Rail Profile Types and Their Impact on Ultrasonic Inspection

Rail profile types change ultrasonic rail testing in measurable ways. The rail height sets when the 0° base echo returns: about 57.1 µs on 115RE against 63.0 µs on 136RE. Head depth sets how far a 70° beam travels to cover the head, and head width and wear decide where the probes must sit to reach the gauge corner.

This guide works through the North American AREMA sections, the older sections still found on short lines and industrial track, and the girder and grooved rail used in street-running transit. For each one it shows what changes in the test setup, with worked timing examples. It ends with what head wear does to the same numbers, and why a fixed probe setup can search "the wrong volume of material" on a worn head.

Two related topics have their own pages. What the AREMA rail purchase specification requires is covered in our guide to AREMA standards for steel rails. How the instrument is calibrated on blocks and reference rail is covered in ultrasonic testing calibration for rail inspection.

What are the main rail profile types in North American track?

Most North American railroad track uses AREMA tee-rail sections, named by nominal weight in pounds per yard plus "RE": 115RE, 119RE, 132RE, 136RE and 141RE are the sections a US mill lists today. Older track also carries earlier ARA and AREMA designs, and street-running transit uses grooved (girder) rail made to European standards.

AREMA publishes its recommended sections in the Manual for Railway Engineering (MRE), Chapter 4, Part 1, Section 1.1, "Recommended Rail Sections" (2025 table of contents). The MRE is paywalled, so the dimensions below come from published mill and supplier data, not from the AREMA drawings themselves.

The word "profile" is used in two ways, which causes confusion. It can mean the whole cross-section (the section a mill rolls), or only the shape of the running surface and gauge corner (the head profile that grinding restores). Both matter to the test: the section sets the timing and the head profile sets the coupling and the beam entry.

Key takeaway: identify the exact section, not the weight class, before setting up a rail flaw test.

Which rail dimensions matter to an ultrasonic test?

Five dimensions drive an ultrasonic rail test setup: rail height, head width, head depth, web thickness and base width. Height sets the 0° base-echo time, head depth sets the 70° path to the bottom of the head, head width sets how many beams the head needs, and the web and base decide what the 0° and angled channels see below the head.

Steel Dynamics publishes nominal dimensions for the five sections it rolls in its Rail Products brochure (08-2023):

Source: Steel Dynamics, Rail Products. The 136RE row agrees with ArcelorMittal's 136RE (TR68) sheet, which gives 185.70 mm height, 74.60 mm head and 49.20 mm head depth, and with the 136 lb AREMA drawing in Harmer Steel's catalog. The 115RE row also matches the Harmer drawing (6-5/8 in. height, 2-23/32 in. head, 1-11/16 in. head depth).

Three things in that table surprise people:

• Heavier does not always mean a deeper head. 132RE is heavier than 119RE, but its head is 1/8 in. shallower (1.7500 against 1.8750 in.). It is wider and the rail is taller instead.
• Head width does not track weight either. 119RE has a narrower head than 115RE.
• The head depth is measured to the fishing surface at the edge of the head, where the joint bar bears, not at the centerline. The head is thicker over the web than at its edge, so treat any 70° calculation from these figures as approximate.

Key takeaway: height, head depth and head width each change a different part of the setup, and the order of sections by weight is not the order by any of them.

How does rail height change the 0° base-echo time?

The 0° (straight-beam) probe times the echo from the bottom of the rail, so the base-echo time equals twice the rail height divided by the longitudinal velocity. In steel at about 5,890 m/s (0.2320 in./µs), each inch of rail height adds 8.62 µs to the round trip, and each 1/16 in. adds 0.54 µs.

The base echo is the 0° channel's reference. Its loss signals either a reflector in the path (a horizontal split head, a head-web separation, a split web) or a loss of coupling. That is why test systems watch it with a gate, and the gate must sit where the base echo actually arrives on the section under test.

