What Are the Main Types of Ultrasonic Testing in NDT?

The main types of ultrasonic testing in NDT are straight-beam (0°) and angle-beam shear wave contact testing, phased array (PAUT), time-of-flight diffraction (TOFD), thickness gauging, immersion and guided wave testing. For structural welds under AWS D1.1 and rail under 49 CFR 213.237, 0° probes and 45°, 60° or 70° shear wave probes still do most of the work.

EMAT, IRIS and laser ultrasonics are real techniques too, but each one serves a narrow job. This guide sorts the techniques three ways (by coupling, by wave mode and by display), gives each one's typical use, governing document and limits in one table, and then picks the technique for three real jobs: a 1 in. butt weld under AWS D1.1, rail in track, and corrosion mapping.

It's written for rail owners and for structural and pipe QA managers who have to specify, buy or review ultrasonic work. Where EMA already has a deep-dive page on a technique, this guide gives a one-paragraph summary and links to it.

How are the types of ultrasonic testing classified?

Ultrasonic techniques are usually grouped three ways: by how the sound gets into the part (contact, immersion or non-contact), by the wave mode it travels in, and by how the results are displayed. The same job can sit in all three groups at once. A rail test car, for example, uses liquid-coupled contact probes, sends both longitudinal and shear waves, and shows the results as a B-scan.

Key takeaway: name the coupling, the wave mode and the display, and you have described the technique more precisely than any brand name does.

By coupling: contact, immersion or non-contact

A separate distinction is pulse-echo (one probe, or a transmit-receive pair on the same side, listens for reflections) versus through-transmission (a probe on each side, and a flaw shows as lost signal). Nearly all field work on welds and rail is pulse-echo, because you rarely have access to both sides. Through-transmission is mostly used on composites and bonded panels.

By wave mode

• Longitudinal (compression) waves travel straight in from a 0° probe. In steel they move at about 5,890 m/s (Evident, steel 1020).
• Shear (transverse) waves come from an angled wedge that mode-converts the beam at the surface. In steel they travel at about 3,250 m/s, at refracted angles that are usually 45°, 60° or 70°.
• Surface (Rayleigh) waves travel along the surface at about 3,000 m/s in steel and reach about one wavelength deep.
• Guided waves fill the whole wall of a pipe or plate and travel along it, at low frequencies (ASTM E2929 gives 10 to about 250 kHz for its method).

The physics of why each mode behaves the way it does (Snell's law, mode conversion, near field and attenuation) is covered in EMA's explainer on the science behind ultrasonic testing, so it isn't repeated here.

By display: A-scan, B-scan, C-scan and S-scan

A conventional angle-beam test shows only an A-scan. A picture of the weld or the wall needs either a phased array probe or an encoder that records probe position, and the image is only as good as the setup behind it.

Which of the types of ultrasonic testing fits which job?

The workhorses are straight-beam and angle-beam contact UT; PAUT and TOFD add imaging and sizing; thickness gauging and guided wave handle wall loss; and EMAT, IRIS and laser UT are specialist tools. The table gives the governing document for each, but codes change by edition, so check the edition named in your contract.

Sources: the ASTM store scope pages listed at the end; Evident velocity data; KB entries for AWS D1.1 (paywalled, hedged).

What are straight-beam and angle-beam testing, and why are they the workhorses?

Straight-beam (0°) testing sends a longitudinal wave straight into the part to find flaws parallel to the surface and to measure thickness; angle-beam testing sends a shear wave in at 45°, 60° or 70° to find flaws that lie across the beam path, such as weld fusion faces and transverse cracks. Between them they cover most structural weld UT and most rail testing.

Key takeaway: the probe angle has to suit the flaw's orientation; a 0° probe can't find a vertical crack, and an angle probe isn't the tool for laminations.

Straight beam (0°)

A 0° probe sees reflectors that lie roughly parallel to the scanning surface: laminations, horizontal split heads in rail, and the back wall. Loss of the back-wall echo is itself a signal, because something between the probe and the far surface is blocking the sound. AWS D1.1 users know straight beam as the base metal scan for laminations that is run before angle-beam testing of a CJP weld (AWS forum quotations of earlier editions; confirm the clause in your edition).

