UT Weld Inspection: The Complete Guide to Ultrasonic Weld Testing

A 42-inch gas transmission line had passed every visual check and a full radiographic film review when an operator finally insisted on a second look with an angle-beam UT scanner. Buried under a clean-looking root pass sat a lack-of-fusion planar defect oriented so unfavorably that the sound beam of the radiographic exposure never caught it — and it was heading for a rupture in a populated area. That is the quiet power of UT weld inspection: it sees into the thickness of metal with sound waves, finding the flaws that eyes and X-rays can miss, and it is why pipelines, pressure vessels, and bridges cannot be built or maintained without it.

What Is UT Weld Inspection and Why Does It Matter?

It all started with a routine review of a 42-inch gas transmission line. Every visual check had passed, and a full radiographic film review had come back clean. Still, an operator insisted on one final look with an angle-beam ultrasonic scanner. Buried underneath an otherwise pristine root pass sat a lack-of-fusion, planar defect. Due to its orientation, the X-ray beam had never caught it at all. That one decision prevented a potential rupture in a populated area. This is the quiet, preventative power of UT weld inspection.

At its core, UT weld inspection is a nondestructive volumetric method that uses high-frequency sound to detect and size internal weld discontinuities without damaging the part. Rather than relying on sight or radiation, a transducer sends sound waves into the weld metal and reads the reflections that bounce back. By analyzing those echoes, a qualified inspector can locate, measure, and evaluate flaws that would otherwise remain hidden deep within the joint.

The Difference Between a Discontinuity and a Defect

Before anything can be rejected, it’s crucial to understand one of the most important distinctions in the trade: a discontinuity is not automatically a defect. A discontinuity is simply an interruption in the physical structure of the weld. A defect is a discontinuity that exceeds the acceptance limits of the referencing code and must be repaired.

This distinction is made explicitly within NASA’s PRC-6510 process specification for ultrasonic inspection of welds, which states a discontinuity is “not necessarily rejectable”. With that being said, acceptance is never up to opinion. It is always dictated by the governing code for that specific component.

Where UT Fits In The Inspection Ecosystem

No single nondestructive examination method can do it all. Each one has its place depending on the location and type of flaw that is being looked for. Here’s how UT stacks up against the others:

• Visual Testing (VT): The first line of defense. VT can only evaluate the weld face, toe, and surrounding surface condition.
• Magnetic Particle Testing (MT): Great for ferromagnetic materials to find surface or slightly subsurface discontinuities, however, depth capability is extremely limited.
• Liquid Penetrant Testing (PT): Ideal for non-porous materials to reveal open-to-surface flaws, but like MT, it cannot see below the surface.
• Radiographic Testing (RT): A volumetric method that uses radiation to project an image of the weld. It can find many internal flaws, however, it can struggle to detect tight, planar flaws oriented parallel to the radiation beam, just like the pipeline example above.
• Ultrasonic Testing (UT): A volumetric method that uses sound. Unlike MT and PT, UT interrogates the full thickness of the weld. According to the FHWA/NCHRP ultrasonic NDT manual hosted by the National Transportation Library, ultrasonic testing locates cracks, blowholes, insufficient penetration, lack of fusion, and other discontinuities in welds.

This depth of coverage is exactly why UT is so heavily relied on for critical joints, especially when safety is non-negotiable.

Why Volumetric Examination Matters For CJP Welds

For Complete Joint Penetration (CJP) welds, there is no un-welded land to fall back on. The weld is designed to carry load through the full thickness of the base metal. With that load transferring through the entire cross-section, a flaw could exist anywhere from the root, all the way up to the cap.

Surface examinations simply cannot verify this. Only a volumetric examination, such as UT or RT, can interrogate that full thickness. This level of scrutiny is the reason that ASME Section V was created, specifically to provide one consolidated document for nondestructive examination methods, including ultrasonic examination of welds, so that every section of the ASME Boiler and Pressure Vessel Code could reference the same requirements.

Industries That Depend On UT Weld Inspection

Due to its ability to find internal flaws without damaging components, UT weld inspection is utilized across countless high-consequence industries. These include oil and gas transmission pipelines, power generation (including nuclear and fossil fueled plants), petrochemical and refinery vessels, structural steel and bridge fabrication, shipbuilding and marine repairs such as those performed under AJ Marine Welding practices, pressure vessel manufacturing, and even specialized aluminum work such as alloy wheel welding.

In each of these fields, the question isn’t “if” a hidden flaw could form, it’s whether or not it will be caught before it sees service. With its volumetric coverage, portability, immediate results, and adaptability to virtually any weldable thickness, UT weld inspection remains one of the most trusted answers to that question.

How Ultrasonic Testing Works: The Physics Behind the Screen

Ultrasonic weld inspection is, at its core, an echo. The instrument sends a short pulse of high-frequency sound into the steel, waits, and listens. A flawless weld returns one clean reflection from the far surface. A discontinuity interrupts that path and throws energy back early, showing up as a spike on the A-scan before the back-wall echo arrives. Everything that follows — wedge angles, skip legs, DAC curves — exists to make that spike real, repeatable, and measured against something traceable.

