TOFD rail inspection means applying time-of-flight diffraction, an ultrasonic technique that times signals diffracted from crack tips, to rail and rail welds. ASTM E2373 says TOFD "has proven effective on thicknesses from 9 mm to 300 mm" in welded joints. It is not a routine rail flaw test method: in-track testing under 49 CFR 213.237 relies on pulse-echo ultrasonic testing (UT).
This article explains the technique itself: the four signals on a TOFD image, the depth equation with a worked example, the two dead zones and what published trials say about sizing accuracy. It then sets out the standards that govern TOFD and where the method does and doesn't fit on rail. For a side-by-side comparison of TOFD and phased array on rail welds, see our separate article on PAUT vs TOFD for rail weld inspection.
What is TOFD and how does it work?
Time-of-flight diffraction (TOFD) is a pitch-catch ultrasonic technique in which a transmitting probe and a receiving probe sit on the same surface, one each side of the weld, and the instrument measures when diffracted signals from a flaw's tips arrive. Depth comes from timing and geometry, not from how loud the echo is.
The probes are angled longitudinal-wave probes on wedges, fixed at a constant separation in a scanner. TWI's Kenzie and Speck describe the usual setup as "two angled compression wave probes (typically between 2 to 10MHz frequency)" in transmit-receive mode. Each probe's beam diverges widely, so one probe pair covers much of the wall thickness in a single pass along the weld.
When the sound strikes the tip of a crack, the tip acts as a secondary source and scatters a weak signal in all directions. Some of that signal reaches the receiver. Because the arrival time depends only on where the tip is, TOFD can measure the through-wall height of a crack even when the crack face is badly oriented for a pulse-echo reflection.
ASTM E2373/E2373M-19(2024), the ASTM practice for TOFD, puts the core idea in one sentence:
"TOFD is a nondestructive ultrasonic examination technique that is not based on amplitude response." (ASTM E2373/E2373M-19(2024), section 4.2)
The same section adds that "sufficient sensitivity is required to identify indications for evaluation". Amplitude-independent sizing is not the same as gain-independent detection, and it isn't operator-independent either.
The probe placement is a common point of confusion. The original version of this article said the two probes sit "on opposite sides of the rail", and its FAQ said TOFD needs "access to both sides of a rail". That's wrong. Both probes are on the scanning surface, straddling the weld or the region of interest. They are on opposite sides of the weld centerline, not on opposite faces of the part.
Key takeaway: TOFD is a two-probe, same-surface technique that sizes flaws from tip-signal timing.
What signals appear on a TOFD scan?
A TOFD A-scan from sound material shows two reference signals, the lateral wave and the back-wall echo, and any flaw adds tip-diffraction signals between them. Reading a TOFD image means placing every indication relative to those two references.
| Signal | Path | Arrives | What it tells the analyst |
|---|---|---|---|
| Lateral wave | Straight from transmitter to receiver, just under the scanning surface | First (shortest path) | The time zero for depth; its loss can mean poor coupling or a surface-breaking crack |
| Upper tip diffraction | Transmitter to the flaw's upper tip to the receiver | After the lateral wave | Depth of the top of the flaw |
| Lower tip diffraction | Transmitter to the lower tip to the receiver | After the upper tip | Depth of the bottom of the flaw; the gap between tips gives height |
| Back-wall echo | Reflected from the far surface | Last of the main signals | Wall thickness; a break in it can mean a flaw reaching the back wall |
| Mode-converted signals | Part of the path as a shear wave | After the back wall | Usually outside the evaluation window; must not be read as flaws below the back wall |
Two practical points help when reading the image. First, phase matters. Baker Hughes notes that the lateral wave and back-wall echo are "180° out of phase", that the upper-tip signal is out of phase with the lateral wave, and that the lower-tip signal has the same phase as the lateral wave. Phase is one of the cues an analyst uses to tell a crack's upper tip from its lower tip.
Second, the time-to-depth scale is not linear. Baker Hughes, a probe and instrument maker, puts it as "not linear but follows a hyperbolic function". Equal time steps represent much larger depth steps near the scanning surface than near the back wall. The worked example below shows how large that effect is.
Most TOFD scans are non-parallel scans: the probe pair travels along the weld, and the image plots scan distance against depth. That image gives flaw length and through-wall height, but not the flaw's exact position across the weld. TWI notes that full characterization of an indication may need "additional scans" to place it in the remaining dimension.
