Ultrasonic Testing

Rail Ultrasonic Testing Technology: What Changed Since 2014

Updated

ultrasonic testing technology rail safety

Rail ultrasonic testing technology in the US has changed more through rules than through hardware since 2014. FRA's January 24, 2014 rule made 49 CFR 213.237 performance-based and created the 213.238 qualified operator. The October 7, 2020 rule added 213.240 continuous testing. Ultrasound remains the primary method, and no public data shows a step change in detection.

This article is a dated, sourced timeline for track owners, transit engineers and short-line supervisors who buy rail flaw testing. It covers the two FRA rules that reshaped the work, the shift to software-assisted data review, the records rules, the phased array and other trials reported by TTCI (now MxV Rail) and FRA, the vendor systems that followed, and what FRA did in 2024 to 2026. It ends with what an owner should do about each change.

For how each detection method ranks by maturity, see internal rail defect detection methods. For a side-by-side of the two test processes, see continuous vs stop-and-verify rail testing.

What has changed in rail ultrasonic testing technology since 2014?

Two FRA rules, not a new kind of probe, account for most of the change. The 2014 rule changed how often rail must be tested and who may run the equipment. The 2020 rule allowed non-stop testing with central data review. Research since then has tested phased array, new statistics for probability of detection (POD) and automated classification, but none of it has replaced the fluid-filled wheel probe and its 0°, 37–45° and 70° beams as everyday practice.

DateEventWhat it changed
Jan 24, 2014 (effective Mar 25, 2014)FRA final rule Track Safety Standards; Improving Rail Integrity, 79 FR 4234213.237 service failure rate targets and segments; valid search defined; plug rail testing; new 213.238 qualified operator; revised 213.113 table; 213.241(c) record items
2014–2016TTCI digests on rolling contact fatigue (RCF) measurement and high-speed phased arrayResearch only; no change to practice or rules
Nov 2016TTCI TD-16-050, phased array roller search unit prototypePrototype tested at up to 20 mph on the FAST test loop
2009–2020Continuous testing run under 213.17 waivers (CSX, CN, NS, KCS, BNSF, UP, LIRR, Metro-North)Operating experience that became 213.240
Oct 7, 2020FRA final rule Rail Integrity and Track Safety Standards, 85 FR 63362New 213.240 continuous rail testing; new 213.241(f) and (g) records; annual report
Sept 2021FRA RR 21-18 and RR 21-19 (MAPOD feasibility and roadmap); TTCI TD-21-027 (thermite weld technique)Research only
Jan 10, 2022A vendor launches a branded hi-rail ultrasonic rail flaw system (see below)Commercial product; no rule change
2022FRA RR 22-25, UT feasibility in special trackworkResearch; cast manganese frogs remain hard to test
Oct 24, 2024FRA NPRM on track geometry measurement (proposed 213.236)A geometry proposal; doesn't touch rail UT
May 15, 2026 (effective July 14, 2026)Part 243 training rule amendment, 91 FR 27849Training program duties for railroads and contractors
July 2026FRA DOT/FRA/ORD-26/07, MAPOD for handheld rail UTResearch on POD, thresholds and operator calibration
Aug 31, 2026 (effective Sept 30, 2026)Part 214 workplace safety rule, 91 FR 55749Roadway worker rules trimmed; no rail testing change

Key takeaway: the rail flaw rules a test contractor works to today, 213.237, 213.238, 213.240 and 213.241, were last amended in 2014 and 2020.

What did the 2014 rule change in 213.237 and 213.238?

The 2014 rule replaced a fixed time-and-tonnage interval with a service failure target backed by maximum intervals. It also defined a valid search, made plug rail testing a rule and set minimum qualifications for the people who run rail flaw detection equipment. FRA issued it under section 403 of the Rail Safety Improvement Act of 2008, under Docket FRA-2011-0058.

FRA described the shift to ultrasonic testing directly in the preamble:

"This rulemaking changes the ultrasonic testing of rail from a standard based on time and tonnage to one based on self-adaptive performance goals." (FRA, Track Safety Standards; Improving Rail Integrity, 79 FR 4234, January 24, 2014)

The preamble explains that railroads were already using a "self-adaptive scheduling method", in which test frequency is "established annually" from the detected defect rate per test, the rate of service failures between tests and the tonnage between tests. The rule wrote that practice into federal law and kept maximum intervals as a backstop.

