Ultrasonic Testing Calibration for Rail Flaw Inspection

Ultrasonic testing calibration for rail is the set of checks that proves a flaw detector measures distance, angle and echo height correctly before and during a test. No federal rule prescribes the method: 49 CFR 213.238(b)(2) only requires a qualified operator to "determine that such equipment is performing as intended". The owner's procedure fills that gap.

In practice that procedure has five layers: instrument performance checks (linearity, under ASTM E317-21), velocity and zero on IIW and V2 blocks, probe angle and index checks, sensitivity on a reference rail with known reflectors, and periodic re-checks through the shift that decide whether rail must be tested again. This guide covers each one, with a worked shift log, and ends with what a rail testing RFP should require.

Does any FRA rule set how rail UT equipment is calibrated?

No. Part 213 requires internal rail inspection but names no test method, calibration block, gain setting or check interval. The word "ultrasonic" doesn't appear anywhere in Part 213; "calibration" appears only in the rule for vehicle-mounted gage restraint measurement systems (213.110).

What the rules do set is who is responsible and what counts as a valid test:

Read together, these mean a calibration failure isn't a paperwork issue. If the equipment wasn't performing as intended over a stretch of rail, that stretch wasn't validly searched. Under 213.237(h) the owner then has three options before the test interval runs out: test it again, run trains over it at no more than 25 mph, or replace the rail.

FRA's own research confirms that the industry hasn't filled the gap with a common method. The 2026 FRA report on probability of detection for handheld rail UT (DOT/FRA/ORD-26/07, ENSCO) found that North American railroads have no standardized calibration procedure for handheld units. Operators typically set zero, velocity, angle and gain (to 80% full screen height on a known flaw), usually on an IIW Type 2 block.

Key takeaway: the rule makes the operator answerable for "performing as intended"; the written procedure decides what that means.

On the European side, the in-track UT standard is EN 16729-1. EN 13674-1, which some calibration articles list as a calibration standard, is the rail product specification for Vignole rail of 46 kg/m and above, not an inspection standard. AREMA's Manual for Railway Engineering Chapter 4 includes a section titled Recommended Calibration Rails for Rail Flaw Detection System (4.6). It is a recommended practice, not a federal requirement, and its contents are behind AREMA's paywall.

What instrument checks come before ultrasonic testing calibration?

Instrument performance checks confirm that the flaw detector's screen and gain control are linear, so later calibration steps mean something. They are done less often than daily calibration, typically at intervals set by the procedure and after repair.

The main US practice is ASTM E317-21, Evaluating Performance Characteristics of Ultrasonic Pulse-Echo Testing Instruments and Systems without the Use of Electronic Measurement Instruments. It is active (last updated June 22, 2021). Its scope covers horizontal limit and linearity, vertical limit and linearity, entry- and far-surface resolution, sensitivity and noise, and the accuracy of calibrated gain controls. Its companion for electronic measurements is ASTM E1324-21(2026).

Two points in E317 matter for buyers. First, it sets no pass marks. Section 1.4 says the practice "establishes no performance limits for examination systems"; the using parties must specify any. Second, it sets no schedule. Section 5.7 states:

"No frequency of system evaluation or calibration is recommended or implied. This is the prerogative of the using parties and is dependent on application, environment, and stability of equipment." (ASTM E317-21, section 5.7)

So any claim that "ASTM recommends recalibrating every 4–8 hours" has no basis in E317.

E317 also notes that not every check matters in every application: vertical linearity may be irrelevant for a go/no-go gate alarm, while horizontal linearity matters for depth measurement. Rail testing uses both, because operators judge echo height against a threshold and locate indications by sound path.

Outside the US, EN ISO 22232-1 (instruments), 22232-2 (probes) and 22232-3 (combined equipment) cover the same ground. ISO 22232-3 sets methods, tolerances and acceptance criteria for verifying the instrument, probe and cable together on standard calibration blocks. It replaced EN 12668-3.

Key takeaway: linearity checks are the foundation; E317 tells you how to run them, and your procedure tells you the limits and the interval.

