Rolling Contact Fatigue in Rail: Causes, Types, and Inspection Methods

Rolling contact fatigue (RCF) is surface cracking of the rail head caused by millions of repeated wheel contacts. It appears as head checks, shelling and squats. Under 49 CFR 213.113(d)(7), these conditions are a recognized origin of detail fractures, and surface RCF can block an ultrasonic search, which triggers the 213.237(h) options for invalid searches.

This guide is written for track supervisors, transit maintenance-of-way engineers and rail testing managers. It explains how rolling contact fatigue starts and grows, which forms matter, and how each one fits the defect terms of 213.113. It also covers why RCF defeats ultrasonic testing (UT), what eddy current can and can't tell you about crack depth, and how a depth reading turns into a grinding decision.

Contents: What it is · Causes · Types · 213.113 terms · UT masking · 213.237(h) · Eddy current · Methods compared · Grinding · Worked example · Limits · FAQ

What is rolling contact fatigue in rail?

Rolling contact fatigue in rail is damage that builds up in the thin surface layer of the rail head where the wheel touches it. No single wheel causes it. It comes from the sum of a very large number of load cycles, each of which pushes the surface metal a tiny amount in the direction of traction.

The FRA's comprehensive review of the subject, Rolling Contact Fatigue: A Comprehensive Review (DOT/FRA/ORD-11/24, Magel, November 2011), puts the scale in numbers. A typical rail in freight or commuter service sees about 3 to 5 million wheel load cycles for every 100 million gross tons (MGT) of traffic. Under high friction and high contact stress, part of those cycles deform the surface metal in small increments, a process called ratcheting.

The surface steel work-hardens as it ratchets, but it can't strain forever. When its ductility is used up, a crack forms at the surface. Early cracks grow at a shallow angle to the running surface, typically in the direction of traffic.

The review describes crack growth in three phases, drawing on a crack-growth model from the University of Sheffield:

• Phase I: cracks start early in rail life through ratcheting.
• Phase II: the cracks grow faster until they pass through the depth of maximum shear stress, then slow down once they leave the zone controlled by contact stress (unless water or lubricant gets into them).
• Phase III: bending and residual stresses take over and decide whether the crack turns up to the surface, which spalls metal out, or turns down into the head, which can break the rail.

The practical point for a track owner is that phase III is where a surface problem becomes a structural one. A crack that turns down into the head is no longer a maintenance nuisance. It is a progressive fracture that 49 CFR 213.113 treats as a defect.

Key takeaway: rolling contact fatigue starts at the surface, and the danger comes when a surface crack turns down into the rail head.

What causes rolling contact fatigue?

Rolling contact fatigue is driven by contact stress, traction, track geometry, moisture and the rail steel itself. Each factor changes either how hard the wheel loads the surface or how well the steel resists it.

Contact stress and wheel load

Contact stress depends on wheel load, wheel diameter, and the transverse profiles of the wheel and rail. ORD-11/24 notes that under Hertz's elastic contact equations, doubling the wheel load raises the peak contact stress by only about 27%, and tripling it raises the stress by about 44%.

Small stress increases still matter because fatigue life is very sensitive to them. The same review cites a laboratory result in which a 10% rise in contact stress (from 3,320 to 3,640 MPa) cut the fatigue life of rail steels by a factor of about 2 under pure rolling.

Loads have grown. ORD-11/24 reports that between 1965 and 1995 the average freight car axle load more than doubled, and annual tonnage on main-line track quadrupled. Dynamic loads from wheel flats, dipped welds, poor tie support and corrugation add to the static load and can be managed.

Traction and creepage

Traction often matters more than load. ORD-11/24 points out that doubling the traction coefficient from 25% to 50% raises the stress by about 50%, far more than doubling the wheel load does.

Friction at the wheel-rail contact isn't constant. Left unmanaged, the coefficient can range from near zero to about 0.7, depending on moisture, leaves, contaminants and lubricant. High creepage in curves, at the gauge corner of the high rail and the field side of the low rail, is why those locations crack first.