Angle beam shear wave

An angle-beam probe uses a plastic wedge to refract the beam into a shear wave. Shear waves have about half the wavelength of longitudinal waves at the same frequency, so they resolve smaller reflectors. AWS D1.1 Clause 8 requires search units within ±2° of 70°, 60° or 45° and sets the angle by thickness and weld zone (quoted from forum reproductions of the code; CONFIRM).

EMA's deep-dive on angle beam testing in UT covers refraction angles, wedge choice and scan patterns in detail. For the wave physics and how shear wave probes are calibrated, see shear wave ultrasonic testing.

In rail

Rail flaw testing uses both at once. A typical test car or walking cart carries 70° probes for transverse defects in the head (transverse fissures and detail fractures), 0° probes for horizontal and vertical split heads and loss of back wall, and probes near 37–45° for the web and bolt holes. The exact beam set varies by system. Rail testing has to meet the 49 CFR 213.237 definition of a valid search:

"Valid search means a continuous inspection for internal rail defects where the equipment performs as intended and equipment responses are interpreted by a qualified operator as defined in § 213.238." — 49 CFR 213.237(j)(4), eCFR

What is the difference between PAUT and TOFD?

PAUT steers and focuses a beam from many small elements to build an image of the weld, while TOFD uses a pair of probes to time the weak signals that diffract from flaw tips and so measure a flaw's through-wall height. They answer different questions, which is why the two are often run together.

Key takeaway: PAUT is best at detection and imaging, TOFD at height sizing, and neither removes the need for a trained analyst.

Phased array (PAUT)

A phased array probe fires its elements with small time delays (focal laws) to sweep a sector of angles (an S-scan) or step a beam along the array (an E-scan). Encoded PAUT records the data with the probe position, so a weld can be reviewed later and audited. The probe still travels along the weld; electronic steering replaces the side-to-side raster, not the scan.

ASTM E2700 covers contact PAUT of welds from 9 to 200 mm and warns that "extreme caution should be used" when sizing indications with phased array, because beam spread can make flaws look larger. EMA's comparison of UT and PAUT works through S-scan coverage on a 1 in. weld and when PAUT is worth its higher setup cost.

Time-of-flight diffraction (TOFD)

TOFD places a transmitter and a receiver on opposite sides of the weld. A lateral wave runs just under the surface, a back-wall echo comes from the far side, and any flaw tips between them diffract sound that arrives in between. Depth comes from diffraction geometry, d = √[(ct/2)² − S²], where S is half the probe center spacing, not from the pulse-echo formula. ASTM E2373 states the key difference:

"TOFD is a nondestructive ultrasonic examination technique that is not based on amplitude response." — ASTM E2373/E2373M-19(2024), section 4.2

TOFD is not operator-proof. Not depending on amplitude makes TOFD good at sizing height, but it doesn't take the operator out of the result. The image still has to be interpreted: an analyst must tell flaw tips from geometry, surface-breaking from embedded flaws, and real indications from noise. Baker Hughes notes that only "an experienced inspector" can judge with reasonable certainty whether a flaw is embedded or open to a surface.

TOFD also has two dead zones. The lateral wave hides flaws just below the scanning surface, a band of roughly 4 to 5.5 mm at a 50 mm probe center spacing in one vendor's example (it depends on frequency and spacing). The back-wall echo can hide small flaws at the far surface, a zone BS EN ISO 16828 defines. For this reason TOFD is normally paired with pulse-echo or PAUT. ASTM E2373 also describes TOFD as a supplement to E164 and E1961 weld examinations, used to "improve sizing estimates".

TOFD has a narrow role in rail. It is a weld-sizing technique for plate and pipe of 9 to 300 mm, not a rail-in-track method, and claims that it is the answer for rail welds overstate it. EMA's page on PAUT vs TOFD for rail weld inspection covers where each one helps.

What are thickness gauging, immersion and guided wave testing used for?

Thickness gauging measures remaining wall from one side; immersion testing gives repeatable coupling for parts that can go in a tank; and guided wave testing screens long lengths of pipe from one probe position. All three are mostly about wall loss and material quality, not weld flaws.