The velocity used here is Evident's published longitudinal value for 1020 steel, 5,890 m/s; many gauges default to 5,920 m/s, which shortens each time by about 0.5%. Times are steel-only, with no wedge, delay line or water path, which every real system adds and calibrates out.

Worked example: 0° base-echo times by section

Round trip t = 2 × H ÷ 0.2320 in./µs.

Check the 136RE line: 2 × 7.3125 = 14.625 in.; 14.625 ÷ 0.2320 = 63.04 µs.

Now suppose a base-echo gate is set on 136RE and runs from 61 to 65 µs. On 132RE the echo at 61.42 µs is barely inside it. On 119RE (58.73 µs) and 115RE (57.11 µs) it falls before the gate opens, so the system sees no base echo and reports continuous loss of bottom. On a line where 100RE meets 136RE at a compromise joint, the base echo jumps by 11.31 µs across the joint.

The reverse error is worse. A gate set wide enough to catch every section can also catch a reflector above the base and treat it as the base echo. For that reason the gate should be set on the section actually being tested, from the reference rail for that section, and the test record should say which section was assumed.

Key takeaway: a 0° setup made on one section can report false loss of bottom on another; set the gate on the section in track.

How does head depth change 70° coverage of the rail head?

A 70° shear-wave probe looks for transverse defects in the head, and the deeper the head, the longer the beam must travel to reach its bottom: on 136RE about 5.67 in. of sound path, against 4.93 in. on 115RE. The beam is angled 70° from the vertical, so it drops only 0.342 in. for each inch it travels along its path.

The geometry is simple. To reach depth D, the 70° beam travels D ÷ cos 70° (D × 2.9238) along its path and moves D × tan 70° (D × 2.7475) along the rail. The round-trip time uses the shear velocity, about 0.1280 in./µs (3,250 m/s).

Worked example: 70° path to the bottom of the head

Check the 136RE line: 1.9375 × 2.9238 = 5.665 in. of path; 2 × 5.665 ÷ 0.1280 = 88.51 µs; 1.9375 × 2.7475 = 5.323 in. along the rail.

Two setup points follow. First, a 70° head gate set for 136RE (to about 88.5 µs) is about 11 µs too long on 115RE. The extra time looks past the bottom of the head into the head-web fillet, where geometry echoes appear. Second, a gate set for 115RE (77.1 µs) and used on 141RE stops about 0.38 in. short of the bottom of a 2.16 in. head, which leaves the lower head unsearched by that channel.

The 0° channel and the 70° channel also need different windows on the same rail. On 136RE the 0° base echo returns at 63.04 µs while the 70° head gate must run to about 88.5 µs, so they can't share one gate.

How wide is one 70° beam?

One 70° beam covers only part of the head width, so systems use several head channels and offsets. A simple ray model of a 0.5 in., 2.25 MHz shear beam (−6 dB half-angle about 3.35°) gives a beam about 0.26 in. wide at a 2.2 in. path and about 0.66 in. wide at 5.67 in., the bottom of a 136RE head. That ignores the element width and is only a guide, but it shows why a single centered 70° probe can't cover a 2.94 in. head from gauge corner to field side.

A wider head (141RE at 3.1875 in.) needs more lateral coverage than a narrower one (119RE at 2.6563 in.). Channel layouts vary by system, and no public source sets a standard North American channel set. A common arrangement is 0°, 37° to 45° and 70° beams, with 70° probes looking forward and backward, plus side-looking probes for the gauge corner.

Key takeaway: head depth sets the 70° gate length and head width sets how many beams it takes; both change with the section.

What changes on older rail sections still in track?

Older sections such as 90 ARA-A, 100 ARA-A, 100RE and 133RE are shorter, with smaller heads, and they often share a line with modern rail through compromise joints and relay rail. None of the four is in the five-section list Steel Dynamics publishes today, but a rail supplier's catalog notes that relay rail "even those not currently manufactured" can still be supplied.