The Transducer and the Piezoelectric Effect

Sound comes from a transducer: a small crystal bonded to a replaceable wear face inside a plastic wedge. Apply a brief electrical voltage and the crystal deforms mechanically, pushing against the steel. Strike it with returning sound and it produces a voltage the instrument displays as an amplitude. This two-way conversion is the piezoelectric effect, and it is why one probe both transmits and receives.

Frequency matters more than most beginners realize. Typical weld probes run at 2 MHz to 5 MHz. Lower frequency penetrates rough, attenuative material further but resolves small flaws poorly; higher frequency gives sharper resolution but scatters faster through coarse grain structure.

Shear Waves, Longitudinal Waves, and the Standard Angles

Straight-beam probes send longitudinal waves straight down through the thickness, which is useful for locating laminations and measuring thickness. Weld examination almost always uses shear waves, because shear energy travels at an angle across the weld volume and reflects off the bevel faces where fusion problems occur.

When the longitudinal wave in the wedge strikes the steel interface, it refracts according to Snell’s Law and mode-converted shear energy enters the part at the desired angle. In practice, technicians carry wedges cut for fixed angles of 45°, 60°, and 70°, with 70° being the angle most often specified for conventional angle-beam work because it covers the root and lower portion of common groove welds efficiently.

Skip Technique, Couplant, and Beam Geometry

That zig-zag is the skip technique. Sound travels from the probe to the opposite surface and reflects back: the first leg, or half-skip, ends at the first reflection; the second leg, or full-skip, ends when the beam returns to the same surface as the probe. Inspectors plan scan paths leg by leg so beam coverage overlaps across the entire weld volume, including the root and the cap.

None of this works without couplant. Air is a terrible transmitter of megahertz sound, so a gel, oil, or water-based couplant fills the microscopic gap between probe and steel. A thin, continuous film matters: dry patches and mill scale reflect sound unpredictably. Even on aluminum joints, surface condition drives reliability, which is why surface preparation deserves as much attention as the probe itself, as our aluminum cleaner for welding guide makes clear.

Calibration Blocks, DAC/TCG, and Reference Level

Before scanning production welds, the technician calibrates on a reference block of known geometry, typically with side-drilled holes at several depths. Because sound loses amplitude as it travels and spreads, the same-sized flaw produces a smaller echo when it is deeper. A Distance Amplitude Correction (DAC) curve plots the responses from those reference holes so indications of equal size can be compared fairly at any depth. TCG, or time-corrected gain, does the same job electronically by normalizing the screen so every reference hole sits at one amplitude, usually 80% of screen height.

Three gain settings build on each other:

• Reference level is the calibrated sensitivity established on the block, tied to a documented hole size and depth.
• Scanning gain adds a set number of decibels above reference (often +6 dB to +14 dB depending on the code) so nothing marginal slips past during the search.
• Evaluation level is the threshold at or above which an indication must be measured and judged.

Written procedures define these values precisely. NRC inspection procedure 57080 requires UT procedures to specify the ultrasonic frequency range, linearity and signal attenuation accuracy, beam angles, scanning rate, calibration block type, and the defined reference level used for monitoring discontinuities. Sensitivity also has a noise ceiling: NASA PRC-6510 requires that background noise not exceed 25% of the amplitude response from the applicable reference reflector during examination.

Why Shear-Wave Velocity Assumptions Matter

All distance measurements depend on an assumed velocity. Reference standards are generally machined from 1018 cold-rolled steel, and wedge manufacturers cut the angle assuming a shear-wave velocity of about 0.127 in/µs. Steel is not perfectly uniform, though. Rolling direction, grain orientation, and heat treatment all change how fast sound travels.

The practical consequence is sobering. FHWA research presented in 2024 showed that acoustic anisotropy in steel can shift the actual beam path, and that only a 3° change in beam angle can produce a 5 dB drop in the reference signal — enough under AWS D1.5 severity classes to make a Class A discontinuity read like a Class D. In other words, the geometry you calculate on paper is only as good as the velocity assumption behind it, which is why calibration on representative material and repeatable positioning are non-negotiable habits.

Codes and Standards That Govern UT Weld Inspection

A UT procedure is never a matter of personal preference. Every amplitude, every skip distance, and every accept-or-reject decision has to trace back to a written rulebook chosen before the first couplant is squeezed out of the tube. A pressure vessel, a highway overpass, a gas transmission line, and a reactor vessel head weld each fall under a different authority. Knowing that map is half the job of a welding or NDT inspector.

ASME BPVC Section V, Article 4: The Method Standard

In the ASME Boiler and Pressure Vessel Code world, Section V is the nondestructive examination section, and Article 4 is its ultrasonic chapter for welds. It sets out technique development, calibration, scan paths, and documentation. What it does not contain is acceptance criteria.

That omission is deliberate, and it catches newcomers every year. Section V tells you how to examine; it never tells you whether the weld is good enough. Acceptance criteria always come from the code that referenced Section V in the first place: Section VIII Division 1 for pressure vessels, Section III for nuclear components, Section XI for in-service inspection. A complete specification therefore reads like a two-part sentence: examination per Section V, Article 4, acceptance per Section VIII (or Section XI).