Key takeaway: a thickness gauge measures, guided wave screens, and immersion is a shop method, not a field one.

Thickness gauging

A thickness gauge times the round trip of a 0° longitudinal pulse and computes T = V × t / 2. ASTM E797 covers manual contact pulse-echo thickness measurement up to 93 °C (200 °F) on material with a constant velocity and a resolvable back wall. Dual-element probes are the corrosion-survey standard because their crossed beams pick up pit bottoms better.

The traps are velocity calibration, coatings (read at steel velocity unless the gauge has an echo-to-echo or through-coat mode) and hot surfaces. EMA's guide to ultrasonic thickness measurement methods covers gauge modes, two-point calibration and coating errors.

Immersion testing

In immersion testing the part sits in water, or under a water column, and the probe scans at a fixed water path. Coupling is uniform and the scan can be automated into a C-scan, which is why bar, plate, forgings and aerospace composites are tested this way. ASTM E1001 covers immersed pulse-echo testing with longitudinal waves and notes that it detects discontinuities "at normal beam incidence"; other orientations need other scanning techniques. Structural steel and rail in track can't go in a tank, so immersion matters to EMA's readers mainly as a mill and supplier test.

Guided wave testing

Guided wave testing (GWT) clamps a ring of transducers around a pipe and sends low-frequency waves along it, reporting changes in cross-section such as corrosion, welds and supports. Guided wave and long-range UT are the same method under two names; ASTM E2775 says GWT was "previously known as long range ultrasonic testing (LRUT)". The standard is blunt about its role:

"GWT is a screening tool. The method does not provide a direct measurement of wall thickness or the exact dimensions of defects/defected area." — ASTM E2775-16(2023), section 1.3

ASTM E2775 covers carbon and low-alloy steel pipe of NPS 2 to 48 with walls of 0.15 to 1 in. Anything it flags needs follow-up with thickness gauging or another method.

Where are EMAT, IRIS and laser UT used?

EMAT is used where liquid couplant won't work (hot, coated or moving metal); IRIS is used inside heat exchanger and boiler tubes; and laser UT is used mostly on production lines and in research. Rail owners and structural QA managers will rarely specify any of the three.

EMAT

An electromagnetic acoustic transducer uses a magnet and a coil to generate sound inside a conductive or ferromagnetic part, with no couplant. ASTM E1774 lists the advantages (moving, hot and rough parts, and horizontally polarized shear waves without mode conversion) and the main cost: EMATs "have very low efficiency", with insertion losses of 40 dB or more. ASTM E1816 covers EMAT thickness measurement up to 650 °C (1200 °F). In rail, EMAT work on the head and foot is mostly laboratory research; revenue rail testing still relies on liquid-coupled wheel and sled probes.

IRIS (internal rotary inspection system)

IRIS sends a pulse along the tube axis onto a mirror spun by a water-driven turbine, which turns the beam onto the tube wall and sweeps the circumference as the probe is pulled. Evident gives a working range of 0.5 to 3 in. ID and a pull speed limited to about 2 in./s (50 mm/s). It needs clean, water-filled tubes and can't pass U-bends. For that reason IRIS is often used to check what a faster eddy current screen has flagged.

Laser ultrasonics

Laser UT generates sound with a pulsed laser and detects it with a second laser and an interferometer, so nothing touches the part. That suits hot, moving or delicate parts on a production line and in research. The equipment is costly and complex, and no general field code governs it, so it isn't a realistic option for weld or rail inspection on site today.

Surface waves

Surface (Rayleigh) waves are another specialist mode. They can find surface-breaking cracks on smooth surfaces but reach only about one wavelength deep. In rail they remain mostly a research topic; see surface wave UT for rail for what they can and can't do on the railhead.

Worked example: which type of ultrasonic testing fits each of three jobs?

The right technique follows from the flaw you're looking for, the access you have and the code that sets acceptance. Here are three jobs that EMA's readers commonly buy.

Job 1: a 1 in. CJP butt weld under AWS D1.1

Inputs: 1.00 in. (25.4 mm) plate, single-V groove, 30° bevel, 1/8 in. root, statically loaded building steel. The contract calls for UT of CJP groove welds (D1.1 itself doesn't require UT; the contract or AISC 360 does).