Track geometry

Local geometry faults concentrate wheel passes and raise dynamic loads. The FRA review gives freight examples: a 6-foot length of tight gauge produced a local band of gauge corner RCF, and a crosslevel error from a poor ballast shoulder produced a deep-seated shell on tangent rail.

Geometry faults tend to produce clusters of RCF rather than continuous cracking. A cluster on otherwise sound rail is a reason to look at the geometry before planning grinding.

Moisture and lubricants

Water changes how cracks grow. Modeling cited in ORD-11/24 found that water drawn into a surface crack and pressurized by the passing wheel raised the Mode I (opening) stress intensity roughly fivefold compared with a dry crack.

The review adds a caution. Some fluid-entrapment models predict growth rates about 35,000 times the shear-mode rate, which doesn't match laboratory measurements, so the size of the effect is uncertain. Grease works differently: it lubricates the crack faces and promotes shear-mode growth instead.

Rail steel and cleanliness

Harder rail resists the surface plastic flow that starts RCF. ORD-11/24 reports that, on the basis of as-manufactured hardness, a 400 HB hypereutectoid premium rail can withstand contact stresses about 45% greater than standard 275 HB rail without yielding. That is a yield comparison, not a promise of 45% longer fatigue life.

Cleanliness matters for the deeper defects. TTCI research on shelling at the FAST test loop (TD-10-004, 2010) found manganese sulfide, aluminum oxide and complex oxide inclusions of 50 to 100 µm in the fatigue marks, and recommended cleaner steel.

Source: compiled from FRA DOT/FRA/ORD-11/24 (2011) and TTCI TD-10-004 (2010).

Key takeaway: traction and local geometry decide where RCF appears; load and steel decide how fast it becomes serious.

What types of rolling contact fatigue defects form on rail?

The main rail forms are head checks (gauge corner cracking), shelling, squats, spalling and flaking, and crushed heads. Some of these become detail fractures, which is the form that breaks rail.

Sources: 49 CFR 213.113(d); FRA DOT/FRA/ORD-11/24; TTCI TD-10-004.

Head checks and gauge corner cracking

Head checks are thin, closely spaced cracks at the gauge corner, most often on the high rail of curves and sometimes on tangent track. ORD-11/24 notes they are often regularly spaced along whole curves; where they appear in clusters, a track geometry fault is usually nearby. North American practice tends to call them gauge corner cracking, and European practice head checks.

Some track inspection manuals describe head checks as running at about 45° across the gauge corner. As they grow, they get longer and deeper and can lead to pits and shells. For the full defect profile, see our guide to head checks in rail.

Head checks sit on the same corner where flanges wear the rail. Wear removes cracked metal, so a rail that wears quickly can carry fewer cracks than a premium rail that barely wears. The interaction is covered in our article on gauge face wear in rail.

Shelling and gauge corner collapse

Shelling is the loss of chunks of metal from the gauge corner. ORD-11/24 describes two routes: surface and subsurface cracks linking until the metal above them is no longer attached, or collapse of the gauge corner when overloading makes the steel yield in shear along a slip line. The second route produces what the review calls the deep-seated shell.

TTCI's FAST study (TD-10-004) found shells starting 1/4 to 3/8 in. below the high-rail gauge corner in 5° and 6° curves under 39-ton axle loads. They took 20 to 40 MGT to grow from initiation to critical failure. Those were accelerated test conditions, so revenue-service lines will differ.

Shells are not only a heavy-haul problem. The FRA review shows gauge corner fatigue on a light rail system where flattened rail overloaded the gauge corner. More on how shells and spalls differ is in rail shelling, spalling and surface defects.

Squats

A squat is a surface-initiated RCF defect, not a subsurface void. ORD-11/24 calls the squat "a surface defect most commonly associated with high-speed rail and areas of high tractive effort." It shows as a shallow, dark depression near the center of the rail head on tangent and mildly curved track.

In cross section a squat has two cracks: a short leading crack (in the direction of travel) and a much longer trailing crack, each running at about 30° to the horizontal. The depression comes from that cracking, the weakened metal above it and debris collecting in the dip. Nothing hollow forms and collapses.

The review gives two proposed mechanisms, and both start at the surface. One is a thin, brittle martensite layer formed by high traction and microslip, which cracks readily (the explanation used in Japan). The other is classic RCF from high surface shear.