ItemBefore the 2014 ruleAfter the 2014 rule (current 213.237)
Class 4–5, and Class 3 with passenger trains40 MGT or once a year, whichever is shorter370 days or 30 MGT, whichever is shorter (Class 3 also when it is a hazmat route)
Other Class 330 MGT or once a year, whichever is longerOnce a calendar year (≤18 months apart) or 30 MGT, whichever is longer, never more than 5 years
Performance targetNone0.10, 0.09 or 0.08 service failures per mile per year, by class, passenger and hazmat status
Rail inspection segmentsNoneDesignated by the owner by March 25, 2014; changes need FRA approval
Valid searchNot definedDefined in 213.237(j)(4); untested track doesn't count as inspected
Plug railFRA Safety Advisory 2006-02 (guidance)Must have been tested; no more than 30 MGT since its last test
OperatorNot definedQualified operator under new 213.238

The valid search definition is the part that matters most in the field. Section 213.237(j)(4) reads: "a continuous inspection for internal rail defects where the equipment performs as intended and equipment responses are interpreted by a qualified operator as defined in § 213.238." If the operator decides a valid search couldn't be made over a length of track, that length isn't inspected (213.237(g)). Before the time or tonnage limit runs out, the owner must re-test it, hold trains to 25 mph, or replace the rail (213.237(h)).

The new 213.238 requires a written training program, at least 160 hours of rail flaw detection experience under a qualified operator or a manufacturer's representative, written employer authorization and demonstrated proficiency on each type of equipment. The operator's duties are to conduct a valid search, determine that the equipment "is performing as intended", and interpret responses under the employer's procedures. FRA created no federal certificate; the employer qualifies the operator.

The rule also revised the 213.113 remedial action table. For transverse fissures, detail fractures and defective welds, the size at which remedial action A2 or E and H applies dropped from 80% to 60% of head area. It added a 4-hour window for the owner to verify an indication. For the size bands and actions, see FRA defect codes explained, and for the interval rule itself, see 49 CFR 213.237 rail flaw detection requirements.

Key takeaway: since 2014 the test interval is set by results, and a search only counts if a qualified operator judged it valid.

Worked example: how the service failure targets work

A service failure rate is the number of service failures in 12 months divided by the segment's miles. A service failure is a broken rail caused by a transverse fissure, compound fissure, detail fracture or vertical split head (213.237(j)(3)). The owner calculates the rate within 45 days of the end of each 12-month period.

Take two rail inspection segments on one owner's network.

InputSegment ASegment B
Length62 miles7 miles
Class and trafficClass 4, scheduled passenger trains and a hazmat routeClass 4, freight only, not a hazmat route
Target under 213.237(a)0.08 per mile per year0.10 per mile per year
Failures allowed before exceeding0.08 × 62 = 4.96, so 40.10 × 7 = 0.7, so 0
Service failures this period51
Rate5 ÷ 62 = 0.0811 ÷ 7 = 0.143
ResultTarget missedTreated as compliant under 213.237(d)(2)

Segment A misses its target by one failure: four failures would have given 4 ÷ 62 = 0.065. The owner must tell FRA within 45 days of the end of the period and may send an explanation and a remedial plan (213.237(d)). If the target is missed for two years in a row, then in the area with the most failures the owner must either cut the test interval to 10 MGT or drop the track to Class 2 until the target is met (213.237(d)(1)).

Segment B looks worse on paper, because a single failure on 7 miles gives 0.143. Section 213.237(d)(2) covers this case: where one failure alone would push the rate over the target, the segment is treated as compliant unless a second failure occurs in the same 12-month period. On any Class 4 segment shorter than 10 miles, one failure trips the rate, so short segments rely on this rule.

The practical lesson is that the 370-day or 30 MGT maximum is a ceiling, not a schedule. An owner whose failure count is climbing has to test more often than the ceiling to stay inside the target.

What did the 2020 continuous testing rule add?

The 2020 rule added 49 CFR 213.240, which lets an owner test non-stop and verify suspect locations afterward, within 84 hours, or 36 hours for the most serious suspects. Before 2020, 213.113(b) required verification within 4 hours, or immediately for suspects that would need remedial action A, A2 or B. FRA wrote that this timeframe "has made it practically impossible for track owners to conduct continuous testing" (85 FR 63362).

