Ultrasonic Testing: The Practitioner's Guide

Table of Content

• Introduction
• Principles of ultrasonic testing
• Pulse-echo and other conventional techniques
• Time-of-flight diffraction (TOFD)
• Phased array ultrasonic testing (PAUT)
• Full matrix capture and the total focusing method (FMC/TFM)
• Automated ultrasonic testing (AUT)
• Guided wave testing (GWT)
• Calibration and verification
• Worked examples
• Codes, standards and personnel qualification
• Limitations and technique selection
• Common field errors
• Frequently asked questions
• References

Introduction

One of the most effective nondestructive testing (NDT) techniques is ultrasonic testing (UT), which may be used for three purposes: detecting internal flaws, sizing them, and figuring out how thick the remaining wall is. The instrument sends high-frequency sound into the component and interprets the echoes that come back. There is no radiation to control, most methods only require access from one side, and depth information is shown immediately on the screen.

Level II and Level III technicians as well as the inspection leads who authorize their work are the target audience for this guide. Since physics is the foundation of all techniques, we start with it and proceed to pulse-echo, TOFD, phased array, TFM, automated UT, and guided waves, followed by calibration, worked examples, codes and limitations.

Principles of ultrasonic testing

Wave modes and velocity

The majority of contact flaw detection occurs between 0.5 MHz and 15 MHz. Longitudinal (compression) waves pass through solids, liquids, and gases. Shear waves only pass through solid materials, and in steel they travel at roughly 55% of the longitudinal speed, which is slightly more than half. Lamb and other guided waves cover the entire wall of thin structures, whereas Rayleigh waves travel along the surface and penetrate only about one wavelength deep.

Wavelength is just velocity divided by frequency, and velocity is dependent on the density and elastic constants of a material. This results in a longitudinal wavelength of approximately 1.2 mm and a shear wavelength of approximately 0.65 mm in steel at 5 MHz. Remember those two figures; they keep coming up.

Treat these as typical values. Alloy, temperature and heat treatment all shift velocity, so set or verify it on a block of the same material as the part you are testing.

Acoustic impedance and reflection

The mismatch between two materials determines how much sound bounces back, and acoustic impedance (Z) is calculated by multiplying density by velocity. The proportion of energy reflected is equal to ((Z2 − Z1) / (Z2 + Z1))². About 88% of the energy is reflected at a steel–water interface, and nearly all of it is reflected at a steel–air interface. This explains why a gas-filled crack is so obvious and why a dry probe yields no results until couplant is added.

Refraction and mode conversion

When sound strikes a boundary at an angle, it does not continue straight ahead. In accordance with Snell's law (sin θ1 / V1 = sin θ2 / V2), it bends, and a portion of it can change from one wave mode to another. The first and second critical angles for an acrylic wedge on steel are approximately 27.5° and 57°, respectively. The longitudinal wave no longer penetrates the steel when the incident angle is set anywhere between those two, leaving only the shear wave. Standard angle probes provide 45°, 60°, or 70° in steel in this manner. Curved surfaces make things more difficult. The scan plan must take into consideration the fact that the contact geometry on pipe changes both the beam path and the refracted angle.

Beam characteristics

Amplitude-based evaluation cannot be trusted in the near field close to the probe because interference causes amplitude to rise and fall unpredictably. The near-field length for a circular element is N = D²f / 4V, where V is the velocity, f is the frequency, and D is the element diameter. The near field of a 10 mm, 5 MHz longitudinal-wave probe in steel is approximately 21 mm. After that, the beam expands and the amplitude gradually decreases with distance; the half-angle to the first null is determined by sin γ = 1.22λ / D.

Attenuation and detectability

As sound moves, it loses energy due to both absorption and scattering. Scattering increases sharply with frequency and becomes significant when grain size is a meaningful fraction of the wavelength. Because of this, testing coarse-grained castings, austenitic welds, and nickel alloys is far more difficult than testing fine-grained carbon steel. Although orientation, surface roughness, and noise all have a role, it is generally accepted that a reflector must be around half a wavelength across in order to be reliably detected.

Pulse-echo and other conventional techniques

One transducer sends a brief pulse and listens for echoes during pulse-echo testing. The A-scan graphs amplitude against time, and when velocity is set, time becomes sound path, depth, and surface distance.

The usual tool for laminations, plates, forgings, and thickness is a straight-beam probe, which transmits a 0° longitudinal wave into the part. Because cracks and lack of fusion are rarely parallel to the surface, an angle-beam probe for welds produces a refracted shear wave. Dual-element (TR) probes provide significantly greater near-surface resolution on thin or corroded walls by separating the transmitter and receiver. Through-transmission works well with highly attenuating materials and composites because it places probes on opposing sides. Immersion testing couples through water, which maintains consistent sensitivity and makes focused probes and automated scanning feasible.