Short lines, industrial spurs and plant tracks are where these sections turn up. Harmer Steel's catalog describes ARA-A and AREMA rails as "primarily designed for railroad tracks, including sidings and spur tracks", and says that relay (second-hand) rail is "often used for industrial sidings and spur tracks". AREMA treats second-hand rail in its own section of Chapter 4: Section 3.12, "Inspection and Classification of Second Hand Rail for Welding" (2025 table of contents; content not reviewed here).

Dimensions from the Harmer Steel catalog drawings (2014):

Each row was checked by adding head depth, fishing height and base thickness, which must equal the height. For 90 ARA-A: 1-15/32 + 3-5/32 + 1 = 5-5/8 in.

What the older sections change in the test:

1. Different gates. 100 ARA-A and 100RE have the same 6 in. height and so the same 51.72 µs base echo, but their heads differ by 3/32 in. in depth. A 0° check alone can't tell them apart.
2. Different joints. The Harmer drawings give standard drilling of 2-11/16 × 5-1/2 in. for 90 ARA-A, 100 ARA-A and 100RE, against 3-1/2 × 6 in. for 115RE and 3-1/2 × 6 × 6 in. for 133RE and 136RE. Bolt-hole echoes therefore appear at different spacings, and the operator reading them needs to know which drilling to expect.
3. Mixed sections on one line. Plug rails and relay rail of another section create short lengths where the setup no longer matches. Each change of section is a place to note in the test record.
4. More wear, more history. Relay rail has already lost some of its head, so the section on paper and the section in track differ (see the wear section below).

A field check helps when the section isn't known. The mill brand rolled on the web usually names the section and year. Where it is unreadable, measure height, base width and head width, and match them to a section table before setting up.

Key takeaway: on older track, confirm the section at each change of rail before trusting a setup made elsewhere on the line.

How do girder and grooved rail change the test?

Grooved (girder) rail used in embedded street track has a narrow running head beside a flangeway groove and a lip, a tall section with a wide base, and a base that can't be reached once the rail is set in pavement. A common European section, 59R2 (Ri59N), is 180 mm high with a 180 mm base and a 12 mm web, so its 0° base echo is close to 132RE's.

European grooved rails are made to EN 14811:2019, which covers grooved rails of 42 kg/m and over for tram systems. ArcelorMittal's grooved-rail brochure lists 59R2 (Ri59N) at 58.20 kg/m with H 180.00 mm, B 180.00 mm, E 12.00 mm and C 113.00 mm, plus C1 55.83 mm and C2 42.35 mm. Read with the brochure's drawing, C appears to be the overall head width including the lip, and C1 and C2 the running head and groove widths; confirm the labels on the drawing before using them in a procedure. A 2007 light-rail guideline notes that "there are no North American producers of girder rails" (Graebner, Jackson and Lovejoy, Trackway Infrastructure Guidelines for Light Rail Circulator Systems), so US systems buy these sections abroad.

That last row matters. EN 16729-1, the European standard for ultrasonic testing of rails in track, "applies only to rail profiles meeting the requirements of EN 13674-1", the Vignole (flat-bottom) rail standard. Grooved rail is outside it, so transit agencies and contractors write their own procedures and set sensitivity on a reference piece of the same grooved section.

Three practical effects come from embedment. The probe can only sit on the running surface. The groove fills with water, sand and de-icing residue that spoil the couplant film. And any length where coupling was lost has to be recorded as untested, not as sound rail. Our page on ultrasonic rail testing for metro and light rail systems covers possession time, third rail and transit regulation.

Key takeaway: grooved rail needs its own reference piece, its own probe positions and an honest record of what the head-only search could not reach.

What does head wear do to ultrasonic coverage?

Head wear changes the section under the probe: vertical wear shortens the rail and the head, and gauge-face (side) wear narrows the head and moves the gauge corner. Both shift the echo times and beam positions that a new-rail setup assumes, and the beam itself enters a worn head at a different angle.