The rule was effective on publication, October 7, 2020 (Docket FRA-2018-0104, RIN 2130-AC53). It followed about a decade of waivers under 213.17, starting in 2009, for CSX, CN, NS, KCS, BNSF, UP, the Long Island Rail Road and Metro-North. In the preamble FRA reported results from those waivers, including 34% fewer main-line service failures on NS in 2018 than in 2014. These are FRA's statements about waiver operations, not controlled comparisons.

The main parts of 213.240 are:

  • Choice. Continuous testing is optional. An owner who uses it is exempt from 213.113(b) only; "all other requirements of § 213.113 apply" (213.240(a)).
  • Designation. The type of test, continuous or stop-and-verify, is recorded before testing starts. A change mid-run is documented with times and mileposts (213.240(c)).
  • People. Vehicle operators are qualified under 213.238 except for the interpretation duty, the data is "reviewed and interpreted by a person qualified to interpret the equipment responses", and every suspect is field-verified by a 213.238 person (213.240(d)).
  • Clocks. Field verification within 84 hours of the indication, or 36 hours for suspects that would need A, A2 or B. A broken rail with separation needs immediate protection. An unverified suspect gets the most restrictive remedial action for its suspected type and size (213.240(e)).
  • Report. Owners using continuous testing send FRA an annual report within 45 days after each calendar year (213.240(g)).

The rule didn't change the 213.237 intervals or the 213.113 remedial action table. For the full comparison of the two processes, including cost and track fit, see continuous vs stop-and-verify rail testing.

Key takeaway: continuous testing moved interpretation from the truck to an analyst and moved verification to a scheduled field crew, but it kept every suspect subject to human field verification.

How have automated recognition and B-scan review changed the operator's job?

Software now sorts most of the data, but a qualified person still decides what each indication is. Rail test vehicles display the channel data as B-scan style side views with A-scans for detail, and their software flags patterns for the operator or analyst to review. Part 213 has no rule for the software. It has rules for the people.

The 2020 preamble describes how waiver railroads used the data. Suspect locations "are categorized based on the ultrasonic reflective responses viewed by the analyst", and "analysts interpret the collected ultrasonic reflective responses, estimate each indication type and size". The analyst then assigns a priority. In waiver practice, priority one suspects (would need A, A2 or B) were about 95–99% confirmed in the field, while priority three suspects (questionable responses) were under 2% confirmed. Even so, FRA's waiver data shows priority threes accounted for about 85% of the field-verified defects found by continuous testing, so a low confirmation rate doesn't make a suspect safe to ignore. Those are industry categories from the waivers, not rule text.

Three points follow for anyone buying testing:

  1. Automated recognition is a filter, not a call. Under 213.240(e)(4), "A suspect location is not considered a defect under § 213.113(c) until it has been field-verified by a person qualified under § 213.238." On a stop-and-verify test the operator verifies on the spot.
  2. The analyst is a separate qualification. Section 213.240(d)(2) requires the analyst's employer to keep records of each qualification, the equipment it covers and the most recent reevaluation. FRA left the content of that training to the owner.
  3. Classifier accuracy is not POD. A 2025 paper (Hu and co-authors, International Journal of Steel Structures) reported 88.56% accuracy for a two-stage classifier on rail UT B-scan images from the authors' own dataset. That figure measures one model on one image set. It isn't a probability of detection and doesn't transfer to another vehicle's data.

For how to read the displays themselves, see A-scan, B-scan and C-scan results in rail UT, and for what drives benign suspects, see false calls in rail UT.

Key takeaway: the change since 2014 is where interpretation happens and who signs it, not whether a person does it.

What changed in rail testing data and records?

The 2014 rule made internal rail records show exactly what was tested and what wasn't, and the 2020 rule added timestamped records for continuous testing. Both sit in 49 CFR 213.241, whose amendment history runs 1998, 2005, 2014 and 2020.

The 2014 change answered a recommendation from the DOT Office of Inspector General that FRA require railroads to report all track locations covered during internal rail flaw testing. Section 213.241(c) now lists six items for each internal rail inspection: the date; the track inspected, with beginning and end points; the location and type of defects found; their size if not removed before the next train; the initial remedial action and its date; and the "location of any track not tested pursuant to § 213.237(g)".