Technique: a 0° base metal scan for laminations, then manual angle-beam shear wave testing to D1.1 Clause 8, from both sides of the weld axis where possible. PAUT under the D1.1 PAUT annex is an option if the contract allows it and the owner wants an encoded record.

Geometry check (sound path for one leg = t ÷ cos θ; full skip = 2t × tan θ):

The 1.28 in. cap is 2 × 1.00 × tan 30° + 0.125 = 1.155 + 0.125 = 1.28 in. An indication at the bottom of the first leg with the 70° probe lies at 1.00 in. depth, which matches the plate thickness, so the geometry is consistent. D1.1 rates each indication with d = a − b − c, where the attenuation factor c is 2 dB per inch of sound path beyond the first inch. At 2.92 in., c = (2.92 − 1.00) × 2 = 3.8 dB (rating method per the Evident AWS tutorial and earlier code editions; confirm the class limits in the edition you work to).

Why not TOFD? At 25.4 mm the weld sits inside the TOFD range, but a 4 to 5.5 mm lateral-wave dead zone covers roughly 16 to 22% of a 1 in. (25.4 mm) wall, and D1.1 acceptance is amplitude-based. TOFD would be a supplement for sizing, not the test of record.

Job 2: rail in track

Inputs: a Class 4 freight line, so 213.237(c) requires internal inspection at least every 370 days or 30 MGT, whichever comes first.

Technique: continuous multi-channel pulse-echo from a test car or walking cart, with liquid-coupled wheel or sled probes at 0°, 70° and an intermediate angle (often 37–45°). The output is a B-scan style channel map. Suspect indications are then verified by hand with an A-scan flaw detector, and the operator must be qualified under 213.238.

What's added: thermite field welds get separate hand UT of the base, because test cars don't cover most of the rail base (TTCI TD-00-013). PAUT is used for verification and research in rail, not routine testing, and TOFD, immersion and IRIS don't apply. The rules are covered in EMA's guide to 49 CFR 213.237 rail flaw detection.

Job 3: corrosion mapping on an aboveground line

Inputs: 0.375 in. nominal carbon steel pipe, some of it insulated, with external corrosion suspected.

Technique, in two stages:

1. Screen the insulated runs with guided wave testing (the pipe is inside the E2775 range of 0.15 to 1 in. wall). GWT flags the locations but gives no thickness.
2. Measure at the flagged locations and the exposed runs. Use dual-element spot gauging to ASTM E797 for a grid of readings, or encoded 0° PAUT for a C-scan map where the owner needs coverage of an area rather than points.

Worked reading: at the steel default of 0.2331 in./µs, a round trip of 2.57 µs gives T = 0.2331 × 2.57 ÷ 2 = 0.300 in. That is 0.075 in. lost from 0.375 in., or 20% of nominal wall. Calibrate on a block of the same material first, because the default velocity is only an assumption. Whether 0.300 in. is acceptable depends on the owner's code, such as API 570 for in-service process piping, not on the UT method.

What are the limits of each type of ultrasonic testing?

Every ultrasonic technique has a blind spot, and naming it in the procedure is part of doing the job properly. The main ones are:

• Orientation. A reflector parallel to the beam returns almost nothing, so a single angle never gives full coverage. D1.1 Clause 8 calls for testing from both sides of the weld axis where mechanically possible (per secondary copies of the code) for this reason.
• Near-surface and back-wall zones. Single-element contact probes have a dead zone just under the surface, and TOFD has lateral-wave and back-wall dead zones. Dual-element probes, delay lines and pulse-echo pairing cover them. See EMA's page on the dead zone in ultrasonic testing.
• Coarse grain. Cast austenitic manganese frogs and austenitic welds scatter sound, so standard UT struggles on them and sizing becomes unreliable.
• Surface condition. Shelling, head checks and corrugation on rail, and paint, scale and pitting on steel, block or distort the beam. On rail, lost coupling means untested rail, not a passed test.
• Sizing. Amplitude does not equal flaw size. ASTM E2700 cautions about PAUT oversizing, and TOFD gives height but needs interpretation.
• Screening methods don't measure. Guided wave gives locations, not wall thickness. Test-car rail data shows indications that still need hand verification.
• People. Every technique here depends on the procedure, the calibration and a qualified operator. Under SNT-TC-1A, the employer certifies its own NDT personnel.