This shows up in real regulatory traffic. NRC correspondence concerning San Onofre Unit 3 cites ASME Section XI Subsection IWA-2232, which requires ultrasonic examination of reactor vessel welds in accordance with Section V, Article 4. The National Board has published a good historical overview of Section V and its evolution for background.

AWS D1.1 Clause 6: Structural Steel

For structural steel in the United States, the governing book is AWS D1.1/D1.1M, Structural Welding Code – Steel. Clause 6 addresses ultrasonic inspection directly: examination level and technique, calibration requirements, and the acceptance criteria themselves, including separate static and dynamic (severity-level) criteria used when the connection is subject to cyclic loading. Unlike Section V, everything you need to accept or reject sits in one code.

One boundary matters: D1.1 is a steel code. Structural aluminum falls under AWS D1.2, whose procedure variables differ enough to deserve their own study, as covered in Aluminum MIG Welding: The Complete Practical Guide. Industry-specific code families go further still; aerospace work lives under AWS D17.1 and AMS specifications, outlined in Aerospace Welding: Processes, Standards, and Careers Guide.

AWS D1.5: The Bridge Welding Code, Table 6.3, and Annex J

Bridge welds carry their own book: AWS D1.5M/D1.5 Bridge Welding Code. Its authority is not merely contractual. The Federal Highway Administration incorporates it by reference at 23 CFR 625.4(d)(2)(iii), which makes the D1.5 ultrasonic requirements federal requirements for bridges built with federal-aid highway funds.

The heart of D1.5 ultrasonic work is Table 6.3, which organizes acceptance into severity classes A through D. Class A is the most restrictive, applied to the most critical load-carrying members; Class D is the most permissive, applied where a flaw’s consequence is lowest. Each class sets its own amplitude limits (measured against the reference level established with a DAC curve) and its own length limit for a reportable indication. Reading that table quickly is a genuine field skill.

D1.5 also led the way on advanced methods. Annex J defines phased array ultrasonic testing procedures and acceptance for bridge welds, giving PAUT a codified seat alongside conventional angle-beam UT.

API 1104: Pipeline Welds

Pipeline girth welds are governed by API 1104, Welding of Pipelines and Related Facilities, currently in its 22nd Edition. API 1104 sets its own ultrasonic technique requirements and acceptance criteria for pipeline welds, and permits UT as supplementary to or, under stated conditions, in lieu of radiography, with tightly controlled procedure and operator qualification requirements. API 1104 is also one of the code books used in the AWS CWI Part C examination, alongside AWS D1.1, D1.2, D1.5, D15.1, and D17.1, per AWS certification information for CWIs.

ISO 17640: The International Standard

Work outside North America, or on international projects, usually points to ISO 17640, Non-destructive testing — Testing of welds — Ultrasonic testing of welds, most recently updated in 2018. It defines basic and enhanced examination levels and points to ISO 5817 (steel) or ISO 10042 (aluminum and castings) for acceptance quality levels. Its lineage matters because old drawings reference dead standards: ISO 17640 replaced EN 1714, which itself replaced DIN 54125 and BS 3923, as Sonatest’s reference material on weld ultrasonics documents.

Code	Typical Work	Where Acceptance Lives
ASME Section V, Article 4	Pressure vessels, nuclear, in-service	Referencing code (VIII, III, XI)
AWS D1.1 Clause 6	Structural steel	D1.1 itself
AWS D1.5 Table 6.3	Bridges	Severity classes A–D
API 1104	Pipelines	API 1104 itself
ISO 17640	International welds	ISO 5817 / ISO 10042

The practical takeaway is simple. Before you scan anything, confirm in writing which code governs, which edition applies, and which acceptance criteria control. When two documents conflict, the specification or the most restrictive requirement wins, and your procedure should say so on page one.

Who Can Perform UT Weld Inspection: CWI, NDT Levels, and Qualifications

Knowing how ultrasonic testing works is one thing; being signed off to make a call on a pressure vessel weld is another. In practice, two separate credential families govern who is allowed to stand at a weld with a transducer in hand, and they answer different questions. Welding inspector certifications from the American Welding Society say you understand welds, drawings, codes, and procedures. Nondestructive testing (NDT) certifications say you understand the specific method — in this case ultrasonics — well enough to operate equipment, evaluate indications, and issue a report. On most serious work, the person running the UT scan needs both a working knowledge of the applicable welding code and a method-level NDT certificate.

The AWS Welding Inspector Ladder: CAWI, CWI, and SCWI

The American Welding Society offers three certification levels under AWS QC1, currently administered as AWS QC1:2016-AMD1: Certified Associate Welding Inspector (CAWI), Certified Welding Inspector (CWI), and Senior Certified Welding Inspector (SCWI). The CWI sits in the middle and is the one most employers ask for by name.

Earning a CWI means passing three separate exams, each scored independently: Part A (Fundamentals), a 150-question written exam on welding processes, metallurgy, and inspection theory; Part B (Practical), a hands-on 46-question exercise using gauges, visual samples, and weld measurements; and Part C, an open-book exam on a single reference code book you select in advance, such as AWS D1.1 or ASME Section IX. Candidates must achieve a minimum score of 72% on each part, according to the official AWS CWI program page. There is no averaging across parts — a strong code score does not rescue a weak practical result.