Key takeaway: every TOFD indication is read against the lateral wave above and the back-wall echo below.
How is flaw depth calculated in TOFD?
For a flaw tip midway between the probes, depth is d = √[(c·t/2)² − S²], where c is the longitudinal velocity, t is the tip signal's time of flight and S is half the probe center spacing. It comes from the right-triangle geometry of the two sound paths, not from the pulse-echo formula.
The derivation is short. With the probes 2S apart and the tip at depth d on the centerline, each leg of the path from probe to tip has length √(S² + d²). The total path is twice that, so c·t = 2√(S² + d²). Solving for d gives the equation above.
In practice the probe delay in the wedges is not known exactly, so time is measured from the lateral wave instead of from the pulse. The lateral wave travels the straight 2S path, so its arrival is t_L = 2S/c. A tip signal arriving Δt after the lateral wave has t = t_L + Δt. This cancels the wedge delay, which is why the lateral wave is the depth reference.
Three conditions sit behind this equation:
- The tip is on the centerline between the probes. If it is off to one side, the path is longer and the centered equation reads it deeper than it is.
- The velocity is right. ISO 10863 bases its material-dependent parameters on steel with a longitudinal velocity of 5,920 ± 50 m/s. A different steel or a weld with a different structure needs its own velocity.
- The probe center spacing is fixed and measured. Baker Hughes notes that the probe center spacing (PCS) "must remain constant" during the scan. A common setup rule is to aim the beam axes to cross at about two-thirds of the wall thickness.
The pulse-echo depth formula, d = c·t/2, gives badly wrong answers when applied to TOFD timing. The worked example shows by how much.
Key takeaway: TOFD depth is a square-root geometry with the lateral wave as time zero, and it assumes the tip sits midway between the probes.
Worked example: sizing a flaw from tip arrival times
On a 25 mm (about 1 in.) plate butt weld scanned with 60 mm probe center spacing, tip signals 1.20 and 1.80 µs after the lateral wave place a flaw from 15.0 to 18.7 mm deep, giving a through-wall height of 3.6 mm. The example is illustrative, worked on a plate weld because that is where TOFD standards apply.
Inputs
| Input | Value | Basis |
|---|---|---|
| Wall thickness, T | 25 mm | Illustrative plate butt weld |
| Longitudinal velocity, c | 5.92 mm/µs | ISO 10863 reference steel, 5,920 m/s |
| Probe center spacing, 2S | 60 mm (S = 30 mm) | 60° probes crossing at two-thirds of T: 2 × tan 60° × 16.7 mm = 57.7 mm, rounded to 60 mm |
| Upper tip signal after lateral wave, Δt₁ | 1.20 µs | Illustrative reading |
| Lower tip signal after lateral wave, Δt₂ | 1.80 µs | Illustrative reading |
Step 1: reference times
- Lateral wave: t_L = 2S/c = 60 / 5.92 = 10.135 µs.
- Back wall: path = 2√(30² + 25²) = 2 × 39.051 = 78.102 mm, so t_BW = 78.102 / 5.92 = 13.193 µs. That is 3.058 µs after the lateral wave.
Both tip readings, 1.20 and 1.80 µs, fall inside that 3.058 µs window, so both tips are between the scanning surface and the back wall. That's the first check on any TOFD reading.
Step 2: depth of each tip
| Tip | t = t_L + Δt (µs) | c·t/2 (mm) | d = √[(c·t/2)² − 30²] (mm) | Pulse-echo formula c·Δt/2, wrongly applied (mm) |
|---|---|---|---|---|
| Upper | 10.135 + 1.20 = 11.335 | 33.552 | 15.03 | 3.55 |
| Lower | 10.135 + 1.80 = 11.935 | 35.328 | 18.66 | 5.33 |
Step 3: height and checks
- Through-wall height = 18.66 − 15.03 = 3.63 mm.
- Both depths are less than T = 25 mm, and the lower tip sits about 6.3 mm above the back wall.
- The pulse-echo formula would put the same flaw 3.6 to 5.3 mm below the surface, about 11 to 13 mm too shallow. That is why TOFD software always uses the diffraction geometry.
