Ultrasonic Thickness Measurement: 5 Setups and When to Use Each

Ultrasonic thickness measurement finds wall thickness from one side by timing a sound pulse through the material: thickness = velocity × round-trip time ÷ 2. In steel the pulse travels at about 5,920 m/s (0.233 in/µs), so a 10 mm wall returns its echo in about 3.4 µs. ASTM E797/E797M-21 is the governing practice for contact pulse-echo gauging.

Five setups cover most field work: single-element pulse-echo, dual-element for corrosion, echo-to-echo through coatings, delay-line for thin material, and high-temperature gauging. This guide gives the figures behind each, a worked calibration and corrosion-rate example, and the limits an inspector or QA manager needs before trusting the readings.

How does ultrasonic thickness measurement work?

An ultrasonic thickness gauge times an echo and multiplies by the material's sound velocity. Evident's thickness-gauging tutorial gives the relation as T = V × (t/2), where T is thickness, V is the longitudinal velocity in the test material and t is the measured round-trip transit time.

A transducer, typically 2 to 10 MHz, sends a short longitudinal-wave pulse through a couplant film into the part. The pulse reflects from the far wall (the back wall) and returns. The gauge measures the transit time and divides by two because the sound crosses the wall twice.

Two consequences follow, and they drive every error discussed below:

• Velocity is an input, not a measurement. The gauge has no way of knowing the velocity of the material in front of it. If the velocity setting is 1% wrong, the reading is 1% wrong.
• The gauge reports the first strong echo it sees. That is usually the back wall, but it can be a lamination, an inclusion, the base of a pit or the second echo of a thin plate.

For the underlying physics (wave modes, impedance and why a couplant is needed), see the science behind ultrasonic testing and the basic principles of ultrasonic testing.

Typical longitudinal velocities for thickness gauging

Evident calls its table "only a general guide" and notes that composition, microstructure, grain orientation and temperature all shift velocity. The 5,920 m/s and 5,890 m/s figures differ by 0.5%, which is why ASTM E797 and every gauge manual tell you to calibrate on a sample of the actual material rather than trust a table.

Key takeaway: a thickness reading is only as accurate as the velocity behind it.

Which 5 setups cover most ultrasonic thickness measurement work?

The five setups are single-element pulse-echo, dual-element, echo-to-echo, delay-line and high-temperature. They differ in which two events the gauge times and in which transducer it uses. Evident's tutorial groups single-element timing into three modes, and duals have their own mode.

1. Single-element contact pulse-echo

Contact pulse-echo with a single-element probe is the default for clean, smooth, uncoated metal. Evident describes Mode 1 as "the most common approach" and says it typically gives the greatest maximum thickness capability.

The same element sends and receives, so the gauge must wait for the element to stop ringing before it can hear an echo. That sets a higher minimum measurable thickness than the other setups. The dead zone in ultrasonic testing explains the same effect for flaw detection. Evident also notes that contact transducers in Mode 1 can be used only on surfaces below 50 °C (125 °F).

Single-element gauging is the method ASTM E797 describes, and it is the right choice for machined parts, new plate and pipe, and wall checks on smooth material. On a corroded back wall it can miss the thinnest point, because a rough or pitted surface scatters the echo. See pulse-echo ultrasonic inspection for the general technique.

2. Dual-element transducers for corrosion

Dual-element transducers are the standard tool for remaining-wall surveys on corroded equipment. The probe has separate transmitting and receiving elements on delay lines, separated by an acoustic barrier. Evident's corrosion application note says the two beam paths "cross beneath the surface" of the part, which gives a pseudo-focusing effect.

That crossing makes duals more sensitive than single-element probes to echoes from the base of pits, which is where the minimum remaining wall is. Because the transmitter and receiver are separate, there is no ringdown to wait for, so duals also handle rough surfaces better.

Probe choice sets the working range. Evident's corrosion note lists these approximate figures for three of its duals, and says they should be verified experimentally:

Higher frequency means a lower minimum thickness but less penetration. The note also advises taking data at increments no greater than half the transducer diameter, which is about 5.5 mm for an 11 mm D790, and aligning the acoustic barrier perpendicular to the pipe axis on curved surfaces.

3. Echo-to-echo through coatings

Echo-to-echo measurement ignores paint by timing between two back-wall echoes instead of from the surface. Both echoes have crossed the coating the same number of times, so the coating drops out of the interval.

