Railroad Switch and Crossing Inspection: Critical UT Defects

Railroad switch and crossing inspection is the check of every switch, turnout and rail-to-rail track crossing that 49 CFR 213.235 requires on foot at least monthly, plus internal ultrasonic testing of the rail through them. When testing finds a defect, 49 CFR 213.113 sets the remedial action by defect type and size.

A note on the word "crossing". In this article a crossing is a place where rails cross rails: a diamond where two tracks intersect, or the frog of a turnout (in UK usage the frog itself is often called the "crossing"). It does not mean a highway-rail grade crossing. For road crossings and the web corrosion hidden under crossing panels, see grade crossing rail inspection.

Switches and crossings are harder to test than plain rail for two reasons. The geometry changes every few inches, and most mainline frogs and diamonds are cast austenitic manganese steel, whose coarse grain scatters ultrasound. This guide covers the inspection rules, the defects that matter, why frog UT is difficult, and what Part 213 says to do once a defect is confirmed.

What does "crossing" mean in switch and crossing inspection?

In track work, "crossing" means rails crossing rails, not a road crossing. The term is used three ways, and mixing them up sends readers to the wrong rule.

Section 213.235 lists "track crossing" beside switches and turnouts, and in that context it means a rail-to-rail crossing. Most research on cast manganese "crossings", including the UK RailSAFT project, is about frogs and diamonds.

Key takeaway: if a report or RFP says "crossing inspection", ask whether it means a diamond, a frog or a road crossing before you scope the work.

How often must switches, turnouts and crossings be inspected?

Under 49 CFR 213.235(a), every switch, turnout and track crossing on excepted and Class 1 to 5 track must be inspected on foot at least monthly. The rule reads:

"Except as provided in paragraph (c) of this section, each switch, turnout, track crossing, and moveable bridge lift rail assembly or other transition device shall be inspected on foot at least monthly." — 49 CFR 213.235(a), eCFR

Several other rules apply to the same locations. They cover different things, so a turnout is usually inspected under more than one of them.

Two points are often missed. First, 213.235 is a visual rule. A monthly walk can't see a transverse defect inside a switch point. Second, 213.237 sets the internal testing interval for rail in general and has no turnout-specific provision. The text of 213.237 never mentions frogs or switch points, so it gives no guidance on how a test car should handle a frog casting. Our 49 CFR 213.237 guide explains the intervals and what counts as a valid search.

Records of 213.235 inspections, like other visual track inspection records, are prepared on the day of inspection, signed and kept for at least one year (213.241). Class 6 and higher track follows Subpart G instead, including 213.353 for turnouts and crossovers.

Key takeaway: monthly on foot is the visual floor; internal testing of the rail through a turnout comes from 213.237, not 213.235.

What do 49 CFR 213.133 to 213.143 require for turnouts and frogs?

Sections 213.133 to 213.143 set the physical condition limits for turnouts, switches and frogs on Class 1 to 5 track. These are the numbers an inspector measures during the monthly walk. They were read on the Cornell LII copy of Part 213 and the 2024 CFR text on govinfo.

Paragraph 213.143(b) lets a heavy-point frog on Class 5 track run a guard check gage down to 4 ft 6-3/8 in. if it has at least three serviceable through-gage plates with elastic fasteners and adjustable braces, and carries a mark identifying it as a heavy-point frog. Paragraph 213.137(e) exempts flange-bearing frogs in crossing diamonds on Class 2 to 5 track from the flangeway depth rule when the crossing angle is greater than 20 degrees, unless movable guard rails are used.

Note that these are wear and geometry limits, not internal-flaw limits. Part 213 has no ultrasonic acceptance criterion for a frog casting.

Worked example: does a chipped frog point trigger the 10 mph limit?

An inspector measures a frog point on Class 3 track. The rule in 213.137(b) needs both conditions: more than 5/8 in. down and more than 6 in. back.

A separate check applies to the tread: if the tread of the casting is worn 7/16 in. below the original contour, that exceeds 3/8 in., and 213.137(c) limits speed over the frog to 10 mph regardless of the point.

Which UT defects matter most in switches and crossings?

The defects that matter most are the same internal rail defects listed in 213.113 (transverse and detail fractures, split heads, head-web separations, bolt hole cracks and defective welds), plus fatigue cracking in cast frogs. Where each one appears depends on the component.