The trailing crack is the dangerous part. Small branch cracks grow from it, and one may turn down into the head and start a transverse fracture. ORD-11/24 notes that those branch cracks, shielded by the trailing crack, are often missed by conventional ultrasonic systems.

Spalling, flaking and crushed heads

Flaking is very shallow surface cracking. Some manuals put it at about 1/32 in. deep, shallower than shelling, and it can develop into shelling. Some manuals also call very shallow shells spalls; the terms aren't used the same way everywhere, so check the owner's defect manual.

A crushed head is the end state of RCF that has worked toward the center of the head. ORD-11/24 says crushed heads are typically about 0.25 to 0.5 m long, often triggered by a local stress raiser such as poor grinding, no grinding, track irregularities or dirty steel.

Part 213 sets its own size threshold. Under 49 CFR 213.113(d)(4), a crushed head is a short length of rail, not at a joint, that has sagged across the head width at least 3/8 in. below the rest of the head and is at least 8 in. long, with the sag also visible in the head-web fillet.

Detail fractures: where RCF becomes a structural defect

A detail fracture is the defect that turns rolling contact fatigue into a broken rail. Part 213 defines it by origin, which ties it directly to RCF:

"Detail fracture means a progressive fracture originating at or near the surface of the rail head. These fractures should not be confused with transverse fissures, compound fissures, or other defects which have internal origins. Detail fractures may arise from shelled spots, head checks, or flaking." (49 CFR 213.113(d)(7))

The distinction matters for diagnosis. A transverse fissure under 213.113(d)(15) starts from a crystalline center or nucleus inside the head, not from surface damage. A detail fracture grows from a head check, shell or flake. Inclusions matter because they help start the shell, as ORD-11/24 notes for oxide stringers, not because the transverse crack itself starts at an inclusion.

Detail fractures are a leading cause of broken-rail derailments. In FRA accident data for main track, 2016 to 2025, cause code T207 (detail fracture from shelling or head check) accounted for 110 of the 384 broken-rail derailments, the largest single rail code. The 1992 Superior, Wisconsin, hazardous materials derailment was traced by the NTSB to detail fractures from shelling at the upper gauge corner.

For how these defects grow and are sized, see detail fracture development in rail and, for the internal-origin defect it is often confused with, transverse fissure in rail.

Wheels suffer a parallel set of RCF defects (tread cracking, shelling and shattered rims), but those belong to rolling stock maintenance, not to a track testing program.

Key takeaway: head checks, shells and squats all start at or near the surface, and any of them can seed the detail fracture that breaks the rail.

How does rolling contact fatigue fit the defect terms of 49 CFR 213.113?

Most forms of rolling contact fatigue are not 213.113 defects in themselves. The remedial action table has no row for head checks, shelled rail, spalling, flaking or squats. Those conditions enter the rule only through the detail fracture definition in 213.113(d)(7) and through the crushed head row.

That is easy to get wrong, because "shelled rail" sounds like a defect type. A rail with shells but no detail fracture carries no 213.113 remedial action. What it does carry is a testing problem, because the shells may stop UT from seeing a detail fracture below them.

Source: 49 CFR 213.113(c) table and (d) definitions, as revised by the 2014 Improving Rail Integrity final rule (79 FR 4234).

In summary, the lettered actions mean the following. A is visual supervision of each operation over the rail by a person designated under 213.7. A2 is a visual inspection, then no more than 10 mph for up to 24 hours. C and D require joint bars within 10 days and 7 days, with speed limits until they're applied. E is joint bars bolted under 213.121, and H limits speed to 50 mph or the class maximum, whichever is lower.

The 2014 rule lowered the threshold for A2 or E and H from 80% to 60% of head area. FRA cited detail fracture growth of more than 5% of head area per MGT as part of the reason, which shows how quickly an RCF-born defect can move up the table.

Two other parts of 213.113 apply. Under 213.113(a), a person designated under 213.7 decides whether the track may stay in use. Under 213.113(b), an indication found during an internal inspection must be verified within 4 hours, or immediately for defects that require action A, A2 or B.