Without it, a pulse-echo gauge counts the coating as if it were steel. Coatings are slower than steel, so the error is larger than the coating itself. For example, take a 0.30 mm coating with an assumed velocity of 2,600 m/s, in the range Evident lists for plastics such as nylon. It adds 0.30 × 5,920 ÷ 2,600 = 0.68 mm to the steel reading. On a 6 mm wall that is an 11% overstatement, in the unsafe direction for a corrosion survey.

Evident's 45MG gauge page describes two options:

• Echo-to-Echo measures between successive back-wall echoes and shows the true metal thickness.
• THRU-COAT uses a single back-wall echo and dedicated duals to show both metal and coating thickness.

The trade-off: echo-to-echo needs at least two clean back-wall echoes, and Evident's corrosion note says it can be less effective at detecting pitting. Badly pitted walls under thick coatings may still need the coating removed at critical points.

Measuring the coating itself is a different job. Dry-film thickness of organic coatings by ultrasonic gauge is covered by ASTM D6132-13(2022), not by E797.

4. Delay-line probes for thin material

Delay-line transducers are the choice for thin walls, because the delay line moves the surface echo away from the excitation pulse. In Mode 2 the gauge times from the delay-line tip echo to the first back-wall echo, and in Mode 3 between successive back-wall echoes. Evident says Mode 3 typically offers the highest accuracy and best minimum-thickness resolution.

Higher frequencies help on thin material because the pulse is shorter. At 5,920 m/s the wavelength is about 1.18 mm at 5 MHz and 0.59 mm at 10 MHz. The limit is the other end of the range: Evident says the main disadvantage of Mode 2 is that maximum thickness is limited by delay-line length. If the back-wall echo arrives after the second delay-line echo, the gauge can lock on the wrong signal.

Note that thickness gauging is not a near-field problem in the way amplitude sizing is. Evident states that gauging in the near field "is not a problem". See near field vs far field in UT for when it does matter.

5. High-temperature gauging

High-temperature gauging uses probes and couplants rated for the surface temperature, plus a velocity correction. Standard contact transducers are limited to about 50 °C (125 °F). Evident's high-temperature article says a dual such as the D790 can be used on surfaces up to 500 °C (930 °F) with intermittent contact of 5 seconds or less.

Couplant is rated separately. Evident's corrosion note lists one couplant for about −18 to 400 °C (0 to 750 °F) and another for about 371 to 538 °C (700 to 1,000 °F).

Two errors appear at temperature:

• Velocity drops as steel heats. The size of this error is worked through in the next section.
• The dual's zero offset rises as the probe heats up. Evident's dual-element setup tutorial says to recalibrate zero at the working temperature if the part is much hotter than the calibration block.

High-temperature work is outside the scope of ASTM E797, which covers temperatures not exceeding 93 °C (200 °F). A written procedure for hot work therefore needs its own basis, normally the probe and gauge maker's instructions plus the owner's inspection program.

Key takeaway: pick the setup from the surface (coated, corroded, hot) and the thickness range, not from the gauge's feature list. For gauge-versus-flaw-detector choices, see thickness gauge or flaw detector and the ultrasonic thickness gauge guide.

Which standards and codes govern thickness readings?

ASTM E797/E797M-21 governs how to make a contact pulse-echo thickness measurement. API and ASME codes govern what the readings mean for in-service equipment. The ASTM store lists E797/E797M-21 as the active edition (checked 24 October 2026). Its scope says the practice:

"provides guidelines for measuring the thickness of materials using the contact pulse-echo method"

— ASTM E797/E797M-21, Scope 1.1

The ASTM store page states three further scope limits, and each one matters in the field:

• Temperature: surfaces not exceeding 93 °C (200 °F).
• Material: any material in which sound travels at a constant velocity and gives a resolvable back-wall echo.
• Geometry: primarily for flat parts with parallel surfaces, with limited applicability to non-parallel or concentric surfaces.

The scope also notes that the method is used to find wall thinning in process equipment caused by corrosion and erosion.

API and ASME codes are paywalled, so the clause-level details below come from study guides and secondary summaries and are hedged. The current code text and the owner's inspection program control.

Key takeaway: E797 tells you how to take the reading. The owner's code (API 510 for vessels, API 570 for piping) tells you how often and what to do with it.

