A thickness reading can look precise to two decimal places and still be wrong enough to affect a remaining-life decision. That is the field problem this ultrasonic thickness measurement guide addresses. Reliable results depend less on pressing a probe to steel and more on understanding the material, setting up the instrument correctly, preparing the surface, and knowing when a reading does not represent the true minimum wall.

For corrosion work, ultrasonic thickness testing is fast, portable and highly useful. It can also hide localised attack, misidentify a backwall echo or produce unstable values on rough surfaces. Good practice is about making those limitations visible before the report goes out.

What an ultrasonic thickness reading actually measures

A conventional thickness gauge measures the time taken for an ultrasonic pulse to travel through a component and return from the far surface. The instrument converts that transit time into thickness using the programmed sound velocity. In simple, uniform plate, the method is straightforward. In service-aged plant, it often is not.

The displayed number represents the sound path at the point being tested. It is not automatically the minimum thickness across a vessel shell, pipe circumference or corroded area. If the probe footprint misses a narrow pit, the result can be sound but incomplete. That distinction matters when inspection data is being used for fitness-for-service assessment, maintenance planning or statutory records.

Material velocity is equally important. Most gauges are set up for carbon steel, but alloy composition, temperature, coatings and material condition can change the effective result. A gauge that has been zeroed but not velocity-verified can produce readings that are consistently wrong.

Know the job before selecting the gauge

Start with the inspection question. Is the task a general corrosion survey, spot checks at nominated condition monitoring locations, or mapping a known damage mechanism? The answer determines the probe frequency, diameter, measurement grid and whether a single-element gauge is enough.

For clean carbon steel with accessible surfaces, a standard dual-element probe is usually practical. Its separate transmit and receive elements help manage the near-surface dead zone and make it suitable for many corrosion applications. Higher frequencies can improve resolution on thinner sections, but attenuation increases in coarse-grained, coated or noisy material. Lower frequencies are more forgiving through difficult materials, though small local pits may be less clearly resolved.

Where thickness variation needs to be documented across an area rather than found by spot readings, encoded scanning changes the quality of the evidence. A controlled scan path and position data can show a profile of the wall rather than a collection of isolated values. It takes more setup, but it reduces the chance that two technicians sample different locations and reach different conclusions.

Ultrasonic thickness measurement guide: setup before testing

The instrument should be configured and checked at the work area, not assumed to be correct because it worked on the previous job. Temperature, probe condition, battery state and the surface being inspected all affect practical performance.

First, select the correct measurement mode. Echo-to-echo mode can be useful where a stable coating is present because it measures between successive backwall echoes and can exclude coating thickness. It is not a universal answer. Weak multiple echoes, heavy corrosion or unsuitable geometry can make the mode unreliable. If coating thickness is relevant to the result, use a mode and procedure that includes it, or measure the coating separately.

Then set the material velocity from a verified reference. A calibration block of the same material and similar thickness is best. If that is not available, use a representative area with a known thickness confirmed by an appropriate method. Do not treat a handbook velocity as proof for critical work.

A proper field check includes a zero check and a two-point calibration across the expected thickness range. A one-point check may confirm probe delay, but it does not adequately establish accuracy across a broad range. Recheck calibration whenever the probe is changed, the instrument setting is altered, conditions change substantially, or readings begin to behave unexpectedly.

Surface condition is part of the measurement system

Loose scale, dirt, failed paint and weld spatter interfere with coupling. They can create unstable readings or make the instrument lock onto noise. Remove only what is needed to obtain consistent coupling and comply with the site coating and surface-preparation requirements. Excessive grinding can alter the component, particularly on thin wall or localised corrosion.

Use enough couplant to exclude air, but not so much that the probe floats and shifts from the nominated point. In vertical and overhead work, a higher-viscosity couplant may hold better. In hot-service work, use a suitable high-temperature couplant and probe, then allow for the effect temperature has on velocity and probe performance.