Once certified, a CWI is valid for nine years, with renewal applications due at years three and six to keep the certification active. AWS also offers optional endorsements, and the AWS D1.1 Structural Welding Code endorsement is a common one: it requires its own 72% pass and splits the exam evenly, with 25% of questions each drawn from the Qualification, Fabrication, and Inspection sections of the code.

NDT Method Certification: ASNT Level I, II, and III

The NDT side follows the personnel qualification framework published by the American Society for Nondestructive Testing, most commonly ASNT Recommended Practice SNT-TC-1A and ASNT CP-189, both applied through an employer-written practice that defines training hours, experience, and examinations for each level.

• Level I works under direct supervision: calibrates equipment as instructed, scans, and records indications without evaluating them.
• Level II selects techniques, calibrates independently, interprets and evaluates results against acceptance criteria, and can train and supervise Level I personnel.
• Level III writes and approves procedures, establishes techniques and acceptance criteria, and supervises the program.

The key threshold for weld UT is that Level II is the minimum for making accept/reject decisions, which is exactly how NASA Procedure PRC-6510 states it. If you are a welder or QC lead hoping to sign off UT results in-house, that is the certification target to aim for.

Underneath every level sits a practical requirement people often overlook: documented visual acuity. Both AWS and most ASNT-based employer practices require a current near- and far-vision screening, because reading a small A-scan trace, spotting a hairline surface crack, and scanning dense code tables are all visual tasks.

Performance Qualification: D1.8 and the PAUT Path

Some codes demand proof beyond written exams. AWS D1.8, the Structural Welding Code Seismic Supplement, defines a UT technician performance qualification using at least 20 known indications distributed across multiple test specimens, with results confirmed by two ASNT Level III inspectors. It is a realistic blind test: if you cannot find and size what is actually in the plate, the certificate does not matter.

Phased array adds another layer. Candidates for PAUT Level II must already hold UT Level II certification and then complete an additional 80 hours of training plus 320 method-specific hours before sitting the exam, a pathway outlined in ASNT’s phased array UT weld inspection guidance. That stacking requirement is why PAUT technicians are scarcer and better paid than conventional UT operators.

For anyone building a career here, formal training shortens the climb considerably, whether that is a structured aluminium welding course that starts you on the shop floor or an NDT certificate program that puts you behind the scanner. Either way, the credential only opens the door: the acceptance decisions still have to be right, every time.

Step-by-Step: The UT Weld Inspection Process on the Job

Everything up to this point is classroom work. On the job, UT weld inspection runs as a fixed sequence: paperwork, surface, calibration, then scanning. Get the order wrong and you get a plausible-looking screen full of echoes that mean nothing. Here is the routine a technician or QC lead actually walks through on a weld map.

Step 1: Read Before You Probe

No competent scan starts at the weld. It starts with the welding procedure specification (WPS), the supporting procedure qualification record (PQR), the fabrication drawing, and the inspection and test plan (ITP), which tells you which welds get 100% examination and which get a sampling percentage. You are hunting for material type, thickness, joint geometry, and the governing acceptance code. NRC procedure 57080 directs inspectors to verify that the written UT procedure specifies the apparatus type, search unit sizes and frequencies, beam angles, calibration requirements, and scanning technique consistent with the applicable code. If the procedure in your hands does not say those things, stop and get one that does.

Step 2: Prepare the Surface and Pick the Couplant

Sound does not jump gaps. Remove spatter, grinding burrs, and heavy mill scale from the scan path, and make sure the probe can travel smoothly. On stainless and aluminum work, surface chemistry drives the cleaning choice, and a dedicated prep such as Aluminum Cleaner for Welding shows how specific those routines get. Then select couplant to match the job: water-soluble gels for quick indoor work, oil-based or high-temperature couplants for hot or outdoor scans. A thin, continuous film is the goal. Dry spots show up as sudden signal loss, a ghost beginners chase constantly.

Step 3: Choose the Search Unit

Frequency, crystal size, and angle are your three decisions. Common practice lands on 2.25 MHz for general steel welds, stepping up to 4 or 5 MHz on thinner material where finer resolution pays, and dropping lower on coarse-grained or highly attenuative alloys. Angle follows geometry: 70° probes work well on thin plate because the skip distance stays short, 60° covers mid-range work, and 45° reaches deep into thick sections. Pick the angle that puts the beam square into the expected flaw face, not the one that is easiest to hold. Material-specific procedures deserve the same care as filler and heat-input selection in an Alloy 20 Welding job.

Step 4: Calibrate and Establish DAC or TCG

On the reference block, establish sound velocity, probe zero (index point), and sensitivity using reflectors at known depths. Then build your distance amplitude correction curve, or its electronic equivalent, time-corrected gain, so identical flaws at different depths read at comparable heights. The NRC procedure specifically requires verification that DAC methods or TCG mechanisms are specified and that transfer-compensation reference reflectors are defined, meaning you correct for signal loss through the actual part, not just the block. Recheck calibration at intervals through the shift and after any cable or probe handling; a drifting zero mid-job invalidates everything after it.