How do you calibrate a thickness gauge?

Calibrate velocity and zero together with a two-point calibration on two samples of the same material, one thick and one thin. Evident recommends a thickness ratio of at least 2:1 between the blocks, and 5:1 or more as optimum, with both inside the range you will measure.

Two-point calibration procedure

1. Choose two reference samples of the same material as the part, ideally with a similar surface finish. For a 3 to 12 mm survey, a step block with 2.5 mm and 12.5 mm steps gives a 5:1 ratio. In inches, 0.100 in. and 0.500 in. does the same.
2. Velocity step: couple to the thick sample, let the reading settle, and adjust the display to the sample's known thickness.
3. Zero step: couple to the thin sample, let it settle, and adjust to its known thickness.
4. Verify on at least one thickness between the two calibration points.
5. Recheck at intervals set by the procedure, after a probe change, and whenever the surface temperature changes significantly.

Evident's calibration tutorial gives a practical diagnosis rule. An error that grows with thickness points to velocity. A fixed offset at every thickness points to zero. Either way, repeat the two-point calibration.

Resolution is not accuracy

Display resolution is the smallest step the gauge shows; accuracy is how close the reading is to the true thickness. Evident's 45MG specification lists selectable resolution of 0.1 mm (0.01 in.) or 0.01 mm (0.001 in.), and 0.001 mm (0.0001 in.) with the single-element option. A display that reads 0.01 mm can still be 0.1 mm or more wrong if velocity, zero, coupling or temperature are off.

On corroded surfaces, quote readings to the precision the surface supports. A rough, pitted wall rarely justifies the third decimal place in inches.

Key takeaway: two points, same material, 2:1 or better, then verify in the middle.

How big are the velocity, temperature and V-path errors?

Each error source can be estimated, and most of them push the reading in a direction you can predict. The worked figures below use stated inputs. They are illustrations of scale, not acceptance limits.

Velocity error

A 10.00 mm wall of 4340 steel (5,850 m/s) is measured with the gauge left at the 5,920 m/s default:

• True transit time = 2 × 10.00 ÷ 5.850 = 3.419 µs
• Gauge reading = 5.920 × 3.419 ÷ 2 = 10.12 mm, about 1.2% thick

The same wall in 1020 steel (5,890 m/s) would read 10.05 mm. On a line corroding at 0.15 mm/yr, a 0.12 mm error is most of a year's metal loss.

Temperature error

Evident's high-temperature article cites ASTM E797-95 for the rule of thumb that carbon steel velocity falls by about 1% per 55 °C (100 °F) rise. Treat it as a linear approximation.

Take a 12.70 mm (0.500 in.) steel wall at 200 °C, gauged with a room-temperature (20 °C) calibration:

• Temperature rise = 180 °C, so velocity is about 180 ÷ 55 = 3.3% lower
• Reading ≈ 12.70 ÷ (1 − 0.033) = 13.13 mm (0.517 in.)

The gauge overstates the wall by about 0.4 mm. The fix is a temperature-compensation setting, where the gauge offers one, or a calibration on a block at the working temperature. Evident notes that calibrating on a hot block is slow and may be imprecise if the block and the part are at different temperatures.

V-path error on dual probes

On a dual, the sound travels a V-shaped path, so thin walls read thicker than they are unless the gauge corrects for it. An Olympus NDT (now Evident) patent, US 8,156,784, states that V-path errors occur when the material is thinner than the probe's focal thickness, and that the time distortion increases as thickness decreases without correction. Gauges apply stored V-path correction tables for their duals. General-purpose flaw detectors typically don't, so Evident advises checking a dual setup on reference blocks of several thicknesses.

A simplified straight-ray model shows the trend. Assume the beam enters and leaves 2 mm either side of center (an illustrative value, not a probe specification). The uncorrected reading is then √(t² + 2²):

Real probes differ (roof angle, delay-line travel, refraction), and zero calibration absorbs part of the offset. The trend is what matters: V-path error is negligible on thick walls and serious on thin ones, so a dual calibration should include a block near the thinnest wall you expect to report.

How do thickness readings become a corrosion rate and remaining life?