TTCI's report on phased array rail inspection states the problem directly:

"As such, the direction of the ultrasonic wave within the material will be different on a worn rail than on a new rail. As a result, flaws may be missed during dynamic testing on worn rail because the wrong volume of material is inspected." (Witte and Poudel, TTCI Technology Digest TD-16-030, July 2016)

Wear is measured against a new rail of the same section. Harmer Steel's catalog describes "top wear" and "side wear" as the difference between the worn rail's height and head width and those of "a newly manufactured rail section of the same weight and size". Measurement methods and gauge face angle are covered in our guide to gauge face wear on rail.

Worked example: vertical wear on 136RE

Assume the head loses metal evenly from the top, so the rail height and the head depth each drop by the wear amount.

Check the 1/2 in. row: 2 × 6.8125 ÷ 0.2320 = 58.73 µs; 1.4375 × 2.9238 = 4.203 in.; 2 × 4.203 ÷ 0.1280 = 65.67 µs.

Three results stand out:

• A 136RE head worn 1/4 in. times exactly like new 133RE (60.88 µs), and one worn 1/2 in. times exactly like new 119RE (58.73 µs). The base-echo time alone can't tell a worn heavy section from a new lighter one.
• The remaining head of a 136RE worn 1/4 in. is as deep as a new 115RE head (1.6875 in.). A 70° gate left at the new-136RE length now runs about 11 µs past the bottom of the head.
• The 0° gate drifts by about 0.54 µs per 1/16 in. of wear. On a curve where wear varies along the rail, a narrow base-echo gate can produce intermittent loss of bottom from wear alone.

For scale, one rail supplier says its Number One relay grade "generally" allows a maximum of 1/4 in. of top wear on 136 lb and heavier rail, 3/16 in. for 119 to 133 lb and 1/8 in. up to 115 lb (Harmer Steel catalog, 2014). That is a commercial grading note, not a safety limit, but it shows that relay rail arrives with part of its head already gone.

Worked example: gauge-face wear on 136RE

Take a 136RE high rail with 1/2 in. of side wear.

So 1/2 in. of side wear moves every head channel 0.25 in. off the worn head's center, unless the carriage follows the worn gauge face. The gauge-corner channel loses the most. The corner it was aimed at has been worn away, and the new corner is lower and further toward the field side. A rail test patent (US 5,419,196, 1995) describes this case, noting that an earlier 40° side-looking probe misses the corner on a worn head.

Why the gauge corner is the hardest place to cover

The lower gauge corner is the part of the head ultrasonic equipment reaches least well, and it is where wear and fatigue meet. In its brief on the 2014 Lynchburg, Virginia derailment (NTSB/RAB-16/01), NTSB wrote that detecting fatigue in the lower corner of the gage face is "more difficult" because that part of the railhead is "not easily scanned by ultrasonic equipment".

Side wear makes this worse in three ways. It moves the corner away from where the probes were set, it steepens the gauge face so beams enter at a changed angle, and it often comes with shelling and head checks on high rails. Shells block the beam from what lies beneath them; see our page on rail shelling and spalling.

Key takeaway: wear turns one rail section into many; the setup has to follow the section that is actually in track, and the record has to show where coverage was lost.

Does any FRA rule set a rail head wear limit?

No. The FRA Track Safety Standards (49 CFR Part 213) set no limit on vertical head wear or gauge-face wear for running rail. A full-text search of Part 213 (govinfo annual CFR XML, Title 49, volume 4, 2025) finds no use of "head wear", "side wear" or "rail wear", and the word "ultrasonic" does not appear in the part at all.

Every use of "wear" or "worn" in Part 213 concerns something other than the running rail head:

Two rules are close enough to be confused with head wear. Section 213.113 defines "flattened rail" and "crushed head" as short local depressions of 3/8 in. or more over 8 in. or more, not at a joint. The definition of flattened rail separates it from a condition "such as head flow on the low rail of curves", so neither is a wear limit. And 213.53 limits track gage by class, measured 5/8 in. below the top of the rail; side wear can widen gage past that limit, but the rule limits the gage, not the wear.