Thickness gauging

One of the most frequent tasks in UT is thickness measurement (ISO 16809, ASTM E797). While precision gauges employ single-element probes on smooth surfaces, corrosion gauges use dual-element probes to deal with pitted back walls. If the part is painted, echo-to-echo mode measures between consecutive back-wall echoes, excluding the coating from the measurement.

Conventional sizing

The standard technique for sizing length is the 6 dB drop method, which is effective when the reflector is larger than the beam. Length is overestimated when the reflector is smaller since you end up measuring the beam rather than the flaw. The equivalent reflector size, not the actual size, can be found by comparing amplitude with a reference curve. Tip diffraction provides the optimum result for through-wall height, and TOFD automates this exact technique.

Time-of-flight diffraction (TOFD)

TOFD employs two angled longitudinal-wave probes in pitch-catch, one on each side of the weld. The sound is diffracted by the upper and lower tips of a planar flaw, and the diffracted signals arrive at the receiver between the lateral wave, which skims just below the scan surface, and the back-wall echo. Through-wall height is determined accurately since sizing is based on arrival time rather than amplitude; under good conditions, the uncertainty is often about ±1 mm.

The encoded data is displayed as a grayscale B-scan. A two-tipped planar flaw can be distinguished from a point reflector thanks to the opposite phases of the upper- and lower-tip signals.

There are blind spots in TOFD. There are two dead zones: one near the back wall and one near the scan surface, where signals are obscured by the lateral wave. For complete coverage, TOFD is typically combined with PAUT or pulse-echo. Additionally, it has trouble with coarse-grained materials and with transverse flaws unless additional scans are included.

For a deeper look at how TOFD compares with radiographic methods in weld inspection, read our dedicated guide on TOFD in Lieu of Radiography.

Phased array ultrasonic testing (PAUT)

A phased array probe contains many small elements, usually between 16 and 128, which are fired with precisely timed delays. Every group of delays is referred to as a focal law. A single probe can steer the beam at various angles, focus it at specific depths, and sweep it along the array without moving by electronically switching focal laws.

The majority of the work is done by two scan types. A sectorial scan (S-scan) creates a fan-shaped image by sweeping one aperture through a range of angles, such as 40° to 70°. Without moving the probe, an electronic linear scan (E-scan) covers a strip by stepping the active aperture along the array while maintaining a single angle.

Encoded scanning is a significant improvement over manual A-scan work because it provides recorded, reproducible data that can be reanalyzed and audited at a later time. Every focal law still requires its own sensitivity calibration, typically using TCG and angle-corrected gain (ACG).

Additionally, element pitch is important. Grating lobes can appear as false indications if the pitch is too coarse for the wavelength. They are suppressed at all steering angles by keeping the pitch at no more than half a wavelength.

Full matrix capture and the total focusing method (FMC/TFM)

Full matrix capture (FMC) records an A-scan for each transmitter-receiver pair (N × N A-scans for an N-element array) by firing each element sequentially while each element listens. Instead of focusing at a few focal depths as PAUT does, the total focusing method (TFM) creates an image by synthetically focusing at every pixel in a designated region. The result is crisper images of flaw tips and geometry and constant focusing over the whole zone.

TFM images are computed for wave sets, which are particular propagation paths. Anything that the beam reaches directly is imaged by direct modes like L-L or T-T. Planar flaws such as sidewall lack of fusion can be detected from far more advantageous angles by skip modes like TT-T or TT-TT that bounce off the back wall. The drawback of TFM is that it is less forgiving; mistakes in velocity, wall thickness, or geometry can produce images that look real but aren't.

The codes now cover FMC/TFM. FMC is covered by ASME Section V, Article 4, Mandatory Appendix XI, which was added in the 2019 edition. Its amplitude fidelity limit of 2 dB limits how coarse the image grid can be. FMC/TFM is covered by ISO 23865 for general use and ISO 23864 for welds.

Automated ultrasonic testing (AUT)

Because the probes in AUT travel on a motorized or mechanized scanner with position encoding, coverage, probe location, and data recording are no longer dependent on manual handling.

Pipeline girth welds are where AUT made its name. ASTM E1961 divides the weld into roughly equal vertical zones, each examined by a pair of focused search units. The procedure typically applies to pipe diameters of at least 15 cm and walls ranging from 7 mm to 25 mm. TOFD channels are commonly added, and phased arrays are frequently used in contemporary systems to create the zones. The results are displayed as strip charts, so evaluation keeps pace with production welding, and acceptance follows the pipeline code's workmanship or ECA-based criteria. Encoded corrosion mapping is another common application for AUT.

Guided wave testing (GWT)

Long-range UT, or guided wave testing, uses low-frequency waves that move along the pipe wall rather than through it. These waves are usually in the tens of kilohertz. Axisymmetric modes like the longitudinal L(0,2) mode and the torsional T(0,1) mode are excited by a transducer ring or collar clamped around the pipe. From a single test site, it records reflections from welds, supports, and changes in cross-section in both directions.