Corrosion rate is metal lost per operating year at a fixed location, and remaining life is the margin above the required thickness divided by that rate. Study guides for API 570 give the long-term (LT) and short-term (ST) rates as:

• CR(LT) = (t_initial − t_actual) ÷ years between them
• CR(ST) = (t_previous − t_actual) ÷ years between them
• Remaining life = (t_actual − t_required) ÷ governing corrosion rate

The same guides say the inspector normally uses the larger of the two rates, or the rate that best reflects current operation after consulting the piping engineer. They describe the piping thickness interval as the lesser of half the remaining life and a maximum set by piping class. Secondary summaries of API 510 describe a similar rule for vessels: the lesser of half the remaining life or 10 years, unless a risk-based inspection assessment justifies otherwise. Confirm both rules and their clause numbers in the edition your owner uses.

The required thickness comes from the design code, for example the ASME B31.3 pressure-design formula or a structural minimum, whichever is greater. That number is an engineering input; it is not something the UT technician measures.

Worked example: piping corrosion rate and remaining life

A carbon steel line has a thickness measurement location (CML) read at the same point, with the same method, in three surveys. The required thickness is 0.180 in. (4.57 mm), supplied by the piping engineer. The line ran continuously, so operating years equal calendar years.

1. Long-term rate: (0.280 − 0.232) ÷ 10 = 0.048 ÷ 10 = 0.0048 in./yr (4.8 mpy, 0.122 mm/yr)
2. Short-term rate: (0.262 − 0.232) ÷ 5 = 0.030 ÷ 5 = 0.0060 in./yr (6.0 mpy, 0.152 mm/yr)
3. Governing rate: the short-term rate is higher, so use 0.0060 in./yr unless the engineer justifies otherwise
4. Remaining life: (0.232 − 0.180) ÷ 0.0060 = 0.052 ÷ 0.0060 = 8.7 years
5. Next thickness inspection: half the remaining life is 4.3 years. That is shorter than the class maximums the study guides give (5 or 10 years), so it governs

Why reading quality changes the answer

Suppose the 2026 survey missed the bottom of a pit, and the true minimum at that CML is 0.227 in. (5.77 mm), only 0.005 in. thinner:

A 5-mil difference in one reading removes two years of calculated life. In the velocity example above, an uncorrected default velocity alone produced a 0.12 mm (about 5 mil) error. That is why calibration, CML location control and the minimum-reading method matter as much as the gauge.

The readings do not decide fitness for service on their own. Local thin areas and pitting are assessed under API 579-1/ASME FFS-1, which secondary summaries say treats general metal loss, local metal loss and pitting in separate parts.

What can't ultrasonic thickness measurement tell you?

A thickness gauge reports one number per spot; it cannot tell you the thinnest point on a part it didn't cover, or what the reflector is. These are the main limits.

Pitting and isolated wall loss

A pit narrower than the beam may return a weak echo that the gauge ignores, so the gauge reads the surrounding wall. Duals are more sensitive to pit bottoms than single-element probes, but no spot gauge finds a pit it never sat on. Echo-to-echo modes are weaker still on pitting, according to Evident. Where pitting is expected, the procedure should call for a scan or grid at close spacing, not isolated spots.

Minimum reading versus scan

A spot reading measures one point; a scan with a minimum-capture function measures the thinnest point across an area. Many gauges hold the lowest value seen while the probe is moved, which Evident calls Freeze Minimum on the 45MG. A B-scan or encoded corrosion map shows the profile. The procedure must say which is required at each CML, because a spot reading and a scanned minimum are not comparable in a corrosion-rate calculation. For display types, see A-scan, B-scan and C-scan in UT.

Coatings

Pulse-echo through a coating overstates the wall, as shown above. Echo-to-echo needs clean multiple echoes, which heavily corroded or thick coatings may not give. Thick, disbonded or multilayer coatings may need removal at the CML.

Curvature and small pipe

ASTM E797 says the practice has limited applicability to non-parallel or concentric surfaces. On small-diameter pipe a flat probe rocks, couples on a line and can read low or unstable. Use a probe rated for the radius (Evident lists 20 mm minimum radius for the D790), orient the dual's barrier perpendicular to the pipe axis, and check the setup on a curved reference of similar diameter.

What else it misses

• Laminations and inclusions can return an echo above the back wall and be reported as thin wall. An A-scan display helps tell them apart; see the UT formulas guide for time-of-flight checks.
• Cracks perpendicular to the surface are not found by a 0° thickness probe. Crack detection needs angle-beam UT or another method.
• Coarse-grained or attenuative material (some castings, austenitic welds) may give no usable back-wall echo.
• Non-uniform velocity breaks the E797 assumption of constant velocity through the part.