Step 4: what an off-center tip does
The calculation assumes the tip is on the centerline. Recomputing the path for a tip truly at 15 mm depth but offset across the weld shows the error:
| Lateral offset of the tip from the centerline | Apparent depth from the centered equation | Error |
|---|---|---|
| 0 mm | 15.0 mm | 0 |
| 5 mm | 15.2 mm | +0.2 mm |
| 10 mm | 15.7 mm | +0.7 mm |
| 15 mm | 16.7 mm | +1.7 mm |
Both tips of a vertical crack at one offset shift by similar amounts, so the height error is smaller than the depth error. An absolute depth used in an engineering assessment, though, needs the flaw's position across the weld. That comes from a parallel scan or a pulse-echo check.
Step 5: why timing errors matter more near the surface
Differentiating the depth equation gives the depth change for a small timing error. For a 0.05 µs error in this setup:
| Tip depth | Depth change per 0.05 µs |
|---|---|
| 3 mm | 1.49 mm |
| 6 mm | 0.76 mm |
| 12 mm | 0.40 mm |
| 20 mm | 0.27 mm |
| 24 mm | 0.24 mm |
The same timing error costs six times more depth accuracy at 3 mm than at 20 mm. This is the hyperbolic scale described above, and it's one reason the zone near the scanning surface is the weak part of a TOFD examination.
Where are the TOFD dead zones?
TOFD has two dead zones: a near-surface zone hidden under the lateral wave and a back-wall zone where tip signals merge with the back-wall echo. Both depend on pulse length and probe spacing, and both are why TOFD is normally paired with pulse-echo UT or phased array.
The near-surface (lateral wave) dead zone
A tip signal that arrives while the lateral wave is still ringing can't be separated from it. Baker Hughes describes the dead zone as "approximately equal to the pulse width of the lateral wave". It gives "at least 3,9 mm at 10 MHz and 5,5 mm when testing with 5 MHz" for a 50 mm PCS.
Those figures can be checked with the depth equation. If the lateral-wave pulse lasts t_p, a tip is hidden when its path exceeds 2S by less than c·t_p. The dead-zone depth is then d = √[(S + c·t_p/2)² − S²].
| PCS (2S) | Pulse length t_p | Dead-zone depth |
|---|---|---|
| 50 mm | 0.2 µs (one cycle at 5 MHz) | 5.5 mm |
| 50 mm | 0.1 µs (one cycle at 10 MHz) | 3.9 mm |
| 60 mm | 0.2 µs (one cycle at 5 MHz) | 6.0 mm |
| 60 mm | 0.1 µs (one cycle at 10 MHz) | 4.2 mm |
The first two rows reproduce the vendor's published figures, which implies a pulse of about one cycle. Real pulses are often longer, so the dead zone can be larger. A wider spacing also makes it larger, which is why thick welds are scanned with several probe pairs at different spacings.
TWI reports that shock-wave probes can cut the lateral-wave effect to "2-3mm", and that lateral-wave straightening and removal in software also help. Even so, TWI warns that surface-breaking planar flaws can "fail to be detected by TOFD" in initial scans, and that where very near-surface sizing is critical, TOFD may need a surface method alongside it.
The back-wall dead zone
Near the far surface, the tip signal and the back-wall echo arrive so close together that they can't be resolved. Kenzie and Speck describe TOFD as restricted where "signal time differences are small", for example for "small flaws close to the backwall". BINDT's index of terms lists the back-wall dead zone as defined in BS EN ISO 16828.
Using the worked-example setup and a 0.2 µs pulse, a tip has to arrive at least one pulse length before the back-wall echo to be resolved:
| Tip position across the weld | Deepest resolvable tip | Back-wall dead zone |
|---|---|---|
| On the centerline | 24.1 mm | 0.9 mm |
| 5 mm off center | 23.9 mm | 1.1 mm |
| 10 mm off center | 23.2 mm | 1.8 mm |
| 15 mm off center | 22.2 mm | 2.8 mm |
The back-wall zone grows away from the centerline because the back-wall echo always comes first from the point midway between the probes. A root crack offset to one side of a weld can sit inside that zone.
How the dead zones are covered
ISO 10863 lets TOFD be used "as a stand-alone method or in combination with other" NDT methods, and its testing levels set when extra coverage is required. ASTM E2373 describes TOFD used with the pulse-echo practices E164 and E1961. ASME procedures that apply Section V, Article 4 commonly specify a pulse-echo supplement for the near-surface and back-wall zones (see the standards section; the clause wording needs checking against the current edition).
