

What Comes Next

The clear trend is automation: crawlers and scanners feeding phased array data into software that can be re-evaluated long after the truck has left site. Closely behind is codified PAUT acceptance, with pipeline and offshore standards already writing explicit criteria so that digital ultrasonic data carries the same contractual weight as a radiograph.

So, to restate the central answer of this guide: ultrasonic inspection is the technique that examines the full volume of a weld from one side, in seconds, without radiation or damage, detecting the planar lack-of-fusion flaws that visual checks and even some radiographs miss. Choose it when your code allows it, qualify the people running it, calibrate honestly, and pair it with another method whenever the decision is critical. Do that, and the hidden flaw that starts quietly inside a butt weld gets found long before it ever fails in service.

Frequently Asked Questions

What is ultrasonic inspection of welds?

Ultrasonic inspection of welds, or weld UT, is a nondestructive test in which a transducer sends high-frequency sound into the weld and measures echoes returned from internal discontinuities. Because the sound travels through the full thickness, UT can find cracks, lack of fusion and lack of penetration without cutting or irradiating the part.

Which defects can ultrasonic testing find in a weld?

UT is especially effective on planar flaws such as cracks, lack of fusion and lack of penetration, and it also detects slag inclusions, porosity and cluster porosity. Volumetric defects like incomplete root penetration are sometimes easier to characterize radiographically, which is why codes sometimes pair the methods.

What codes and standards apply to weld ultrasonic testing?

Common references are AWS D1.1 (including Annex O for the alternative UT technique), AWS D1.5 for bridges, ASME Boiler and Pressure Vessel Code Section V Articles 4 and 14, ISO 17640 with ISO 11666 acceptance levels, and ASTM E273 for pipe and tube welds. Acceptance criteria must be specified separately from the examination method.

What qualifications does a weld UT technician need?

Interpretation and evaluation require certification to at least Level II in ultrasonic testing under a written practice based on ASNT SNT-TC-1A, ISO 9712 or an equivalent national scheme, plus a current vision test. A certified Level III normally writes or approves procedures and signs reports on critical work.

Is ultrasonic testing better than radiography for welds?

Neither is universally better. UT detects planar defects such as lack of fusion more reliably and uses no ionizing radiation, while radiography gives a permanent film record and is stronger on some volumetric flaws. Probability-of-detection studies show PAUT achieving a90 sizes around 6.9 mm versus 12.7 mm for RT, but with a higher false-positive rate.

What are the limitations of ultrasonic weld inspection?

UT can be confused by geometric reflections, is sensitive to probe position and coupling, struggles with coarse-grained austenitic weld metals, and has reduced reliability very close to the scanning surface with TOFD. Base metal laminations and rough weld surfaces can also interfere with the sound path.

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

Conventional manual UT on a typical butt weld is faster than radiography because there is no film handling, exposure area evacuation or processing, and it can run alongside other shop work. Advanced methods such as phased array cost more in equipment and training but often pay back through scan speed, stored digital data and fewer re-inspections.

Is ultrasonic testing safe for inspectors?

It produces no ionizing radiation, so it is generally safer than radiography, but inspectors should still consider high-frequency airborne ultrasound and audible subharmonics, hot surfaces, couplant handling and confined-space hazards. Health Canada’s Safety Code 24 and OSHA’s adopted ultrasound TLVs give exposure guidance.