Where wear does enter the FRA rules is through the test itself. Section 213.237 defines a valid search as "a continuous inspection for internal rail defects where the equipment performs as intended and equipment responses are interpreted by a qualified operator", and paragraph (g) says what happens when worn or damaged rail prevents one:

"If the person assigned to operate the rail defect detection equipment (i.e., the qualified operator) determines that a valid search for internal defects could not be made over a particular length of track, that particular length of track may not be considered as internally inspected under paragraphs (a) and (c) of this section." (49 CFR 213.237(g))

Under 213.237(h), the owner must then re-search the rail, cut speed to 25 mph until a valid search is made, or replace the rail, before the test interval runs out. Our guide to 49 CFR 213.237 covers the intervals.

Wear limits themselves come from railroad and transit maintenance standards, which typically draw on AREMA practice and research. Use the table in the owner's standard; no single number applies everywhere. NTSB tied wear to testing after the 2006 New Brighton, Pennsylvania derailment (NTSB/RAR-08/02), recommending in R-08-11 that FRA require railroads to "use methods that accurately measure railhead wear to ensure that deformation of the head does not affect the accuracy of the measurements".

Key takeaway: FRA limits the result of wear (gage, joint bars, frogs) and the validity of the search, not the amount of head wear.

How should a test setup account for section and wear?

Set up for the section in track, check it on reference rail of that section, and record where the rail differed from the setup. The steps below are a typical sequence to confirm with the test provider's Level III and written procedure.

1. List the sections on the territory. Use track charts, rail brands and relay records. Flag compromise joints, plug rails and relay sections.
2. Get wear data where it exists. Curve wear surveys and profile measurements tell you where heads are narrower or shorter than nominal.
3. Set gates by section. Set the 0° base-echo gate and the 70° head gates for each section, using the timing from the actual reference rail rather than a calculated number.
4. Check probe positions on worn rail. On high rails with heavy side wear, confirm that the head and gauge-corner channels still respond where they should.
5. Watch coupling and loss of bottom. Distinguish loss of bottom from a reflector, from wear-shifted timing and from lost coupling. Untested rail is untested rail, not a false call; see false calls in rail UT.
6. Hand-verify where coverage is doubtful. Gauge corners on worn high rails, transitions between sections and grooved rail at street crossings are the usual places.
7. Record the assumptions. The test record should state the section assumed for each length and list every length not validly searched, as 213.241(c)(6) requires for track not tested under 213.237(g).

The calibration steps themselves (velocity, zero, angle, reference-rail sensitivity, drift checks) are covered in our rail UT calibration guide, and reading the resulting A-scan in how to read a flaw detector.

What rail section and wear can't tell you

Knowing the section and the wear tells you where the beams should go; it doesn't tell you whether a defect is there or whether the search found it. The limits below apply to every setup in this article.

• Calculated times are steel-only. Wedges, delay lines and water paths add time, and the real velocity varies with the rail steel. Set gates from reference rail, not from the tables here.
• Head depth is an edge dimension. Mill data gives depth to the fishing surface at the head edge. The head is deeper over the web, so the true 70° path to the head underside varies across the width.
• Even wear is an assumption. Real heads wear unevenly, with flow, lipping and a sloped gauge face. Only a measured profile shows the actual shape.
• Mill and supplier dimensions are nominal. Rolling tolerances and AREMA's own drawings (MRE Section 1.1) govern; they are not reproduced here. The older-section dimensions come from one supplier's catalog.
• The base flanges are out of reach of a head-contact search. This applies to tee rail and grooved rail alike; field welds and embedded track need other arrangements.
• Surface damage masks the head. Shells and head checks block or scatter the beam whatever the section, and grinding is often the fix that restores the search.
• Channel layouts are system-specific. No public standard sets the North American channel set, so any probe table here is a typical arrangement.

How EMA Industries applies this

EMA Industries provides ultrasonic rail flaw testing for transit agencies, short lines and industrial track owners, based in Southern California and mobilizing nationwide. On mixed-section and worn track, the useful starting point is a list of the rail sections on the territory, any wear survey data, and the locations of compromise joints and relay rail, so the setup can follow the rail that is actually in place. Contact EMA Industries to scope a rail test or a review of an existing testing program.