Probe pressure should be firm and repeatable, not excessive. Rocking the probe can help locate the strongest response on a curved surface, but record the measurement point consistently. On small-diameter pipe, a probe designed for curvature or a suitable shoe can improve contact and reduce operator variation.

Finding the real minimum wall

The most common thickness error in corrosion inspection is not an incorrect displayed value. It is reporting a valid value from the wrong place.

At each condition monitoring location, begin by establishing the nominated position from drawings, weld references, clock position or fixed dimensions. Then search around that point where the procedure permits. A small cross pattern is often enough for general thinning; an expanded grid is needed when pitting, under-deposit corrosion, flow-accelerated corrosion or local damage is suspected.

Watch the signal, not only the number. A stable, repeatable backwall response is more credible than a value that jumps as the probe moves a few millimetres. If the gauge offers A-scan display, use it where a result is questionable. It helps distinguish a genuine backwall echo from a multiple, a coating interface, mode-converted energy or noise.

For localised corrosion, move the probe slowly and deliberately. The lowest reading is not automatically the reportable minimum unless it can be repeated and located. Confirm it from more than one direction, clean the area if necessary, and assess whether the probe face is bridging a narrow pit. A large probe can average over a small defect, while a smaller probe may better identify the local condition but be more sensitive to roughness.

On pipework, do not assume the bottom dead centre is the only concern. The likely damage location depends on the process, fluid, flow regime, supports, insulation condition and geometry. Elbows, reducers, injection points, dead legs, low points and areas downstream of disturbances deserve inspection plans that reflect the damage mechanism rather than a generic grid.

When a spot gauge is not enough

A hand-held thickness gauge is efficient for routine monitoring locations and straightforward access. It becomes less effective when the inspection needs repeatable area coverage, traceable position data or a defensible corrosion profile.

This is where mechanised or encoded scanning earns its setup time. A scanner can hold the probe orientation, control scan direction and capture data against distance. For an NDT contractor, that can mean less rework between shifts and clearer evidence for the client. It also removes some of the dependence on one operator's hand technique.

The trade-off is practical. Scanner selection must suit the geometry, access, coating condition and available setup time. A large, expensive system is not always the right answer for a short pipe spool or a tight plant location. Purpose-built, modular hardware is often more useful when several inspection jobs need to run at once. PAUT.Tech applies that same field-first thinking to scanning hardware for ultrasonic inspection work.

Common causes of misleading readings

Poor coupling is obvious when readings drop out, but other issues are easier to miss. Thick or variable coatings can shift results when the wrong mode is used. Laminations can generate an early reflector that looks like a backwall. Coarse-grained material can attenuate and scatter sound until the apparent echo is unreliable.

Geometry creates its own traps. Curved pipe, small radii, weld caps and transitions can deflect sound or prevent the probe from seating properly. Near welds, use an appropriate scan plan and recognise that the material and surface condition may differ from the parent plate. A thickness gauge is not a substitute for a weld examination method when the question is weld integrity.

Incorrect velocity, worn probe faces and damaged cables can produce believable but repeatable error. That is why calibration checks must be recorded, and why a questionable result should be verified before it drives a repair decision. If there is no credible backwall response, report the limitation rather than forcing a number into the data sheet.

Record data so another technician can trust it

A useful thickness record identifies the component, location reference, date, instrument, probe, calibration standard, measurement mode and result. It should also capture relevant qualifiers such as coating condition, surface preparation, temperature, access limitations and whether the reading was a spot value or mapped minimum.

Photographs and marked-up sketches are often as valuable as another decimal place. They allow the next inspection team to return to the same location and tell whether apparent wall loss is real, localised or simply the result of measuring somewhere different.

The best thickness measurement is not the fastest one or the lowest one. It is the reading that is traceable, repeatable and appropriate to the decision it supports. When the condition is unclear, spend the extra few minutes checking the response, expanding the scan area or selecting a better method. That time is cheaper than building a maintenance decision on a number that only looked certain.