Step 5: Scan With a Plan, Not a Habit

Mark index lines on the surface, then run overlapping passes from at least two directions. NASA’s PRC-6510 technique requirement is a useful benchmark: coverage of the entire weld bead and heat-affected zone, extending a minimum of ¼ inch beyond the weld bead on each side. Keep probe motion slow enough that the operator can read a rising echo; roughly 100 millimeters per second is a common ceiling. The Sonatest ultrasonic weld inspection guide walks through standard skip sequences for butt, T, and lap joints if you want a refresher on pass patterns.

Step 6: Mark, Measure, and Evaluate

Every reportable indication gets marked on the part and sketched on the weld map. Measure its length by a defined technique, usually the 6 dB drop method as the probe passes each end, and record peak amplitude, depth, and position. ASME Section V, Article 4 requires dimensional evaluation of indications, so “I saw something” is not a finding; length, amplitude severity, and location are. Evaluate against the acceptance criteria in the referencing code, never against intuition.

Step 7: Report, Repair, and Re-Examine

Close with documentation: procedure number, equipment and serial numbers, calibration data, a scan coverage sketch, and the disposition of every indication. Rejected welds follow a defined repair cycle: excavate, examine the open cavity by MT or PT, re-weld to the repair procedure, then re-examine the completed weld by UT. Re-scanning the repaired area from scratch, rather than re-reading old marks, is what actually closes the loop.

What UT Finds: Common Weld Discontinuities and Acceptance Decisions

A UT weld inspection doesn’t hand you a simple pass or fail. It maps what sits inside the joint — shape, depth, orientation, and echo strength — and puts that picture in front of someone who has to make a judgment call. That call starts with recognizing the signature each discontinuity leaves on the A-scan, and it ends with comparing what you found against the acceptance criteria written into the referencing code.

The Major Discontinuities UT Weld Inspection Detects

Every common weld discontinuity behaves differently under the beam. Learning those behaviors is half the inspector’s job.

Discontinuity	Typical A-scan Behavior	Key Note
Crack	Sharp, high-amplitude, highly directional	May disappear from certain probe angles entirely
Lack of side-wall fusion	Flat planar reflector; strong from one direction, weak from another	Often hides along the fusion line
Lack of root fusion / incomplete penetration	Steady reflector at predictable root depth	Common in single-V joints with poor root prep
Slag or oxide inclusion	Rounded or irregular, lower amplitude, less directional	Multi-pass SMAW and FCAW welds are prone
Porosity	Cluster of small, round, repeating signals	Often called “pea signals” by technicians
Delamination	Strong planar echo at mid-thickness across a wide area	Plate rolling defect, not a weld defect per se
Undercut / misalignment	Geometric reflector at weld toe or root line	Strong but legitimate; must be distinguished from true flaws

Sonatest’s ultrasonic inspection of welds handbook puts it plainly: lack of fusion can occur anywhere in a weld, but root and side-wall fusion defects are the most common types found in practice. The handbook also warns that flat, fusion-face reflectors demand an inspection approach specifically designed to detect them — meaning probe angle selected for the joint geometry, not a default 70° pulled off the cart. This matters especially in stick welding, where slag entrapment between passes can either mimic or mask fusion problems; the process trade-offs in Advantages of Stick Welding explain why SMAW welds carry distinct defect profiles.

The Root Zone Pitfall: Fusion Defect or Crack?

The most uncomfortable call in weld UT happens at the root. A lack of root fusion and a root crack can look nearly identical on screen — similar amplitude, similar depth, similar refusal to respond from certain angles. Sonatest’s guidance is blunt: when you cannot confidently distinguish the two, you assess them together conservatively as a “root defect.” A crack is rejectable under virtually every code; a lack-of-fusion indication at the root usually is as well. Guessing wrong on a pressure boundary is not an option, so the safe call is the strict call.

How Acceptance Criteria Combine Amplitude, Length, and Location

Weld UT acceptance criteria almost never rely on echo height alone. They combine three variables:

• Amplitude — signal strength relative to a calibrated reference, expressed as a percentage of DAC or a severity class
• Length — the linear extent of the indication measured during scanning
• Location — depth within the thickness and position relative to the weld centerline and critical zones like the root or fusion line

Under AWS D1.5, reflectors are sorted into severity classes A through D. FHWA reports that Class A reflectors are “less severe than a 1/16 in. side-drilled hole,” while Class D reflectors are “more severe.” The critical detail: at most 5 dB separates automatic acceptance from rejection. That margin is razor thin, and a calibration drift of just a few decibels can reclassify an indication from acceptable to rejectable without anyone noticing the shift.

When the Material Itself Works Against You

Even a technically correct reading can mislead. FHWA research into acoustic anisotropy in rolled steel plate found that shear-wave velocity changes depending on whether the beam travels along or across the rolling direction. That shift alters an indication’s apparent depth and, by extension, its severity classification under the D1.5 table. An inspector who assumes a single shear-wave velocity on heavy rolled plate can place a reflector at the wrong depth and miscategorize it entirely.

Sonatest’s full ultrasonic inspection of welds guide covers each discontinuity type in greater depth, including scan patterns optimized specifically for flat planar reflectors — the ones most likely to slip past a technique that wasn’t designed with them in mind.