Key takeaway: the dead zones sit exactly at the scanning surface and the back wall, where many service cracks start, so TOFD is rarely the only technique on a weld.
How accurate is TOFD sizing?
Published trials put TOFD height-sizing error at a mean of about +0.5 mm with a standard deviation of 1.8 mm, against −1.0 to −2.7 mm mean error and about 3 mm standard deviation for the amplitude-based 20 dB drop method. Those figures come from UK collaborative trials summarized by TWI in 2005, on flaw heights of 1.5 to 30 mm.
| Source | Claim | What it is |
|---|---|---|
| Kenzie and Speck (TWI), Inspectioneering Journal, July/August 2005 | TOFD planar-flaw height error: mean +0.5 mm (oversize), standard deviation 1.8 mm; flaw heights 1.5–30 mm | Summary of UK collaborative UT trials |
| Same paper | 20 dB drop height error: mean −1.0 mm, s.d. 3.1 mm, to mean −2.7 mm, s.d. 3.0 mm, for "shop floor" and "laboratory" operators | The amplitude-based comparison |
| TWI time-of-flight diffraction capability page (undated) | "accuracy of greater than ±1mm" across a range of thicknesses; reproducibility "better than 0.5mm" | A testing provider's own claim |
| Baker Hughes, What is time of flight diffraction (TOFD)? (undated) | Depth extents to "approximately ±0,2 mm", for diffraction centers midway between the probes | A probe and instrument maker's claim |
| ASTM E2373, 4.4 | TOFD may be used with pulse-echo examinations to "improve sizing estimates" | Standard practice; gives no number |
Read these numbers with their conditions attached:
- The ±0.2 mm figure applies to tips centered between the probes. The same vendor page warns that the deviation grows with "lateral displacement of the diffraction centre, smaller PCS, and decreasing time of flight differences". The worked example shows 0.7 mm of depth error at a 10 mm offset.
- The trial figures are for planar flaws in welds of regular geometry. They don't transfer to a part where the back wall is curved or the sound path is unknown.
- Very shallow flaws size poorly. Kenzie and Speck note that flaws near the scanning surface are harder to detect because the lateral wave "often occupies several millimetres of the depth axis".
- Interpretation still matters. The same authors write that "a high degree of operator skill and experience" is needed to interpret and report TOFD findings.
No probability-of-detection or sizing study for TOFD on rail was found in a search of public sources (October 2026). The accuracy figures above should not be quoted for rail.
Key takeaway: TOFD is the most accurate common UT technique for embedded crack height in plate and pipe welds, but every accuracy number has a geometry and a flaw position behind it.
Which standards govern TOFD?
The main TOFD documents are ASTM E2373 (US practice), ISO 16828 (general principles), ISO 10863 (TOFD of welds), ISO 15626 (acceptance levels) and ASME Section V, Article 4, Mandatory Appendix III (ASME procedures). None of them is a rail standard, and only ISO 15626 sets acceptance levels.
| Document | What it covers | Current status (checked October 2026) | Rail scope? |
|---|---|---|---|
| ASTM E2373/E2373M-19(2024), Standard Practice for Use of the Ultrasonic Time of Flight Diffraction (TOFD) Technique | Requirements for developing TOFD procedures; "typically applied to welded joints in carbon steel"; proven on 9–300 mm (0.375–12 in.); other ranges "shall be demonstrated" | Active; 2019 edition reapproved 2024 (store page last updated Dec 5, 2024) | No |
| ISO 16828:2025, Ultrasonic testing — Time-of-flight diffraction technique as a method for detection and sizing of discontinuities | General principles of TOFD for low-alloy carbon steel components; refers to ISO 10863 for welds and ISO 15626 for acceptance | Current; published July 2025, replacing ISO 16828:2012 (AFNOR listing) | No |
| ISO 10863:2020, Use of time-of-flight diffraction technique (TOFD) for welds | Semi- or fully automated TOFD of fusion-welded joints at least 6 mm thick; "full penetration welded joints of simple geometry in plates, pipes, and vessels"; testing levels A–D; "does not include acceptance levels" | Current (EN ISO 10863:2020 replaced the 2011 edition) | No |
| ISO 15626:2018, Time-of-flight diffraction technique (TOFD) — Acceptance levels | Acceptance levels for TOFD of full-penetration welds in ferritic steels 6–300 mm thick, matched to ISO 5817 quality levels, for indications classified per ISO 10863 | Current; reported confirmed by ISO in 2023. A revision is under way (CEN project approved Nov 14, 2024; ISO committee draft for a third edition) | No |
| ASME BPVC Section V, Article 4, Mandatory Appendix III, TOFD technique | TOFD requirements within the Article 4 weld examination rules; Article 4 applies "unless modified by this Appendix" | Section V current edition is 2025. Appendix III content is reported in secondary sources only (CONFIRM) | No |
Some details on each:
ASTM E2373. It is a procedure-writing practice. Section 4.1 calls TOFD "a tool for detection and sizing of discontinuities", and section 4.4 lets it stand alone or work with the pulse-echo practices E164 and E1961. It sets no accept or reject criteria; those come from the referencing code or contract.