How are velocity, zero and range set on IIW and V2 blocks?

Range calibration sets the screen's distance scale by placing echoes from known radii at their true sound paths, adjusting velocity until the spacing is right and zero (probe delay) until the position is right. Velocity scales the screen; zero shifts it.

Two blocks do most of this work in rail:

Neither block is a sensitivity reference for rail flaws. The IIW block has no flat-bottom holes and no depth ladder. It sets distance and angle; sensitivity comes from the reference rail (next sections).

A V2 range check, step by step

A 70° shear-wave probe on a V2 block shows a repeating echo pattern. With the beam aimed at the 25 mm radius, sound returns from that radius (displayed at 25 mm), then part of it crosses the block to the 50 mm radius and back before returning. Working the geometry, echoes appear at 25, 100 and 175 mm. Aimed at the 50 mm radius, they appear at 50, 125 and 200 mm.

1. Set the instrument's shear velocity to the steel default (about 3,250 m/s for shear waves in carbon steel).
2. Aim at the 25 mm radius and peak the first echo.
3. Adjust zero (probe delay) until the 25 mm echo sits at 25 mm.
4. Adjust velocity until the 100 mm echo sits at 100 mm. Repeat steps 3 and 4 until both are right.
5. Confirm the 175 mm echo. On a 250 mm range, a reading of 175.2 mm is a 0.11% error.

Rail steel and block steel are close but not identical, so a procedure may also require a velocity check on rail itself. For a 0° channel, a common check is the base echo through a known rail section. A 136RE rail is 7-5/16 in. (185.7 mm) tall, so the base echo is a round trip of about 63 µs at 5,890 m/s.

Key takeaway: velocity sets spacing, zero sets position, and the sensitivity reference is never the IIW block.

How are probe angle and index point checked?

The index point (beam exit point) is found by peaking the echo from a radius whose center is marked on the block, and the refracted angle is read by peaking a hole against the engraved angle scale. Both drift as plastic wedges and sled shoes wear.

• Index point: place the probe on the IIW block over the 100 mm radius center mark (or the V2 radius center), move it back and forth to peak the radius echo, and mark the wedge where it lines up with the center mark.
• Refracted angle: peak the echo from the hole (the 50 mm hole or the 1.5 mm hole on the IIW block, the 5 mm hole on the V2) and read the angle at the index mark on the engraved scale.

A wrong angle moves every reported depth and surface position. Suppose a probe sold as 70° actually refracts at 68.5° and an indication appears at an 80 mm sound path. The operator using 70° would report a depth of 27.4 mm (80 × cos 70°) and a surface distance of 75.2 mm. The true values are 29.3 mm deep and 74.4 mm ahead. That 1.9 mm error stays inside a rail head about 49 mm deep on a 136RE section, but it can put an indication in the wrong zone of the head, which matters when a hand verification is being lined up.

Angles also shift with temperature, because the wedge plastic's sound velocity changes and the refracted angle follows Snell's law. Steel itself changes little; the cold-weather problem is mainly in the wedge.

In rail testing the 70° beam is the main tool for transverse defects in the head, and 0° beams find horizontal flaws and check for loss of the base echo. Typical rail probe sets include 0°, 37–45° and 70° channels, with side-looking channels on test cars. Each one needs its own index and angle check.

Key takeaway: check index before angle, and re-check both after any wedge, shoe or wheel probe change.

How is sensitivity set on a reference rail?

Sensitivity for rail is set on a reference rail: a piece of rail of the section being tested, with artificial reflectors at known positions in the head, web and base, so each channel's reference echo can be set to a defined screen height. It is the step that ties gain to the defects the test is meant to find.

A reference rail does what a flat block can't. It reproduces the rail head's curved running surface, the coupling through a wheel probe or sled, and the beam paths through the head, web and base. Reflector types vary by procedure: side-drilled holes and flat-bottom holes in the head, saw cuts or notches at bolt-hole positions, and holes in the base under the web. FRA research libraries of rail samples use mostly flat-bottom holes sized as a percentage of head area, plus natural flaws.