Key takeaway: state in the report whether each reading is a spot or a scanned minimum, the probe, the velocity, the calibration blocks and the surface condition.

How EMA Industries applies this

EMA Industries is a Southern California NDT contractor that mobilizes nationwide. It offers NDT Level III services and ultrasonic examination, including AWS D1.1 structural weld UT. For thickness work, a Level III review addresses the questions in this guide: which setup suits the surface and temperature, whether the calibration and verification steps meet ASTM E797, and whether readings are recorded so a corrosion rate can be calculated from them later. To discuss a survey procedure or a review of existing thickness data, contact EMA Industries.

Related reading: ultrasonic thickness testing in railway inspection and ultrasonic testing equipment.

Frequently asked questions

How accurate is ultrasonic thickness measurement?

Accuracy depends mostly on calibration, not the display. A standard gauge displays 0.01 mm (0.001 in.), but the reading is only as good as the velocity and zero settings. An uncorrected velocity error of about 1% gives a 1% thickness error, and heat adds about 1% per 55 °C on carbon steel. With a two-point calibration on the same material, the remaining limits are surface condition and coupling. A gauge maker's accuracy statement applies only under its own stated conditions, and corroded surfaces support much less precision.

What ASTM standard covers ultrasonic thickness measurement?

ASTM E797/E797M, Standard Practice for Measuring Thickness by Manual Ultrasonic Pulse-Echo Contact Method, is the governing practice. The active edition is E797/E797M-21. It covers contact pulse-echo measurement on surfaces up to 93 °C (200 °F), mainly on flat parts with parallel surfaces. Coating dry-film thickness by ultrasonic gauge is a separate standard, ASTM D6132. In-service evaluation of the readings comes from codes such as API 510 and API 570.

How do you calibrate an ultrasonic thickness gauge?

Use a two-point calibration on two samples of the same material: a thick one to set velocity and a thin one to set zero. Evident recommends at least a 2:1 thickness ratio, ideally 5:1 or more, with both inside the measuring range. Verify on a thickness between the two. If the error grows with thickness, recheck velocity; if it is a fixed offset, recheck zero.

Can an ultrasonic gauge measure steel thickness through paint?

Yes, if the gauge has an echo-to-echo or through-coating mode and the coating is well bonded. Echo-to-echo times between two back-wall echoes, so the coating drops out. Standard pulse-echo counts the coating as steel and overstates the wall, often by more than the coating thickness itself. Echo-to-echo is less effective at finding pits, so critical locations on pitted walls may still need the coating removed.

What is the thinnest wall an ultrasonic thickness gauge can measure?

It depends on the probe and mode. Evident lists about 1 mm (0.04 in.) for a 5 MHz dual and about 0.5 mm (0.02 in.) for a 10 MHz dual, and says to verify these experimentally. Delay-line probes in echo-to-echo mode reach thinner walls on smooth material. Below a probe's range, the gauge may lock on the second echo and read roughly double, so check thin readings on an A-scan or a thin reference block.

Sources

• ASTM International, E797/E797M-21 Standard Practice for Measuring Thickness by Manual Ultrasonic Pulse-Echo Contact Method (store page: scope, significance and use, active status; checked 24 Oct 2026)
• ASTM International, D6132-13(2022) Dry Film Thickness of Applied Organic Coatings Using an Ultrasonic Coating Thickness Gage
• Evident, Material Sound Velocities
• Evident, How do thickness gauges work? (single-element modes)
• Evident, Velocity and Zero Calibration
• Evident, Corrosion Gaging with Dual Element Transducers
• Evident, Typical Dual Element Setup Procedure
• Evident, 2 Challenges of High-Temperature Thickness Gauging
• Evident, 45MG Ultrasonic Thickness Gauge specifications
• US Patent 8,156,784, System and method for derivation and real-time application of acoustic V-path correction data (Olympus NDT, 2012)
• American Petroleum Institute, API 510, 11th edition announcement
• Hydrocarbon Processing, API strengthens piping inspection standards (API 570 and RP 574 fifth editions), 29 Feb 2024
• Open Exam Prep, API 570 corrosion rate calculations and thickness inspection intervals (study guide; secondary)