Key takeaway: cite 213.113(d)(7) for shelling and head checks, and size any detail fracture they produce against the table.

Why does rolling contact fatigue mask ultrasonic testing?

Surface cracks reflect and scatter ultrasound before it reaches the metal below, so RCF can hide the detail fracture it creates. The FRA review states the problem plainly:

"Surface fatigue is a serious hazard because it prevents the successful transmission of ultrasonic waves to the base of the rail and back." (FRA DOT/FRA/ORD-11/24, section 1.2.1)

The result is what the industry calls a no-test. A test car's 0° channel loses its back-wall echo (often called loss of bottom), and its angle channels can't put enough energy past the cracked layer. The inspector usually has to stop and hand test, and on many railroads a failed test means the affected rail comes out.

UT has a second blind spot at the surface. ORD-11/24 notes that noise and scatter from the first few millimeters are so large that analysis systems generally ignore that zone to focus on internal flaws. Conventional probes can still pick up larger (over 4 mm, about 0.16 in.) surface-originated defects such as deep head checks.

TTCI field data puts numbers on the masking. In a 1999 benchmark of six hi-rail test systems (TD-99-006), about 62% of transverse defects under shells were found, against about 77% of those not under shells. In TD-18-027 (2018), drilled reference holes in five rails pulled for low-confidence UT returned on average 67% of their expected response, and microscopy showed 90% of 880 RCF cracks and spalls were no deeper than 0.040 in.

Masking is mainly a detail fracture problem. Shells and head checks hide what grows from them, and detail fractures grow from them. A transverse fissure starts inside the head, so it is less likely to sit under the surface damage. Our article on surface-wave UT for rail explains why surface-wave methods are still a research tool here, not a substitute for a valid search.

Key takeaway: the shallow cracks UT can't size are the ones that block it from finding the deep crack that matters.

What does 49 CFR 213.237(h) require when RCF blocks a valid search?

If a valid search could not be made over a length of track, the owner must re-search it, slow trains to 25 mph, or replace the rail, before the inspection deadline expires. Masking by RCF is one of the most common reasons a search is invalid.

The chain of rules runs as follows. Under 213.238(b)(4), the qualified operator must determine that each valid search "has not been compromised due to environmental contamination, rail conditions, or equipment malfunction." Heavy head checking and shelling are rail conditions.

Under 213.237(g), if the qualified operator determines a valid search could not be made over a length of track, that length "may not be considered as internally inspected." Under 213.237(h), before the time or tonnage limit expires, the owner must:

1. conduct a valid search for internal defects;
2. reduce operating speed to a maximum of 25 mph until a valid search can be made; or
3. replace the rail that had not been inspected.

The record matters too. 49 CFR 213.241(c)(6) requires the internal inspection record to show the location of any track not tested under 213.237(g). An untested stretch is not a false call and not a clean result; it is missing coverage that has to be closed.

Grinding is the usual way to make a re-search possible, but it is a means, not a remedy. Part 213 doesn't mention grinding, and a grinding pass earns no 213.237 credit. The owner still needs a valid search after the rail is ground.

A practical sequence for a masked stretch:

1. Record the start and end of the no-test zone by milepost and rail.
2. Hand test the zone with probes and angles suited to the head condition, from the field side as well if the gauge corner is cracked.
3. If hand testing still can't produce a valid search, plan corrective grinding and a re-test before the deadline.
4. If the deadline will pass first, apply the 25 mph limit or replace the rail.
5. Close the record when the valid search is complete.

The interval that sets the deadline comes from 213.237(c). On Class 4 and 5 track, and Class 3 track with scheduled passenger trains or on a hazardous materials route, it is 370 days or 30 MGT, whichever comes first. See 49 CFR 213.237 rail flaw detection requirements and how often rail must be tested.

Key takeaway: a no-test caused by RCF starts a 213.237(h) clock, and grinding only helps if a valid search follows it.

What can eddy current tell you about RCF crack depth?

Eddy current testing (ECT) estimates the depth of shallow, surface-breaking RCF cracks so that grinding can be planned and checked. It is not a deep inspection method, it doesn't find internal defects, and it doesn't give a 213.237 valid search.