How EMA Industries applies this

EMA Industries performs ultrasonic rail flaw testing for transit agencies, short lines and industrial track owners, rail weld inspection, and AWS D1.1 structural weld UT, with MT and PT and NDT Level III services. We're based in Southern California and mobilize nationwide. The technique choice belongs in the written procedure: the probe angles and coverage for a weld, the beam set and verification steps for rail, and the reporting of any area that couldn't be tested. To scope a job, see our ultrasonic examination services or contact us.

Related reading: AWS D1.1 weld inspection criteria · Guided wave ultrasonic testing · TOFD rail inspection

Frequently asked questions

What is the most common type of ultrasonic testing for welds?

Manual angle-beam shear wave testing at 45°, 60° or 70°, read on an A-scan. AWS D1.1 Clause 8 and ASME Section V Article 4 are both written around it, and PAUT is increasingly used where the contract allows it.

Is TOFD more accurate than phased array?

For through-wall height, TOFD is usually the better sizing tool because it times diffracted tip signals instead of relying on amplitude. It has dead zones at both surfaces and still needs a trained analyst, so most weld procedures pair it with PAUT or pulse-echo rather than using it alone.

What type of ultrasonic testing is used on railroad rail?

Multi-channel pulse-echo testing from a test car or walking cart, with 0°, 70° and intermediate-angle probes coupled through water-filled wheels or sleds. Suspect indications are verified with a hand flaw detector, and the search must meet the valid-search definition in 49 CFR 213.237.

Can ultrasonic testing measure corrosion through insulation?

Guided wave testing can screen insulated pipe for changes in cross-section from one clamp position, but ASTM E2775 calls it a screening tool that gives no direct wall thickness. You still need thickness gauging or PAUT at the flagged spots, usually with the insulation removed there.

When is EMAT used instead of conventional UT?

When liquid couplant can't be used: hot surfaces (ASTM E1816 covers thickness to 650 °C), moving parts, or scaled and coated metal. EMAT works only on conductive or ferromagnetic material, and its low efficiency means more specialized instruments.

Sources

• 49 CFR 213.237, Inspection of rail, eCFR (text read on Cornell LII)
• 49 CFR 213.238, Qualified operator, eCFR
• ASTM E2373/E2373M-19(2024), Ultrasonic Testing Using the Time-of-Flight Diffraction (TOFD) Technique, ASTM store page
• ASTM E2700-26, Contact Ultrasonic Testing of Welds Using Phased Arrays, ASTM store page
• ASTM E164-24, Contact Ultrasonic Testing of Weldments, ASTM store page
• ASTM E587-15(2025), Ultrasonic Angle-Beam Contact Testing, ASTM store page
• ASTM E114-20, Ultrasonic Pulse-Echo Straight-Beam Contact Testing, ASTM store page
• ASTM E797/E797M-21, Measuring Thickness by Manual Ultrasonic Pulse-Echo Contact Method, ASTM store page
• ASTM E1001-21(2026), Immersed Pulse-Echo Ultrasonic Method Using Longitudinal Waves, ASTM store page
• ASTM E2775-16(2023), Guided Wave Testing of Above Ground Steel Pipework Using Piezoelectric Effect Transduction, ASTM store page
• ASTM E1774-17(2022), Electromagnetic Acoustic Transducers (EMATs), ASTM store page
• ASTM E1816-18(2022), Measuring Thickness by Pulse-Echo EMAT Methods, ASTM store page
• Evident, Material sound velocities
• Evident, 3 methods to inspect heat exchanger tubes (IRIS)
• BINDT, Dead zone (DZ) definition, citing BS EN ISO 16828
• Baker Hughes, What is time of flight diffraction (TOFD)? (vendor)
• AWS D1.1/D1.1M, Structural Welding Code – Steel, Clause 8 (paywalled; clause and table details from secondary sources and marked for confirmation)