ISO 10863 and ISO 15626. These two work as a pair. ISO 10863 says how to carry out and classify a TOFD examination of a weld, and ISO 15626 says which classified indications are acceptable for each ISO 5817 quality level. ISO 10863 also says TOFD "can also be used on other types of materials that exhibit low ultrasonic attenuation (especially that due to scatter)". That clause matters for cast manganese rail crossings, as the next section explains. ISO 15626:2018 replaced ISO 15626:2011 (EN ISO 15626:2013), which itself replaced EN 15617:2009.
ASME Section V, Article 4, Appendix III. Secondary sources describe Appendix III as the Article 4 TOFD appendix, and an ASME interpretation from the 2010 edition refers to it. A published vendor procedure cites paragraph III-471.9 for pulse-echo coverage of the TOFD blind zones. One secondary summary of the 2025 edition says TOFD "can now be used as a stand-alone inspection method in certain conditions". ASME Code Case 2235, which allows UT in place of radiography for pressure vessel welds, is the usual route by which construction codes call TOFD into use. Treat all Appendix III paragraph numbers and the 2025 change as CONFIRM items until checked against the purchased code.
Acceptance criteria. TWI's TOFD FAQ says that "specific acceptance criteria for defects have yet to be established". That wording predates ISO 15626, so it is out of date for ISO-based work. For ASME work, acceptance still comes from the construction code or code case, not from Section V.
For how examination standards differ from acceptance criteria in UT generally, see our guide to UT acceptance criteria standards.
Where does TOFD rail inspection fit, and where doesn't it?
TOFD does not fit routine rail flaw testing, it is generally unsuitable for rail welds and cast crossings, and its rail use so far has been in research. That is a fit judgment based on the standards' scopes and the rail's geometry, not a ban: no rule forbids TOFD, but no rail standard or published rail procedure found adopts it.
Routine rail flaw testing
Internal rail testing under 49 CFR 213.237 is done by test cars and walking carts that scan the rail head with pulse-echo probes, typically 0°, 37–45° and 70°. The rule requires internal rail inspection but names no technique, and Part 213 doesn't mention ultrasonic testing at all. Nothing in it calls for TOFD.
TOFD isn't a practical fit for that job. A rolling search covers miles of rail per shift looking for transverse, longitudinal and web defects. TOFD needs a fixed probe pair in a scanner, a regular back wall and an image that an analyst reads indication by indication. Our guide to how ultrasonic testing works for railways covers the probe set the test cars actually use.
Rail welds
Thermite and flash-butt welds are where TOFD is most often proposed for rail, and where the original version of this article oversold it. That version said TOFD is "widely used" for rail welds. A search of public sources (October 2026) found no published rail weld TOFD procedure, validation study or acceptance criterion.
The geometry is the main reason. ISO 10863 is written for welds of "simple geometry in plates, pipes, and vessels". A rail has a crowned head, a narrow web and a tapered foot. A TOFD pair on the running surface has no flat back wall to frame a depth window, and a thermite weld's cast collar blocks probe placement. Our PAUT vs TOFD for rail weld inspection article works through that geometry and the rail weld standards in detail.
The dead zones make it worse on rail. Detail fractures start at or near the running surface, which falls inside the lateral-wave dead zone of a probe pair on the head. Thermite weld failures cluster in the web, web-base fillet and base, where a head-mounted pair has no regular back wall. Rail welds in track are tested with conventional multi-angle UT to the owner's specification; see our articles on flash-butt weld inspection and aluminothermic weld inspection.