ECT induces currents in a thin skin of the rail surface and senses how cracks disturb them. The FHWA's guide to eddy current testing puts the skin depth in steel at typically less than 1 to 3 mm, and states that only surface-breaking cracks are detectable in that application. Claims that eddy current reaches about 10 mm into rail steel find no support in the FRA or FHWA material.

ECT is very sensitive to lift-off, the gap between the probe and the rail. ORD-11/24 notes rail systems need a consistent standoff of about 1 to 2 mm, which is hard to hold at speed on rough rail with a changing head shape. That makes ECT more sensitive to surface condition than UT, not less.

ECT doesn't measure depth directly. A TTCI evaluation of a Rohmann eddy current trolley on Norfolk Southern track (TD-14-007, 2014) explained why:

"The Draisine relies on the user to enter an assumed crack angle, as the actual angle cannot be determined through nondestructive testing." (TTCI TD-14-007, 2014)

With the default angle of 25°, the trolley's depth readings differed from sectioned samples by 5% to 33%. Crack angles in the high-rail sections ranged from 18° to 28°. TTCI's companion report (TD-14-008) concluded that, for both eddy current and UT, accurate measurement of crack angle and depth is still a primary limitation.

The angle assumption matters because depth is computed from the crack's extent along its face. The table below shows the same 0.33 in. crack face read at three angles.

Worked from the 18° to 28° angle range reported in TTCI TD-14-007; the 0.33 in. crack face is illustrative.

ECT is used where it is strong. On the same project, readings on a 100 ft segment of low rail in a 6° curve at Wabun, Virginia, dropped from 0.14 in. before grinding to less than 0.04 in. after five passes. ORD-11/24 also describes eddy current scanners carried on rail grinders that record the position and depth of head-check cracks before and after grinding, while noting it is "difficult with eddy current to know the actual depth of damage." For a broader comparison, see eddy current testing vs ultrasonic testing.

Key takeaway: treat an eddy current depth as an estimate with an error band, good for planning metal removal and confirming it, not for sizing internal defects.

How do inspection methods compare for RCF?

No single method covers RCF: UT finds internal defects, and surface methods measure the cracking that masks UT. The table summarizes what each one can and can't do.

Sources: FRA DOT/FRA/ORD-11/24 sections 3.1 to 3.3; TTCI TD-14-008. The ACFM response note is from ORD-11/24; the vision and acceleration notes summarize its review of NRCC and European studies.

A few points from the FRA review are worth keeping. ACFM induces a uniform current in a skin less than 0.1 mm deep, and "the deeper the crack, the larger the response." Axle-box accelerometers on a test car found short vertical irregularities of about 0.1 to 1 mm. A 2004 National Research Council of Canada study found RCF cracks an order of magnitude harder for vision systems to distinguish than spalls.

Key takeaway: pair UT with one surface method that can put a number on the cracking, and use that number to plan grinding.

How does rail grinding control rolling contact fatigue?

Grinding removes the cracked surface layer before cracks grow deep, and it restores the rail's ability to be tested ultrasonically. ORD-11/24 lists improving UT among grinding's purposes because "signals cannot pass through cracks."

Preventive grinding removes a thin layer often, before cracks deepen. ORD-11/24 describes Canadian Pacific's experience: a target wear rate (natural plus ground) of about 0.001 in. per MGT in sharp curves, on a 25 MGT cycle, with about 0.2 to 0.3 mm removed by the grinder each cycle. When one sharp-curve subdivision missed a cycle and went to 37 MGT, the premium rail developed gauge corner spalling.

Corrective grinding takes heavy, multiple-pass cuts after the damage is severe. The review warns it removes the work-hardened layer and can leave the deepest cracks behind. In Japan, preventive grinding that removes about 0.08 mm every 40 MGT is reported to have effectively eliminated squats.

For UT, the numbers come from TTCI. In TD20-002, corrective grinding restored a valid ultrasonic test at 68 of 88 locations (77%) that had failed, and at every location where at least 0.060 in. of metal was removed. A production grinder typically removes about 0.006 to 0.008 in. per pass, so 0.060 in. is roughly 8 to 10 passes.