For a handheld probe, the common field method (per ORD-26/07) is to set gain so a known reflector reaches 80% full screen height, then add scanning gain as the procedure specifies. For a cart or test vehicle, practice is to run every channel over a reference or test rail, often a short section mounted in the shop or yard, and confirm each channel alarms on its reflector. On some lines owners keep a test section in track with known reflectors that the car passes at the start and end of a run. These are common practices rather than rules, and the reflector sizes and positions differ from one owner and provider to the next. CONFIRM the reference rail design in the procedure before quoting it.

ORD-26/07 also found that flat probes couple poorly on heavy gage face wear. If the track is heavily worn and the reference rail is new, sensitivity set on the reference rail may overstate what the probe achieves in track. A transfer correction (the dB difference between a reference echo on the block and the same echo on the part) is common in weld UT. For rail it is set by the procedure if used at all.

Key takeaway: the reference rail sets the gain; the procedure must say which reflectors, what screen height and what scanning gain.

How is a DAC or TCG verified for rail work?

A distance-amplitude correction (DAC) curve records how the echo from one reflector size falls with distance; time-corrected gain (TCG) flattens that curve electronically. Verification means re-peaking the reference reflectors and confirming each point still sits on the curve within the procedure tolerance.

Evident's tutorial puts it this way: a DAC curve "graphically compensates for material attenuation, nearfield effects, and beam spreading". The reference points must be beyond the near field and each must be the true peak response. A DAC gives no flaw size; it gives a reference level against which echo height is judged. For the full method, see our guide to the DAC curve in ultrasonic testing.

In rail, DAC and TCG mostly appear in hand verification and in weld testing (thermite and flash-butt welds), where an operator uses a block with side-drilled holes at several depths. Test cars and carts usually use channel gates and thresholds instead of a drawn curve, but the same idea applies: amplitude must be corrected for distance before a threshold is meaningful.

A DAC or TCG verification at a check point should include:

1. Re-peak at least two DAC points, near and far.
2. Compare each against its recorded height.
3. If any point has dropped beyond the tolerance, treat the setup as invalid back to the last good check.
4. If any point has risen beyond the tolerance, re-evaluate the indications recorded since the last good check.

Many weld procedures written to ASME Section V Article 4 use a band of 20% of the reference height or 2 dB for steps 3 and 4. The exact wording varies by edition, so CONFIRM against the edition your contract names. For rail, the tolerance is whatever the owner's or provider's procedure states.

How often should calibration be checked during a shift?

There is no single code rule. ASTM E317 expressly recommends no frequency, Part 213 sets none, and ORD-26/07 found no standard procedure. Intervals come from the owner's specification or the test provider's written procedure.

Common practice, which buyers can write into a contract, looks like this:

The end-of-shift check is the one that protects the data. If it passes, the whole shift stands. If it fails, everything back to the last passing check is suspect. That is why longer gaps between checks cost more when a check fails: the re-test distance grows with the interval.

ORD-26/07 adds a human-factors reason to check often. With a small exception, no two operators in the study produced statistically similar calibration results, and one operator was inconsistent across six occasions. The report warns that "calibration errors can cause excessive false positives and false negatives". For more on what over-sensitive setups produce, see false calls in rail UT.

Key takeaway: calibrate at the start, check at intervals and after changes, and always check at the end.

Worked example: a shift log with mid-shift drift

This example uses illustrative procedure values: a reference echo of 80% full screen height (FSH), a check tolerance of ±2 dB, and a reporting threshold of 50% FSH. A real procedure may differ. The numbers are recomputed from these inputs.

Setup: a walking-speed cart on 136RE rail, 70° forward channel, reference rail with a head reflector.

Step 1: size the drift. The loss in decibels is 20 × log10(80 ÷ 48) = 4.44 dB. That is more than double the 2 dB tolerance. The 2 dB band around an 80% reference runs from about 64% to 101% FSH, and 48% is well outside it.