FRA's 2022 special trackwork study notes that wear and damage on diamonds and frogs "are mainly associated with the nose or point, and joints at the end of the castings." The rail defect definitions themselves come from 213.113(d): for example, a detail fracture is "a progressive fracture originating at or near the surface of the rail head", which is why heavy wheel contact at a stock rail or frog wing matters.

The probe angles in the table are typical industry practice, not a Part 213 requirement. Equipment and procedures vary by test provider.

Why is ultrasonic testing of cast manganese frogs so difficult?

Cast austenitic manganese steel (AMS) has large, uneven grains that scatter and absorb ultrasound, so the conventional high-frequency probes used on rail lose the signal in noise. That is the main reason frogs and diamonds are rarely tested ultrasonically in service.

Manganese steel is used for frogs because it work-hardens under traffic. The same casting structure that makes it tough makes it hard to inspect. FRA research report RR 22-25, Feasibility of Special Trackwork Inspection Using Ultrasonics (September 2022, R. Wilson of FRA and A. Poudel of TTCI) found:

• "Most of the mainline frogs and crossing diamonds are made with austenitic manganese steel (AMS) castings."
• Conventional high-frequency UT of AMS castings is difficult because the casting "is highly attenuative in nature".
• The microstructure is "coarse-grained, nonhomogeneous, and anisotropic", like cast stainless steel. The mean grain diameter in the diamond sample was 0.037 in. (0.94 mm), measured to ASTM E112, which the authors call "fairly large" compared with rail steel.
• Manufacturers inspect crossing diamonds with radiography before shipment, which is costly and needs special facilities.

The report's most important line for track owners is this:

"Revenue service examinations of these special trackwork components currently use only visual inspections." — FRA Research Results RR 22-25, September 2022

The research team tested a wedge cut from a new AMS diamond casting with drilled holes at different depths. At 500 kHz the holes in the wider sections gave clear echoes. At 2.25 MHz, the echo from a hole 0.75 in. (19 mm) deep was about 26 percent and still detectable. The authors concluded that low-frequency UT "could be helpful" for characterizing internal flaws, and called for more samples and a standard special trackwork test sample. It was a laboratory feasibility study, not a field detection rate.

In the UK, TWI's RailSAFT project reached the same diagnosis. TWI says the "fully austenitic coarse grained structure" makes subsurface examination ineffective with standard ultrasonic techniques, and that in-service crossings were limited to visual and dye-penetrant inspection. RailSAFT combined low-frequency transducers with synthetic aperture focusing (SAFT) image processing, an encoded scanner and a conformable water-filled membrane to couple to the crossing surface. TWI reports the system could "reliably detect and locate real sub-surface and surface breaking defects" in critical regions of manganese crossings. It is a research system, not a routine test-car method.

Worked example: why lower frequency helps in a coarse-grained frog

Grain scattering becomes severe when the grains are no longer small compared with the ultrasonic wavelength. Wavelength is λ = v / f. The longitudinal velocity of cast AMS wasn't verified for this article, so this example uses 5,740 m/s, Evident's figure for 302 stainless steel, as a proxy (RR 22-25 compares AMS to cast stainless). The grain size is the 0.94 mm measured in RR 22-25.

At 5 MHz the wavelength is barely larger than one grain, so much of the energy is scattered back as noise. At 500 kHz the wavelength is about 12 grains long and the casting looks more uniform to the beam. In the low-frequency (Rayleigh) range, scattering attenuation is commonly modeled as proportional to grain size cubed times frequency to the fourth power, so halving the frequency cuts scattering by a factor of about 16 (2⁴). That is why RR 22-25 and RailSAFT both used low frequencies.

The trade-off is resolution. An 11.5 mm wavelength can't resolve small or closely spaced flaws the way a 2.25 MHz probe can in rail steel. Shear waves, which most rail head probes use, have a shorter wavelength than longitudinal waves at the same frequency and scatter more. Treat these numbers as an illustration of the physics, not as a procedure.

Key takeaway: a standard rail test car can't be relied on to find internal flaws in a manganese frog; plan visual and surface inspection of the casting, and treat UT of the frog as specialist work.

How does turnout geometry affect UT coverage?

Turnout geometry limits where a probe can sit and adds echoes that aren't defects. Plain rail has one constant profile; a turnout changes profile along its length.