It's critical to understand that GWT is a screening tool. Every indication requires quantitative follow-up, such as UT thickness mapping, because it identifies changes in cross-section but does not measure wall thickness. Range is longest on bare, straight, above-ground pipe and drops sharply with coatings, soil, bends or viscous contents. The main standards are ISO 18211 and ASTM E2775.

To understand where guided wave fits within a broader pipeline inspection programme, see our article on how Guided Wave Testing fits in a pipeline inspection program.

Calibration and verification

A UT result is only as good as the setup behind it. Calibration works at three levels: periodic verification of the equipment, range and sensitivity setup for the job, and checks during and after the examination.

Equipment verification

ASTM E317 for pulse-echo instruments or ISO 22232-1, -2, and -3 (which superseded the EN 12668 series) are used to characterize instruments, probes, and entire systems. ISO 18563 applies to phased array equipment. ASME Section V, Article 4 mandates screen height linearity within ±5% of full screen height over 20% to 80% of the calibrated screen height, as well as amplitude control linearity. Intervals of no more than three months for analog instruments and one year for digital ones are specified in recent editions.

Range and probe checks

Set velocity and zero offset (probe delay) before each task, and confirm the index point and refracted angle for angle probes. The standard blocks for these checks are ISO 2400 (Calibration Block No. 1, also known as the IIW or V1 block) and ISO 7963 (Calibration Block No. 2, also known as the V2 block).

Sensitivity setting

Reference reflectors, such as side-drilled holes, flat-bottomed holes, or notches in a block that is acoustically comparable to the part, are used to set sensitivity. Sensitivity and range settings are covered by ISO 16811, and there are three primary methods. A DAC (distance–amplitude correction) curve is drawn through the peak responses from reflectors of the same size at different depths or sound paths, such as the side-drilled holes in the ASME basic calibration block. The same reflector displays the same screen height at every depth thanks to TCG (time-corrected gain), which applies gain that changes over time. DGS/AVG (distance–gain–size) reduces the need for physical reference blocks by using a calculated diagram that relates distance, gain, and equivalent flat-bottomed-hole size.

Transfer correction accounts for differences in surface condition and attenuation between the block and the part. Calibration is verified at least at the start and end of each examination. If it has drifted beyond tolerance, the code will typically require the work since the last valid check to be re-examined or re-evaluated.

Worked examples

All four examples assume carbon steel (longitudinal velocity 5.92 mm/µs, shear velocity 3.24 mm/µs), with probe delay already removed.

1. Locating an angle-beam indication

You are using a 60° shear-wave probe to scan a 20 mm plate.

• The half-skip surface distance is T × tan θ = 20 × 1.732 = 34.6 mm, so the full skip is 69.3 mm.
• The half-skip sound path is T / cos θ = 20 / 0.5 = 40 mm, so the full skip is 80 mm.
• At a 50 mm sound path, an indication appears. The beam has already bounced off the back wall because that lies between half and full skip.
• Surface distance from the probe index point is equal to S × sin θ = 50 × 0.866 = 43.3 mm.
• Depth is equal to 2T − S × cos θ = 40 − 25 = 15 mm below the scan surface.

Forget the skip correction and you would place this reflector at 25 mm, which is deeper than the plate itself.

2. Comparing an indication with DAC

On the DAC curve, the reference reflector is set to 80% FSH. You added 4 dB of gain to find an indication that peaks at 50% FSH.

• Screen-height difference = 20 log (50 / 80) = −4.1 dB.
• Removing the additional gain: −4.1 − 4.0 = −8.1 dB in relation to DAC.
• 39% DAC is equivalent to −8.1 dB.

As a fast sanity check, amplitude doubles or halves every 6 dB, thus −6 dB is 50% DAC and −12 dB is 25% DAC.

3. Sizing flaw height with TOFD

The half-separation S is 40 mm when two 5 MHz longitudinal-wave probes are positioned on a 30 mm plate with a probe center separation (PCS) of 80 mm. The depth of a flaw halfway between the probes is d = √((c × t / 2)² − S²).

• Arrival of the lateral wave: 80 / 5.92 = 13.5 µs. Back-wall arrival: 2 × √(40² + 30²) / 5.92 = 16.9 µs.
• Upper-tip signal at 15.4 µs: d = √(45.6² − 40²) ≈ 21.9 mm since path = 91.2 mm.
• Lower-tip signal at 16.2 µs: d = √(48.0² − 40²) ≈ 26.4 mm since path = 95.9 mm.
• Based on the unrounded depths, the flaw through-wall height is ≈ 26.4 − 21.9 ≈ 4.6 mm.