UT vs. Radiography and Other NDE Methods: Choosing the Right Examination

No single NDE method wins every job. Ultrasonic testing excels at looking through the thickness of a weld for planar flaws, but a shop floor deciding between UT and radiography, magnetic particle, or liquid penetrant is really deciding what kind of answer the code, the schedule, and the budget need. Understanding how ultrasonic weld inspection stacks up against the alternatives is what lets a QC lead specify an examination that actually passes the first time.

A Quick Comparison of the Main Weld NDE Methods

Method	What it finds	Access / thickness	Record
UT (angle beam)	Cracks, lack of fusion, incomplete penetration, slag, some porosity	One side; thin to very thick	Traceable numeric data
RT (film or digital)	Porosity, slag, lack of fusion when oriented favorably to the beam	Both sides or source placement; thickness range set by energy	Permanent image
MT	Surface and near-surface linear discontinuities	Magnetic materials only, single side	Photos, mapped indications
PT	Surface-breaking flaws, nonporous materials	One side, any alloy	Photos, mapped indications
VT	Profile, undercut, misalignment, visible cracks	Line of sight, any thickness	Reports, photos
Eddy current	Surface cracks, tube and heat-exchanger flaws	Conductive materials, surface probe	Logged signal data

All of these live together in one referenced document: ASME Boiler and Pressure Vessel Code Section V consolidates radiography in Article 2, ultrasonics in Article 4, penetrant in Article 6, magnetic particle in Article 7, eddy current in Article 8, and visual examination in Article 9. That consolidation is deliberate: Section V tells you how to examine, while the referencing code — Section VIII, Section III, Section XI, B31.3 — decides which method and what acceptance criteria apply.

The “UT in lieu of RT” Movement

Nuclear and pipeline work has pushed hardest to substitute UT for radiography, mainly to eliminate radiation logistics. Under ASME Section XI, Code Case N-659 established ultrasonic examination in lieu of radiography for weld inspections, but only for organizations that have passed a performance demonstration — a qualification process proving the procedure, equipment, and personnel can detect and size flaws reliably before they are trusted on real hardware.

The evidence behind that shift is substantial. EPRI research presented to the NRC reports that automated UT detected all flaws greater than 0.4 in. long in performance demonstrations, with all cracks and incomplete-penetration flaws detected, while porosity was the most difficult discontinuity for UT to detect. The same body of work found UT could be effectively substituted for RT in CANDU feeder-tube inspection, with personnel length-sizing error reduced to an RMS of 0.25 in. (6.4 mm) during performance demonstrations (NRC documentation).

What Radiography Still Does Better

RT keeps three real advantages. First, porosity and rounded indications show up on film naturally — UT operators must chase small, scattered reflectors that can be confused with metallurgical noise. Second, film or a digital image is a permanent, language-independent record that any reviewer can re-read years later without equipment or calibration context. Third, radiography is comparatively indifferent to flaw orientation when the source is positioned correctly, where UT is at its weakest for flaws lying parallel to the beam.

UT’s counterarguments are equally strong: no radiation permits, no evacuation zones, no darkrooms or film handling, and examination can proceed continuously with people working around the area because the energy is non-ionizing sound plus couplant. For thick-wall pressure vessels, heavy wall piping, and field joints where craning a source into position is impractical, UT is simply faster and cheaper. Surface methods add their own niche: transportation and pipeline standards routinely combine visual, MT or PT for surface-breaking flaws, and UT or RT for volumetric examination, because no one method covers everything.

How to Choose: Let the Code Lead

1. Read the referencing code first. D1.1, D1.5, API 1104, Section VIII, and ISO 17640 each specify permitted methods and their acceptance criteria.
2. Match the flaw type. Suspect planar lack of fusion or cracking on a CJP butt weld: UT. Suspect rounded porosity requiring a picture: RT.
3. Check access and thickness. One-sided access, heavy sections, or restricted space favor UT; thin sheet with both sides reachable may favor RT.
4. Factor logistics. Radiation permits, evacuation, and shift shutdowns often cost more than the examination itself.
5. Confirm personnel qualification. A method is only as good as the certified technician running it.

The trend line points toward UT and its advanced variants — phased array and TOFD — displacing film radiography in structural, nuclear, and pipeline work. But the smart specifier never picks a method out of habit. They pick the one the code allows, the flaws demand, and the site can support, then document the reasoning behind the choice.

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

A conventional angle-beam search unit does one job at a time. It fires a shear wave into the weld at a single refracted angle, the operator reads one A-scan trace, and full coverage comes from physically repositioning the probe again and again. It works, but the accuracy of every flaw length and depth call rests heavily on hand speed and interpretation skill. Two advanced methods have spent the last two decades changing that equation: phased array ultrasonic testing (PAUT) and time-of-flight diffraction (TOFD).

Phased Array UT: Steering the Beam Without Moving the Probe

Where a conventional transducer is one crystal, a phased array probe is a row of dozens of small, independently fired elements. By inserting precise time delays between firing each element, the system produces and receives waves at various angles and focal lengths, sweeping a sector of, say, 40 to 75 degrees electronically, in fractions of a second, with the probe sitting still. That allows more precise detection and sizing than conventional fixed-angle UT, because the same volume of weld is interrogated from many angles instead of one.