• Switch points taper from a full rail head to a thin tip, so a probe designed for a full head loses contact or sees the edge of the point instead of a flaw.
• Stock rails and points run side by side, which limits access to the gage side of the stock rail.
• Frogs and wing rails put several running surfaces and flangeways close together. Their edges, bolt holes and casting features all return echoes.
• Guard rails, braces, plates and rods block access to the web and base.
• Surface condition matters more than on plain rail. Batter at the frog point, metal flow on switch points and grinding marks all disturb coupling.

Because of this, test providers usually combine the vehicle or cart search on the straight-through rails with hand UT of selected areas, such as switch heel joints, welds and the full-section part of the points. The test report should state which turnout areas were searched and which were not, because a stretch without a valid search is not considered tested under 213.237(g). Phased array UT can steer the beam in hand testing, but it doesn't solve the coarse-grain problem in a manganese casting on its own.

The basic pulse-echo method is explained in how ultrasonic testing works for railways.

What does 213.113 require when UT finds a defect in a turnout?

For FRA-regulated track, 49 CFR 213.113 prescribes the remedial action for each rail defect type and size, so disposition is not left to case-by-case judgment. The owner must verify an indication within four hours, or immediately if the defect would need remedial action A, A2 or B, and then replace or repair the rail or apply the action in the table.

The 213.113(c) table sets these actions for the defects most often found in turnout rails (2014 rule, current in the 2024 CFR):

What the letters mean, briefly: A is visual supervision of every movement; A2 is a visual inspection then up to 10 mph for up to 24 hours; B is a speed set by a designated person with at least one year of supervisory experience, not over 30 mph; C and D are joint bars within 10 and 7 days; E is joint bars bolted under 213.121; F and G are reinspection within 90 and 30 days; H is 50 mph and I is 30 mph. Each speed is "or the maximum allowable speed for the class, whichever is lower". The full table and notes are in our FRA defect codes guide.

Switch points have their own rule. Joint bars often can't be fitted to a tapered point, so 213.135(b) says:

"Broken or cracked switch point rails will be subject to the requirements of § 213.113, except that where remedial actions C, D, or E require the use of joint bars, and joint bars cannot be placed due to the physical configuration of the switch, remedial action B will govern..." — 49 CFR 213.135(b), eCFR

The rule adds that the designated person setting the speed under B takes into account any added safety from reinforcing bars on the points.

For frogs, Part 213 is less direct. Section 213.139(c) requires a spring rail frog with a bolt hole defect or head-web separation to be replaced. For solid manganese castings, Part 213 gives wear limits (213.137) but doesn't say whether a crack in the casting is a 213.113 "rail" defect. Owners should define this in their own procedures, and a Level III or the owner's 213.7 designated person should confirm the approach.

Worked example: the same defect in a closure rail and in a switch point

A 213.237 test on a Class 4 freight main line (class maximum 60 mph) finds a detail fracture sized at 30% of the head area. The table row for detail fractures puts 30% in the 25% to less than 60% band, so the action is D.

The arithmetic is simple, but the outcome differs: the same 30% defect lets a closure rail return to 50 mph within a week, while the switch point stays at 30 mph or less until the point is changed. That's a strong reason to plan a spare switch point for busy turnouts.

Key takeaway: detection is UT's job; disposition on US track is set by 213.113, with the 213.135(b) exception for switch points.

What can't ultrasonic testing find in switches and crossings?

Ultrasonic testing can't reliably search cast manganese frogs and diamonds with standard probes, the thin end of a switch point, or areas the probe can't reach. An engineer quoting a test report should know these limits.

• Cast manganese frogs and diamonds: coarse grain scatters the beam; FRA RR 22-25 says revenue-service examination is visual only, and TWI says in-service crossings were limited to visual and dye-penetrant inspection.
• Tapered switch point tips: the section is too thin and narrow for a standard rail probe.
• Joint bars: rail flaw equipment searches the rail between the bars (213.237(f)), not the bars themselves.
• Rail base under plates and braces: TTCI research (2000) found that test cars miss much of the rail base, which is why thermite welds often need hand UT of the base.
• Masking by surface damage: shelling, head checks, batter and metal flow can hide internal defects or prevent coupling, giving a stretch that wasn't validly searched.
• Unfavorable orientation: a crack parallel to the beam returns little energy. Several probe angles reduce, but don't remove, this risk.
• Wear and gage: UT doesn't measure frog point wear, flangeway depth or guard check gage. Those come from the 213.137 and 213.143 measurements on the monthly walk.

None of these limits is a reason to skip testing. They are reasons to state coverage honestly in the report and to pair UT with visual, dimensional and surface inspection.