Step 2: decide how much rail is suspect. The last passing check was at 06:55, so every foot tested between 06:55 and 11:25 is suspect: MP 12.00 to MP 16.40, or 4.40 miles. It doesn't matter when in that window the shoe wore; the record can't show when, so the whole window is invalid.

Step 3: see what the drift could hide. With 4.44 dB of sensitivity lost, every echo displayed at 48 ÷ 80 = 0.60 of its true height. An indication that should have reached the 50% threshold displayed at only 30%. Put the other way, during the bad window only indications whose true height was at least 50 ÷ 0.60 = 83.3% FSH could have crossed the threshold.

Step 4: re-test and compare. After recalibration the indication at MP 13.72 displays at 55% FSH. That matches the first pass: 33% ÷ 0.60 = 55.0%. It was a real, reportable indication that the drifted setup had hidden. It now goes to hand verification by a qualified operator, and if confirmed, the owner applies the remedial action for its type and size under 49 CFR 213.113.

Step 5: check the end of shift. At 16:10 the reference reads 79% FSH, a change of 20 × log10(79 ÷ 80) = −0.11 dB. That is inside tolerance, so everything tested from 11:50 onward stands.

Shift totals: 6.10 miles of new rail tested (4.40 + 1.70), plus 4.40 miles re-tested. If the crew had run out of track time before the re-test, the 4.40 miles would have had to be recorded as not validly tested under 213.241(c)(6), and the owner would have had to apply one of the 213.237(h) options before the interval expired.

If the drift had gone the other way (the reference rising to 100% FSH, about +1.9 dB), the usual approach is different: the rail was tested at higher sensitivity, so nothing was missed, but the indications recorded in that window are re-evaluated at the correct gain. That rule is common in ASME-based weld procedures; for rail it must be written into the procedure.

What calibration records belong with 213.241 test records?

49 CFR 213.241 lists what an internal rail inspection record must contain, and calibration isn't on the list, but the calibration log is the evidence that each tested stretch was a valid search. Keep the two together.

The rule's items, from 213.241(c), are: the date of inspection; the track inspected, with beginning and end points; the location and type of defects found under 213.113; their size, if not removed before the next train; the initial remedial action and its date; and the location of any track not tested under 213.237(g). Records are kept at least two years after the inspection and one year after the initial remedial action, under 213.241(d). Electronic records must have unique author identity, no change once complete, and storage started within 72 hours.

A calibration record that supports those items should show:

213.238(e) and (g) separately require the provider to keep qualification records (name, equipment, date of qualification and latest reevaluation) available to FRA. Records stay with the owner and are available to FRA on request; 213.241 doesn't require sending them anywhere.

What should a rail testing RFP require for calibration?

An RFP should make the provider name its calibration procedure, its references, its check intervals and its re-test rule, because none of these is set by regulation. Without them, two bids can quote the same miles and deliver very different tests.

Ask for:

1. The written procedure covering instrument checks, range, angle, index, sensitivity and DAC or TCG, with tolerances stated in dB and in percent of screen height.
2. The reference rail design: rail section, reflector types, sizes and positions, and how it matches the section and wear in your track. For grooved or unusual sections, a reference piece of the same section.
3. Check intervals: start of shift, during the shift, after changes and at the end of the shift.
4. The re-test rule: what happens when a check fails, how far back the re-test goes, and who pays for it.
5. Calibration logs in the deliverable, matched to the mileposts tested, with the untested-track list required by 213.241(c)(6).
6. 213.238 evidence: the provider's training program, and each operator's qualification on the specific equipment type being used.
7. Instrument evaluation dates under ASTM E317 or ISO 22232, and probe change records.
8. Hand verification method: probes, angles and the DAC or reference setup used to confirm car or cart indications.

For the wider bid structure, see the rail inspection RFP guide. For what to expect from the equipment itself, see ultrasonic flaw detectors for rail inspection.