The output is what makes phased array intuitive. A sectorial scan, usually written S-scan or S-scope, displays the swept beam as a cross-sectional slice through the weld, color-coded by signal amplitude, alongside the familiar A-scan. Instead of imagining where a 70-degree ray is heading, the inspector sees the root, cap, sidewall and heat-affected zone drawn on screen with the indication sitting in place.

Mounted on an encoded scanner, manual or motorized, with wheels that track position, PAUT records not just the echo but where it came from. That produces C-scan and strip-chart maps of the entire weld, data that can be re-analyzed months later, reviewed by a Level III, or pulled out during a dispute about whether a flaw was really there. It is also why PAUT is now standard practice in aerospace welding, where high-value joints leave little room for rework.

TOFD: Sizing Flaws by Diffraction, Not Amplitude

Time-of-flight diffraction takes a different physical approach. A transmitter and a receiver straddle the weld, sending longitudinal waves through it. When the beam strikes the tip of a crack, that tip diffracts sound in many directions, and the receiver picks up those arrivals. Because sizing depends on the time it takes sound to travel from tip to receiver, depth accuracy is largely independent of flaw orientation and reflectivity, which is exactly where amplitude-based sizing is weakest. In practice TOFD is normally paired with pulse-echo, because TOFD alone has blind zones near the surfaces and can miss planar flaws running parallel to the beam.

Code Recognition: Annex J and Code Case N-713

Advanced UT is only useful if the governing code accepts it, and the acceptance landscape has caught up. AWS D1.5M/D1.5 Annex J defines PAUT acceptance classes ranging from Class A, above the acceptance reference level, through Class D, below the detection reference level, each carrying its own permitted length. For nuclear inservice inspection, the proposed ASME Section XI Code Case N-713 requires recording encoded position and UT data and sets demanding demonstration bars: personnel and procedure test sets of at least 10 and 30 flaws respectively, as documented in EPRI and NRC submittals on the case.

Benefits, Trade-Offs, and the Trend Toward Replacing RT

The benefits are concrete: better flaw characterization, repeatability, digital archiving, faster scanning with less probe repositioning, and no radiation boundary, no film, and no developing chemicals. That last point drives much of the “UT in lieu of RT” momentum in pipeline and nuclear work.

The trade-offs are equally real:

• Equipment cost. A PAUT/TOFD system runs several times the price of a conventional flaw detector, plus scanners, analysis software, and data storage.
• Procedure complexity. Everything from wedge design to focal law setup must be written, validated, and controlled under a written procedure.
• Personnel depth. A conventional Level II does not automatically become a PAUT Level II; the qualification adds array theory, imaging interpretation, and code-specific performance demonstration.
• Frequency selection. Higher PAUT frequencies improve resolution but are more susceptible to attenuation, so frequency must be matched to the material and geometry. Coarse-grained or thick sections often force a compromise between near-surface detail and penetration.

None of this makes conventional UT obsolete. But for CJP welds where planar flaws matter most, phased array and TOFD increasingly set the standard, and the inspectors who can read an S-scan and defend a written scan plan are the ones getting the work.

Common UT Weld Inspection Mistakes, Limitations, and Practical Tips

Even a certified technician with a top-tier flaw detector can walk past a crack. In ultrasonic testing the instrument rarely fails first; the procedure, the preparation, or the scanning habit does. The same handful of mistakes shows up in nearly every rejected examination, rework loop, and “how did we miss that” post-mortem.

The Field-Tested Failure Modes

Poor surface preparation and a thin or contaminated couplant film. Rough weld crowns, mill scale, spatter, and dried-out couplant break the acoustic path between probe and part. The result is a weak, noisy signal that looks like a small flaw or vanishes entirely. Cleaning and grinding the scan path is unglamorous work, but it is the cheapest insurance on the job, and it matters even more on non-ferrous surfaces where preparation quality drives everything, as our aluminum cleaning and prep guide explains.

The wrong probe angle for the flaw’s orientation. A 70° shear beam scanning a planar lack-of-fusion defect sitting parallel to the weld centerline may return almost nothing, while a 45° probe finds it immediately. Technicians who default to one favorite angle for every joint are guessing.

Coverage gaps at the root and cap. Unscanned strips along the weld toes, the ragged root region, and the change-of-thickness zones are where cracks start. If the scan plan does not demonstrably cover them, they will not be examined.

Lateral crosstalk and geometric false calls. Runout grooves, ladder brackets, and beam scatter produce repeatable indications that are not discontinuities. Without cancelling or characterizing them, crews waste repair time on sound metal and start distrusting real signals.

Worn transducers, damaged cables, and drifting calibration. A degraded element or a cable that cuts out mid-scan silently lowers sensitivity. Calibration checked only at the start of an eight-hour shift is not calibration; drift happens during the job.

Inadequate documentation. No scan plan, no gain record, no sketch of indication locations, and the examination cannot be reproduced, defended, or audited.