Because time and depth are not linearly related, timing errors hurt most near the scan surface.

4. Checking a phased array probe for grating lobes

If |sin θg| ≤ 1, then grating lobes occur at angles where sin θg = sin θs ± mλ / p, where m is a whole number. In this instance, the element pitch is called p and the beam steering angle is called θs. Imagine that a 5 MHz longitudinal-wave array is steered to 30° and in direct contact with steel (λ = 1.18 mm).

• With a 1.0 mm pitch, sin θg = 0.5 − 1.18 = −0.68, so a grating lobe occurs at an angle of around −43°, which is enough to produce false indications.
• With a 0.5 mm pitch, there is no grating lobe since sin θg = 0.5 − 2.36 = −1.86 has no real solution.

Codes, standards and personnel qualification

A code specifies what should be examined and what is allowed, whereas a standard specifies how the examination should be conducted. This is an easy way to keep things straight. A method standard like ASME Section V is typically cited by a construction code like ASME Section VIII or B31.3, which then applies its own acceptance criteria.

Always work to the edition named in the contract or referencing code, which is not necessarily the latest one.

Personnel are certified either under a central scheme to ISO 9712 or under an employer-based scheme such as ASNT SNT-TC-1A or ANSI/ASNT CP-189. Aerospace work typically uses NAS 410 or EN 4179. For PAUT, TOFD and TFM, expect to need technique-specific training as well and, under many codes, a performance demonstration on flawed samples.

Limitations and technique selection

UT is powerful, but it has real limits, and a good inspection lead plans around them in the procedure and scan plan.

The first limitation is orientation. Scan plans require many angles or directions because a planar flaw parallel to the beam returns very little energy. Next is the surface condition: curved, rough, or scaled surfaces scatter energy and cause amplitudes to become unstable. Because dead zones and echoes from roots, caps, and counterbores can both conceal and mimic flaws, geometry creates its own problems. Conventional shear-wave methods are replaced by lower frequencies and dual-element longitudinal-wave probes due to the significant attenuation, noise, and beam skewing caused by coarse-grained and anisotropic materials. Lastly, thin sections are challenging: thinner walls require specialized procedures, and most weld standards begin at 6 or 8 mm.

UT also works best alongside other methods. Radiography is better at characterizing volumetric flaws such as porosity, and eddy current testing is better suited to surface-breaking and near-surface cracks.

Common field errors

In practice, most UT errors come from setup and technique, not from the equipment. These are the ones worth looking for during audits and data review:

• Incorrect velocity: using the default steel velocity on stainless steel, castings, or hot parts shifts every reading.
• Uncorrected probe delay or worn wedges: the index point and refracted angle move unnoticed.
• Incorrect calibration block: unless transfer correction is used, a different finish, curvature, or attenuation renders the sensitivity setting invalid.
• Near-field evaluation: amplitude comparisons in the near field are unreliable.
• Poor couplant: dry spots cause fluctuating amplitude and missed indications.
• Coverage gaps: encoder slip, inadequate index overlap, and missing angles or directions.
• Misread geometry: root, cap, and counterbore echoes reported as flaws, or actual flaws written off as geometry.
• Skipped calibration checks: drift cannot be traced, so the inspection work may need to be repeated.

Frequently asked questions

Can UT replace radiography for welds?

Often, yes. Under certain conditions, such as a qualified procedure, qualified personnel, and specific acceptance criteria, ASME and pipeline codes permit encoded UT, typically PAUT and TOFD, to replace radiography. Always verify what is permitted by the governing code.

What is the difference between PAUT and TFM?

PAUT only focuses at the angles and depths determined by its focal laws. TFM employs FMC data to focus at every point in the image, providing consistent resolution but requiring additional data and processing.

What is the minimum thickness UT can examine?

Thickness gauges can measure about 1 mm or less with the right probes. However, the majority of weld examination standards begin at 6 or 8 mm.

Why must the calibration block match the part?

A mismatched block results in incorrect settings unless adjustments are made, because velocity, attenuation, and surface condition all alter range and amplitude.

References

These are the standards cited in this guide. Always use the edition specified by the contract or governing code.

• ASME BPVC Section V (Articles 4 and 5); AWS D1.1; API 1104
• ASTM E164, E317, E797, E1961, E2491, E2700 and E2775
• ISO 16809, 16810, 16811, 16823, 16826, 16827 and 16828 (general UT)
• ISO 17640, 11666, 23279 and 22825 (welds)
• ISO 10863 and 15626 (TOFD); ISO 13588 and 19285 (PAUT); ISO 23864 and 23865 (FMC/TFM)
• ISO 18211 (guided waves); ISO 22232 and 18563 (equipment); ISO 2400 and 7963 (calibration blocks)
• ISO 9712, ASNT SNT-TC-1A and ANSI/ASNT CP-189 (personnel)