What calibration can't do

Calibration proves the equipment responds correctly to known reflectors; it doesn't prove every real defect will be found. Engineers quoting calibration as a guarantee of detection should keep these limits in mind:

• Real defects aren't drilled holes. A tight or compressed crack reflects less than an open reflector of the same size, and a defect at an unfavorable angle can return little energy to the probe.
• Masking. Shells and head checks on the running surface block sound from reaching defects beneath them. A perfect calibration doesn't change that; the record should show where the surface prevented a valid search.
• Coupling in track. The reference rail is clean and often new. Worn, lubricated, dirty or corrugated rail couples differently, and ORD-26/07 found flat probes couple poorly on heavy gage face wear.
• Coverage. Calibration doesn't extend the beam's reach. Vehicle testing covers the head, the web and the part of the base directly beneath the web, not the whole base.
• Between checks. A check only proves the setup at that moment. The end-of-shift check limits the damage of drift; it can't recover rail that wasn't re-tested.
• Operator judgment. ORD-26/07 found large differences between operators' calibrations. A procedure narrows that spread but doesn't remove it.

If you need help reading the screen once calibration is set, see how to read a flaw detector and our manual for ultrasonic testing. For test intervals under 213.237, see how often rail must be tested.

How EMA Industries applies this

EMA Industries performs ultrasonic rail flaw testing for transit agencies, short lines and industrial track owners, and provides NDT Level III services. We're based in Southern California and mobilize nationwide. When we scope a rail test, we agree on the calibration procedure, the reference rail, the check intervals and the re-test rule with the owner before the first shift, and we deliver calibration logs matched to the mileposts tested. To discuss a program or an RFP, see our ultrasonic rail testing services or contact us.

Frequently asked questions

How is ultrasonic testing equipment calibrated for rail inspection?

In layers. The instrument's linearity is checked under a practice such as ASTM E317-21. Velocity and zero are set on IIW or V2 blocks, then the probe's index point and refracted angle are checked. Sensitivity is set on a reference rail with known reflectors, and the setup is re-checked during and at the end of the shift.

What is the difference between an IIW block and a reference rail?

An IIW block is a flat steel block with radii and holes for setting distance, angle and index. A reference rail is a piece of the rail section being tested, with artificial reflectors in the head, web and base, for setting sensitivity. You need both: the IIW block can't reproduce the rail head's shape or the coupling through a wheel probe.

How often should rail UT calibration be checked?

At the start and end of each shift, after any probe, wedge or operator change, and at intervals set by the written procedure. No federal rule or ASTM standard sets a fixed number of hours; ASTM E317 states that no frequency is recommended or implied.

What happens if a calibration check fails mid-shift?

All rail tested since the last passing check is suspect. The fault is fixed, the equipment recalibrated, and that stretch re-tested. If it can't be re-tested, it is recorded as untested under 49 CFR 213.241(c)(6), and the owner must re-test it, limit speed to 25 mph, or replace the rail before the 213.237 interval expires.

Does FRA certify rail flaw detector calibration?

No. FRA doesn't certify equipment or technicians. Under 49 CFR 213.238 the employer qualifies each operator on each equipment type, and the qualified operator determines that the equipment is performing as intended.

Sources

• 49 CFR 213.237, Inspection of rail, eCFR (read on Cornell LII)
• 49 CFR 213.238, Qualified operator, eCFR (read on Cornell LII)
• 49 CFR 213.241, Inspection records, eCFR (read on Cornell LII)
• 49 CFR 213.113, Defective rails, eCFR
• Bruzek, Einbinder and Tunna (ENSCO), DOT/FRA/ORD-26/07, model-assisted probability of detection for handheld rail UT, FRA, July 2026
• ASTM E317-21, Evaluating Performance Characteristics of Ultrasonic Pulse-Echo Testing Instruments and Systems, ASTM store page (scope and significance)
• ASTM E1324-21(2026), Measuring Some Electronic Characteristics of Ultrasonic Testing Instruments, ASTM store page
• EN ISO 22232-3:2020, Combined equipment, catalog listing
• Calibration Block No. 2 (ISO 7963), PH Tool product page (vendor)
• IIW Type 1 block, Sonatest product page (vendor)
• DAC/TVG tutorial, Evident (vendor)
• AREMA Manual for Railway Engineering, Chapter 4 table of contents, AREMA (section titles only)