Crossings and frogs
Cast austenitic manganese steel (AMS) frogs and crossing diamonds are a poor fit for TOFD for a material reason, not only a geometric one. FRA's RR 22-25 describes AMS as "highly attenuative" and coarse-grained, and says revenue-service examination of these castings "currently use[s] only visual inspections". TOFD relies on weak tip signals carried by longitudinal waves, which grain scatter swamps. ISO 10863 extends TOFD to other materials only where attenuation is low, "especially that due to scatter".
Crossing research has gone in a different direction. TWI's RailSAFT work used low-frequency transducers with synthetic aperture focusing (SAFT) on cast crossings, not TOFD. Our railroad switch and crossing inspection article covers what is inspected at turnouts today.
Research use
The main published rail TOFD work found is from the University of Liverpool. Zahran and Al-Nuaimy (Insight, 2004) combined time-frequency analysis with a neural network to segment TOFD images of rail-track defects, aimed at fishplate (joint bar) and weld areas. Their abstract frames this as a future possibility: "It is anticipated that" TOFD could be used for comprehensive automatic rail-track inspection. A related 2003 paper by Zahran, Shihab and Al-Nuaimy is listed in the RSSB SPARK research library.
That work is two decades old, and no later field adoption or validation was found. It shows that researchers have tried TOFD on rail, not that TOFD is a rail practice.
Where TOFD can sit near rail work
TOFD is a normal technique on the plate and pipe welds that ISO 10863 and ASTM E2373 describe. On a rail owner's property that can include full-penetration butt welds in steel structures or tanks, where the thickness and geometry fit the standard and the governing code permits it. Those are weld examinations under their own codes, not rail flaw testing.
| Application | TOFD fit | Why |
|---|---|---|
| Rail flaw testing in track (213.237) | Not used | Pulse-echo test cars and carts; TOFD needs a fixed scanner and a regular back wall |
| Thermite rail welds | Generally unsuitable | Rail profile, cast collar, dead zones at the head and no regular back wall; no published procedure |
| Flash-butt rail welds | Generally unsuitable | Rail profile; ISO 10863 is written for fusion welds of simple geometry |
| Cast manganese frogs and crossings | Unsuitable | Coarse-grained, attenuative casting; research uses SAFT |
| Rail TOFD image processing | Research only | Liverpool work, 2003–2004; no field adoption found |
| Plate, pipe and vessel butt welds | Suitable within scope | ISO 10863 (6 mm and up) and ASTM E2373 (9–300 mm proven range) |
Key takeaway: on rail, TOFD is a research technique; its home is regular-geometry butt welds in plate, pipe and vessels.
What are the limits of TOFD?
TOFD's limits are its two dead zones, its need for regular geometry and a known velocity, its weak response in coarse-grained material and its dependence on skilled interpretation. These hold on every job, whatever the instrument's specification says.
- Near-surface flaws. The lateral-wave dead zone runs from about 2–3 mm with shock-wave probes and processing to 4–6 mm with ordinary 5 to 10 MHz pulses. Surface-breaking cracks can be missed on the first scan.
- Flaws at the back wall. Small root flaws near the back wall, especially off the centerline, can sit inside the back-wall zone.
- Thin sections. ISO 10863 starts at 6 mm and ASTM E2373's proven range at 9 mm. TWI says that even with 10 or 15 MHz probes, TOFD is "not recommended for use on thicknesses of less than 13mm".
- Geometry. Baker Hughes notes that TOFD "is almost exclusively applied to planar, parallel-sided components". Curved back walls, section changes and cast collars break the depth reference.
- Position across the weld. A standard scan doesn't locate the flaw across the weld, and an off-center tip reads deeper than it is.
- Material. Coarse-grained and highly attenuative material, such as cast manganese steel, scatters the weak tip signals.
- Interpretation. TOFD images need a trained analyst. Amplitude independence doesn't make TOFD operator-independent.
- Acceptance. TOFD supplies no acceptance criteria. ISO 15626 does for ISO 10863 work, and the construction code or code case does for ASME work.