Investigators have reached the same conclusion. In its report on a 1999 broken-rail derailment at Bedford, Nova Scotia (R99M0046), the Transportation Safety Board of Canada wrote that "The recommended remedial action for shelly damage is rail grinding." The last ultrasonic test, about 3.5 months before the derailment, had found nothing.

Grinding works alongside other controls. Friction management (gauge-face lubrication and top-of-rail friction modifiers), better rail and wheel profiles, track geometry correction and harder, cleaner rail all reduce the rate at which new cracks form. Our page on rail grinding and ultrasonic testing covers how to schedule testing around a grinding program.

In Europe, grinding acceptance is set by EN 13231-2:2020; the EN 13231-3 standard cited in the 2011 FRA review was withdrawn in 2020. In North America, AREMA Committee 4 publishes grinding practice in the Manual for Railway Engineering, which is paywalled, so check the edition in your specification.

Key takeaway: grind before cracks pass the shallow zone a grinder can remove, then re-test to get the valid search.

Worked example: from crack depth to a grinding decision

The example below turns eddy current depth readings into a grinding plan for three segments of one curve. The method is illustrative; your owner's procedure sets the real margins.

Inputs:

• An eddy current trolley reports the maximum gauge corner crack depth on the high rail of a curve, using the 25° default angle.
• Measurement uncertainty: up to 33% (the top of the TD-14-007 range), so upper-bound depth = reading × 1.33.
• Grinding target: upper-bound depth plus a 0.010 in. margin (an illustrative owner allowance, not a standard).
• Production grinder removal: 0.006 to 0.008 in. per pass (TTCI TD20-002).
• Passes = target ÷ removal per pass, rounded up. Fewer passes at 0.008 in.; more at 0.006 in.

Checks. Segment A: 0.030 × 1.33 = 0.0399, rounded to 0.040; + 0.010 = 0.050; 0.050 ÷ 0.008 = 6.25 → 7; 0.050 ÷ 0.006 = 8.33 → 9. Segment B: 0.060 × 1.33 = 0.0798 → 0.080; 0.090 ÷ 0.008 = 11.25 → 12; 0.090 ÷ 0.006 = 15. Segment C: 0.140 × 1.33 = 0.1862 → 0.186; 0.196 ÷ 0.008 = 24.5 → 25; 0.196 ÷ 0.006 = 32.7 → 33.

Decisions:

• Segment A needs 7 to 9 passes, close to the 0.060 in. removal at which TD20-002 saw every failed UT location restored. One corrective visit is reasonable, followed by an eddy current check and a UT re-test for the valid search.
• Segment B needs 12 to 15 passes, more than a grinder normally gives one location in one visit. Plan two visits, hand test the zone in between, and track the 213.237 deadline so that 213.237(h) is applied if the re-test can't be done in time.
• Segment C would need 25 to 33 passes and 0.196 in. of metal. Against an allowable head wear of 0.75 in. (the Canadian Pacific example in ORD-11/24), that single cut would use 0.196 ÷ 0.75 = 26% of the rail's wear life. The upper-bound depth of 4.72 mm is also close to the roughly 5 mm depth at which one Chalmers study found RCF cracks tend to turn transverse. Grinding alone is the wrong answer here: hand test for detail fractures now, and plan rail replacement.

On Segment C, the eddy current reading alone isn't evidence of a detail fracture. It is a reason to look for one with UT before trains continue at track speed, under the owner's 213.113 process.

Key takeaway: the crack depth and its error band decide whether you grind once, grind in stages or replace the rail.

What are the limits of RCF inspection?

Every method in this article has blind spots, and the plan should say which ones apply to each segment.

• UT can't see through heavy surface cracking. A no-test is missing coverage, not a pass, and the 213.237(h) options apply.
• UT ignores the first few millimeters. It won't size shallow head checks, so it can't tell you how much metal to grind.
• Squat branch cracks hide behind the trailing crack. Conventional systems often miss them.
• Eddy current depth depends on an assumed crack angle. TD-14-007 found errors of 5% to 33%, and lift-off from wear, debris or rough rail adds more.
• Eddy current and other surface methods find nothing internal. They can't detect a transverse fissure or a detail fracture that has grown below the skin depth.
• Magnetic particle and dye penetrant show only that a crack exists. They give no depth below the surface.
• Grinding leaves the deepest cracks. Corrective grinding can also remove the work-hardened layer and leave softer steel that cracks again.
• Research figures are site-specific. The FAST shelling rates and Japanese grinding results come from particular steels, loads and test conditions.