A switch and crossing inspection checklist

Use this as a scope list for an owner's program or an RFP:

1. Confirm what "crossing" means in the contract: diamond, frog or road crossing.
2. List every switch, turnout and track crossing with its track class and whether it carries passenger or hazmat traffic.
3. Schedule 213.235 monthly on-foot inspections, and 3-monthly operation of single-rod switches on Class 3 to 5 track.
4. Measure frog point and tread wear, flangeway width and depth, and guard check and guard face gage against 213.133 to 213.143.
5. Include the rail through each turnout in the 213.237 internal testing program, and require the report to list unsearched areas by turnout.
6. Hand test switch heel joints, welds and the full-section part of switch points where the vehicle search can't reach.
7. Plan visual and surface inspection (PT or MT where suitable) of manganese frog points and casting joints.
8. Apply 213.113 actions on confirmation, using 213.135(b) for switch points and 213.139(c) for spring frogs.
9. Keep 213.235 records for at least one year and internal rail inspection records under 213.241.

For general turnout maintenance topics beyond UT, see turnout inspection.

How EMA Industries applies this

EMA Industries performs ultrasonic rail flaw testing for transit agencies, short lines and industrial track owners, from Southern California and mobilizing nationwide. On turnouts, the useful deliverable is a report that names each switch and frog, states which parts were searched by UT and which were not, and gives each defect's type, size and location so the owner can apply 213.113 without guesswork. To scope turnout and rail testing, see our ultrasonic rail testing services or contact EMA Industries.

Related reading: Grade crossing rail inspection · Rail joint inspection: how UT detects joint bar cracks · FRA defect codes explained (49 CFR 213.113) · 49 CFR 213.237 rail flaw detection

Frequently asked questions

How often must railroad switches be inspected under FRA rules?

At least monthly, on foot, under 49 CFR 213.235(a), for every switch, turnout and track crossing on Class 1 to 5 track. Track used less than once a month is inspected before use. On Class 3 to 5 track, a switch held only by the operating mechanism and one connecting rod must be operated to all positions once every 3 months.

Why are manganese frogs hard to test with ultrasound?

Cast austenitic manganese steel has coarse, uneven grains (about 0.94 mm mean diameter in FRA's RR 22-25 sample) that scatter conventional high-frequency ultrasound into noise. FRA reports that revenue-service examination of frogs and diamonds is currently visual only. Low-frequency UT and SAFT imaging have shown promise in research.

What remedial action applies to a cracked switch point?

The 213.113 table applies by defect type and size. Where that action is C, D or E, which need joint bars, and the bars can't be fitted to the switch, 213.135(b) substitutes remedial action B: a speed of no more than 30 mph set by a designated person with at least one year of supervisory experience.

What speed limit applies to a worn frog?

Under 213.137, speed over a frog is limited to 10 mph if the point is chipped, broken or worn more than 5/8 in. down and 6 in. back, or if the tread of the casting is worn more than 3/8 in. below its original contour.

Does "crossing" in switch and crossing inspection mean a road crossing?

No. It means a rail-to-rail crossing (a diamond) or, in UK usage, the frog of a turnout. Highway-rail grade crossings raise different problems, mainly hidden web and base corrosion under the crossing surface.

Sources

• 49 CFR 213.235, Inspection of switches, track crossings, and lift rail assemblies, eCFR, read on Cornell LII October 2026
• 49 CFR 213.133, Turnouts and track crossings generally, eCFR
• 49 CFR 213.135, Switches, eCFR
• 49 CFR 213.137, Frogs, eCFR (as amended at 85 FR 63388, Oct. 7, 2020)
• 49 CFR 213.139, Spring rail frogs, eCFR
• 49 CFR 213.141, Self-guarded frogs, eCFR
• 49 CFR 213.143, Frog guard rails and guard faces; gage, eCFR
• 49 CFR 213.113, Defective rails, eCFR; remedial action table from the 2024 CFR edition on govinfo
• 49 CFR 213.233, 213.237, 213.241 and 213.119, eCFR
• Cornell LII text of 49 CFR 213.137 (alternate copy)
• Federal Railroad Administration, Feasibility of Special Trackwork Inspection Using Ultrasonics, Research Results RR 22-25, September 2022 (R. Wilson, FRA; A. Poudel, TTCI)
• TWI, RailSAFT: ultrasonic inspection of cast austenitic manganese steel railway crossings
• Evident, Material sound velocities (302 stainless steel longitudinal velocity, used as a proxy)