Frequently asked questions

Why does the rail section matter for ultrasonic testing?

The section sets the timing and geometry the test relies on. Rail height sets when the 0° base echo returns (57.11 µs on 115RE, 63.04 µs on 136RE at 0.2320 in./µs), head depth sets how long a 70° beam takes to reach the bottom of the head, and head width sets how many beams it takes to cover the head. A setup made for one section can report false loss of bottom, or leave part of the head unsearched, on another.

What are the dimensions of 136RE rail?

Steel Dynamics lists 136RE at 7.3125 in. high, with a 6.0000 in. base, a 2.9375 in. head, a 1.9375 in. head depth and a 0.6875 in. web, 13.3263 in² of area and 135.8826 lb/yd nominal. ArcelorMittal's metric sheet for the same section agrees (185.70 mm high, 74.60 mm head, 49.20 mm head depth). These are nominal mill values; AREMA's own drawing is in MRE Chapter 4, Section 1.1.

How does a worn rail head affect a 70° probe?

Vertical wear makes the head shallower, so a 70° gate set for new rail runs past the bottom of the head: on 136RE worn 1/4 in., the head is 1.6875 in. deep and the gate end falls from 88.5 to 77.1 µs. Side wear moves the head center and the gauge corner, so probes set for the new head sit off-center. TTCI notes that worn rail also changes the beam direction, so "the wrong volume of material" can be inspected.

Is there an FRA limit on rail head wear?

Part 213 sets no limit on running-rail head or side wear. Its wear rules cover joint bars (213.121(b)), switch points (213.135(h)) and frogs (213.137, 213.141). Wear limits come from railroad and transit maintenance standards. FRA's rules reach worn rail through 213.237(g): where a valid search couldn't be made, that length doesn't count as inspected.

Can girder rail be tested with the same setup as tee rail?

A tee-rail setup can't simply be reused on girder rail. Grooved rail has a narrower running head beside a lip and groove, a different height (180 mm for 59R2) and a base sealed in pavement. EN 16729-1 covers only EN 13674-1 flat-bottom rails, so grooved-rail testing needs its own procedure and a reference piece of the same section.

Sources

• 49 CFR 213.237, 213.113, 213.121 and 213.53, read on Cornell LII (October 2026) and linked on eCFR: 213.237, 213.113, 213.121, 213.53; copies at law.cornell.edu/cfr/text/49/213.237
• Part 213 full text, 2025 annual CFR XML: govinfo.gov, CFR-2025 Title 49 vol. 4 Part 213
• Steel Dynamics, Rail Products brochure (08-2023): lpg.steeldynamics.com
• ArcelorMittal, 136RE (TR68) profile: rails.arcelormittal.com
• ArcelorMittal, Grooved Rails for Tramways brochure: rails.arcelormittal.com
• Harmer Steel Products, tee rail catalog (2014), introduction and section drawings: p. I-1, 90 ARA-A, 100 ARA-A, 100RE, 115RE, 133RE, 136RE
• AREMA, Manual for Railway Engineering Chapter 4 table of contents (2025): publications.arema.org
• Witte and Poudel, High-Speed Rail Flaw Detection Using Phased Array Ultrasonics, TTCI TD-16-030 (July 2016): mxvrail.com
• NTSB, Railroad Accident Brief RAB-16/01 (Lynchburg, Virginia): ntsb.gov
• NTSB, Railroad Accident Report RAR-08/02 (New Brighton, Pennsylvania) and Safety Recommendation R-08-11: ntsb.gov
• US Patent 5,419,196, rail side-looking probe (1995): Google Patents
• EN 16729-1:2016 scope: genorma.com; EN 14811:2019: BSI Knowledge
• Graebner, Jackson and Lovejoy, Trackway Infrastructure Guidelines for Light Rail Circulator Systems (April 2007), via nacto.org
• Evident, Material Sound Velocities: ims.evidentscientific.com