How EMA Industries applies this

EMA Industries performs ultrasonic rail flaw testing for transit agencies, short lines and industrial track owners, based in Southern California and mobilizing nationwide. On RCF-affected curves, the test report should mark every stretch where a valid search couldn't be made, by milepost and rail, so the owner can close it under 213.237(h) and record it under 213.241(c)(6). To plan testing around a grinding program, see our ultrasonic rail testing services or request a quote.

Related reading: Head checks in rail · Rail shelling and spalling · Detail fracture development · Transverse fissure in rail · Surface-wave UT for rail · Gauge face wear in rail · Rail grinding and UT

Frequently asked questions

What is rolling contact fatigue in rail?

It is cracking of the rail head's surface layer caused by repeated wheel loads and slip. It shows as head checks, shelling, squats and flaking, and it can grow into a detail fracture, a progressive transverse crack that 49 CFR 213.113 treats as a defect.

Is shelled rail a defect under 49 CFR 213.113?

Not on its own. The remedial action table has no row for shelling or head checks. Shelling appears in the rule only in the detail fracture definition, 213.113(d)(7), so a remedial action applies when a detail fracture is found and sized.

Why does a test car report a no-test on curves with head checks?

The surface cracks scatter and reflect the ultrasound before it reaches the head and base, so the 0° channel loses its back-wall echo and the angle channels lose sensitivity. The operator then can't call the search valid, and 213.237(g) and (h) apply.

How deep can eddy current measure RCF cracks?

Eddy current reaches only the skin depth, typically under 1 to 3 mm in steel, and senses surface-breaking cracks. Its depth readings depend on an assumed crack angle; one TTCI trial found errors of 5% to 33%. Use it to plan and check grinding, not to find internal defects.

How much metal should grinding remove to restore an ultrasonic test?

It depends on the crack depth. In TTCI's TD20-002 work, removing at least 0.060 in. restored a valid UT test at every failed location studied, roughly 8 to 10 passes at 0.006 to 0.008 in. per pass. Measure the crack depth first and re-test after grinding.

Sources

• 49 CFR 213.113, Defective rails, eCFR (text read on Cornell LII, October 2026)
• 49 CFR 213.237, Inspection of rail, eCFR
• 49 CFR 213.238, Qualified operator, eCFR
• 49 CFR 213.241, Inspection records, eCFR
• Track Safety Standards; Improving Rail Integrity, final rule, 79 FR 4234 (January 24, 2014), Federal Register via govinfo
• Magel, E., Rolling Contact Fatigue: A Comprehensive Review, DOT/FRA/ORD-11/24 (November 2011), FRA, via ROSA P
• TTCI TD-14-007, eddy current measurement of RCF crack depth (May 2014), MxV Rail (formerly TTCI)
• TTCI TD-14-008, RCF measurement technologies (May 2014), MxV Rail
• TTCI TD-18-027, RCF and ultrasonic test confidence (November 2018), MxV Rail
• TTCI TD-20-002, corrective grinding and ultrasonic test exceptions, MxV Rail
• TTCI TD-10-004, rail shelling at FAST (March 2010), MxV Rail
• TTCI TD-99-006, benchmarking North American rail flaw detection technologies (February 1999), MxV Rail
• FHWA, Eddy Current Testing, Federal Highway Administration NDE Web Manual
• TSB Canada, Railway Investigation Report R99M0046 (Bedford, Nova Scotia, October 9, 1999)
• NTSB safety recommendation letter R-94-1 and -2 (Superior, Wisconsin, June 30, 1992; report NTSB/HZM-94/01)
• FRA Rail Equipment Accident/Incident data (Form 6180.54), data.transportation.gov, main-track derailments 2016–2025
• Ekberg, A. and Kabo, E., Chalmers University of Technology (2014), abstract only (depth at which RCF cracks turn transverse)