Limitations You Have To Plan Around

UT is powerful, not magic. That same DOT/FHWA manual identifies sensitivity, resolution, noise discrimination, geometry, defect orientation, grain size, and inclusions as inherent UT limitations. Coarse-grained or highly attenuative materials scatter sound; complex geometry floods the screen with reflections; small rounded porosity is RT’s territory more than UT’s. The answer is not to avoid UT — it is to design the examination around the limitation, choosing frequency, angle, and reference level to suit the material and joint rather than the technician’s habit.

Practical Tips That Raise Reliability

• Scan a mock-up first. A known-discontinuity practice plate exposes technique gaps before production welds are on the line.
• Rescan from the opposite side and with a second angle. Orientation-dependent flaws give themselves away when the result changes with direction.
• Test repeatability. Move off an indication, come back, and confirm it reappears at the same depth, location, and amplitude. Anything that will not repeat rarely deserves a repair call.
• Keep gain settings traceable. Record reference level, DAC/TCG curve, and any dB change with the reason for it, on the report.
• Verify scanning parameters. NRC inspection procedure 57080 directs inspectors to confirm that scanning rates, scan index, and scanning directions are consistent with the applicable code, because scanning speed directly affects detection reliability. Fast scanning over a wide index spacing is the most common silent cause of missed flaws.
• Re-verify calibration at defined intervals and inspect probes, cables, and couplant delivery at the end of every shift.

Career, Safety, and Where UT Is Heading

Technique is only half the qualification. AWS welding inspector certification has been running since the first exams in spring and fall 1976, and the AWS QC1 2006 edition noted that over 55,000 individuals worldwide had met AWS welding inspector certification requirements by then. Today examinations are delivered through computer-based testing at more than 450 Prometric centers globally, with the practical Part B held at AWS seminar locations — details are maintained on the official AWS CWI program page. Layer NDT Level II or III credentials on top and the career ceiling rises with it; structured training paths, such as those in our aluminium welding course overview, show what classroom and hands-on hours realistically cost. On the safety side, treat couplant hygiene, wet-floor and working-at-height risks, ladder discipline, and ultrasonic output exposure checks as part of the procedure, not an afterthought.

None of this changes what UT weld inspection is: a nondestructive, volumetric window into the thickness of a weld, and the routine method best suited to finding and sizing a planar flaw buried under a clean-looking root pass. Misses happen when preparation, scan planning, and documentation slip — all controllable things. Control them, and UT keeps doing the job that eyes and X-rays cannot: catching the defect heading for a rupture before the structure ever feels the load.

Frequently Asked Questions

What is UT weld inspection used for?

UT weld inspection uses high-frequency sound waves to detect, locate, and size internal weld discontinuities such as cracks, lack of fusion, incomplete penetration, slag inclusions, and porosity without damaging the weld. It is a volumetric method, so it examines the full thickness of the weld and heat-affected zone rather than just the surface.

Which codes govern ultrasonic testing of welds?

In ASME work, Section V, Article 4 defines the examination methodology, while acceptance criteria come from the referencing section such as Section VIII or Section XI. Structural steel work follows AWS D1.1 Clause 6, bridges follow AWS D1.5 (incorporated by reference at 23 CFR 625.4), pipelines follow API 1104, and international work commonly uses ISO 17640.

What qualifications does a UT weld inspector need?

Welding inspection is typically covered by AWS certifications (CAWI, CWI, or SCWI under AWS QC1:2016-AMD1), while the UT method itself is qualified under an employer’s written practice based on ASNT SNT-TC-1A or ASNT CP-189, usually to Level II to make accept/reject decisions. Many roles require both a CWI and an NDT Level II in UT.

How is UT different from radiographic testing for welds?

UT uses sound waves and works well on thick sections, needs no radiation permits, and can size flaw depth; RT uses ionizing radiation and produces a permanent film record and detects porosity very reliably. Research by EPRI has demonstrated that performance-qualified automated UT can effectively substitute for RT in many nuclear and pipeline applications.

What weld defects can ultrasonic testing find?

UT reliably detects planar, crack-like discontinuities such as cracks, lack of side-wall fusion, lack of root fusion, and incomplete penetration, plus volumetric indications like slag inclusions, oxide inclusions, and larger porosity. Small, rounded porosity is generally the hardest discontinuity type for UT to detect.

What is phased array ultrasonic testing (PAUT) of welds?

PAUT uses an array of elements fired in timed sequences to steer and focus the ultrasonic beam electronically, sweeping through many angles in a single pass and producing real-time cross-sectional images. It improves flaw characterization and sizing but requires more expensive equipment and additional qualification, such as the 80-hour training and 320 method hours required for PAUT Level II on top of UT Level II.

How are UT weld indications accepted or rejected?

Acceptance depends on the referencing code, and typically combines amplitude (compared to a DAC curve or severity class such as AWS D1.5 Classes A–D), measured length, and location in the weld. Cracks and crack-like indications are rejected under essentially every code, while porosity and inclusions may be acceptable within defined limits.

What is the most common mistake in UT weld inspection?

Improper scan planning that fails to aim sound reflectors back at the transducer — a probe can illuminate a sidewall lack-of-fusion face yet return no echo if the geometry is wrong. Poor surface preparation, an unsuitable beam angle, and drifting calibration during long shifts are close seconds, which is why codes require defined scanning techniques, rates, and calibration checks.