How EMA Industries applies this
EMA Industries performs ultrasonic rail flaw testing and rail weld inspection for transit agencies, short lines and industrial track owners, working from Southern California and mobilizing nationwide. Rail welds in track are tested with conventional multi-angle UT to the owner's specification. Where a contractor or owner proposes TOFD on a weld, reviewing the written procedure and its technique demonstration against ASTM E2373, ISO 10863 or ASME Section V falls within the scope of EMA's NDT Level III services. See our ultrasonic examination services or contact EMA Industries to discuss a procedure review or a rail weld inspection scope.
Related reading
- PAUT vs TOFD for rail weld inspection, the method comparison for thermite and flash-butt welds
- What is the difference between AUT and PAUT?, where TOFD channels sit in automated weld systems
- Dead zone in ultrasonic testing, the pulse-echo near-surface zone
- Ultrasonic testing formulas, including time-of-flight checks
Frequently asked questions
Is TOFD used for rail flaw detection in track?
Not as a routine method. Internal rail testing under 49 CFR 213.237 is done with pulse-echo probes on test cars and carts, and the rule names no technique. TOFD's fixed probe pair, need for a regular back wall and near-surface dead zone don't suit a rolling search of the rail head.
Where are the probes placed in a TOFD examination?
Both probes sit on the same scanning surface, one each side of the weld or region of interest, at a fixed probe center spacing. They are not on opposite faces of the part. A common setup aims the beam axes to cross at about two-thirds of the wall thickness.
What is the TOFD depth formula?
For a tip midway between the probes, d = √[(c·t/2)² − S²], where c is the longitudinal velocity, t the tip signal's time of flight and S half the probe center spacing. Time is usually measured from the lateral wave, which arrives at 2S/c. The pulse-echo formula c·t/2 gives badly wrong depths for TOFD timing.
How big is the TOFD dead zone?
The near-surface zone is roughly the depth hidden under the lateral-wave pulse. One maker gives at least 3.9 mm at 10 MHz and 5.5 mm at 5 MHz for a 50 mm probe spacing, and shock-wave probes can reduce it to about 2–3 mm. A smaller back-wall zone, about 1 to 3 mm in the worked example here, sits just above the far surface.
Does any standard set acceptance criteria for TOFD?
ISO 15626:2018 sets acceptance levels for TOFD of full-penetration ferritic steel welds 6 to 300 mm thick, applied to indications classified under ISO 10863. ASTM E2373 and ASME Section V set how to examine, not what to accept. No rail standard found sets TOFD acceptance criteria.
Sources
- ASTM E2373/E2373M-19(2024), Standard Practice for Use of the Ultrasonic Time of Flight Diffraction (TOFD) Technique, ASTM store page (sections 1.1, 4.1–4.5; Active)
- EN ISO 10863:2020, Use of time-of-flight diffraction technique (TOFD), scope listing
- EN ISO 15626:2018, Time-of-flight diffraction technique (TOFD) — Acceptance levels, scope and life-cycle listing
- prEN ISO 15626 rev, revision project listing (new project approved Nov 14, 2024)
- ISO 16828:2025, Time-of-flight diffraction technique as a method for detection and sizing of discontinuities, AFNOR listing
- ASME BPVC Section V, Nondestructive Examination, ASME product page
- Kenzie, B., Speck, J., Structural Integrity with TOFD Ultrasonic Inspection, Inspectioneering Journal, July/August 2005 (TWI copy)
- TWI, Weld inspection with ultrasonic time-of-flight diffraction, FAQ
- TWI, Time of flight diffraction capability page
- Baker Hughes, What is time of flight diffraction (TOFD) in ultrasonic testing?, vendor blog (undated)
- BINDT, Index of acronyms: DZ (dead zone)
- Zahran, O., Al-Nuaimy, W., Automatic segmentation of time-of-flight diffraction images using time-frequency techniques: application to rail-track defect detection, Insight 46(6), 338–343, 2004
- Zahran, O., Shihab, S., Al-Nuaimy, W., Time-frequency techniques applied to TOFD for the automation of rail-track inspection, 2003, RSSB SPARK listing
- Wilson, R., Poudel, A., Feasibility of Special Trackwork Inspection Using Ultrasonics (FRA RR 22-25), FRA Research Results, September 2022, ROSA P record
- TWI, RailSAFT: ultrasonic inspection of cast austenitic manganese steel railway crossings
- 49 CFR 213.237, Inspection of rail, eCFR (text checked on Cornell LII in earlier edits)