Records ruleSourceWhat it requires
Internal rail inspection record213.241(c), 2014Six items, including limits tested and untested track
Retention213.241(d)At least 2 years after the inspection and 1 year after initial remedial action
Service failure rate records213.241(e)Kept to support the 213.237(a) calculation
Continuous test records213.241(f), 2020Everything in the 213.240(e) suspect listing, whether the test counts for 213.237, and dates and times of the run, of each suspect's identification and of its field verification
Continuous testing operations213.241(g), 2020Records sufficient to monitor compliance, available to FRA
Electronic records213.241(j)Unique author identity, no change once complete, storage started within 72 hours, amendments stored separately
Annual report213.240(g), 2020Per segment: class, annual MGT, numbers of stop-and-verify and continuous tests, defects and service failures

Here is how the 2020 timestamps work on one run. A continuous test car flags two suspects at 10:00 on Monday, November 2, 2026. One would need remedial action B if confirmed; the other is a lower-priority suspect. The first must be field-verified by 22:00 on Tuesday, November 3 (36 hours). The second must be verified by 22:00 on Thursday, November 5 (84 hours). The 213.241(f) record must show the identification time and the verification time for each, so an auditor can check both clocks from the record alone.

Two gaps remain. The annual report doesn't count benign suspects, so there is no federal false-call rate. And an instrument's stored files don't meet 213.241(j) on their own unless the export and database supply the author identity and lock-down controls.

Key takeaway: since 2014 a rail test record must prove coverage, not only list defects.

What have phased array and other trials shown?

Trials since 2014 have shown promise for phased array in verification and research, but none has changed what a regulated rail test uses. The reports below are the main public US work, by date. Each is a test-track, lab or simulation result, and none reports a field POD improvement for revenue testing.

DateReportWhat it found
2014TTCI TD-14-007 and TD-14-008RCF measurement: dye penetrant and magnetic particle can't measure below the surface, and for eddy current and UT "accurate measurement of crack angle and depth is still a primary limitation"
July 2016TTCI TD-16-030, High-Speed Rail Flaw Detection Using Phased Array UltrasonicsConventional UT finds "about 90 percent" of flaws, so about 10% go undetected in periodic inspection; on worn rail fixed probes inspect "the wrong volume of material"
Nov 2016TTCI TD-16-050, Characterizing Rail Flaws Using Phased Array UltrasoundPhased array roller search unit fitted in place of a conventional one; detection mode up to 20 mph on the FAST loop; reported as equivalent or better than a conventional truck on the defects tested; can't inspect the base flanges
Nov 2018TTCI TD-18-027Rails removed for low-confidence UT: 90% of RCF cracks were 0.040 in. deep or less; more grinding passes would likely have improved UT confidence
May 2020TTCI TD-20-007An electromagnetic (not UT) RCF system on a hi-rail carriage at up to 20 mph; slight loss of sensitivity with speed from wider sample spacing
Sept 2021FRA RR 21-18 and RR 21-19 (ENSCO)Model-assisted POD (MAPOD) is most feasible for UT rail flaw detection; roadmap for applying it
Sept 2021TTCI TD-21-027Phased array technique for thermite welds after installation; full base coverage needed a full skip
2022FRA RR 22-25Cast austenitic manganese frogs are "highly attenuative" and coarse-grained, so UT in special trackwork stays difficult
2023MxV TD-23-017Low frequency (0.5 MHz) worked best for UT of standard joint bars
July 2026FRA DOT/FRA/ORD-26/07 (ENSCO)MAPOD for handheld rail UT: POD falls as the detection threshold rises; operators' calibrations differed; no standardized calibration procedure for handheld units in North America

Two of these matter most for owners. TD-16-050 is the clearest public evidence for phased array on a test vehicle, and it was a prototype on a test loop. ORD-26/07 matters because it puts numbers on something field crews already knew: the threshold and the operator's calibration move the detection rate. In its simulated dataset, mean POD for a flaw of 25% head area was 95.8% at a 0.25 threshold and 86.7% at 0.40. FRA warns that "too low a detection threshold can result in an unacceptable number of false positives".

For where each method stands today, see internal rail defect detection methods, which ranks them by maturity, and phased array ultrasonic testing for the technique itself.

Key takeaway: phased array in rail is still verification and research; the public studies since 2014 measured limits as much as gains.

Where do new vendor rail flaw systems fit?

New commercial systems package wheel probes, data software and other inspection sensors, but they work under the same 213.237, 213.238 and 213.240 rules as older equipment. No standard or FRA rule defines a product category for them.

One example: on January 10, 2022, ENSCO Rail announced its Ultrasonic Rail Flaw System (URFS) as part of its suite of track inspection products. The company's own page lists goals of fewer false stops, more automation and integration with its track geometry and machine vision systems. Those are the vendor's stated goals, not independent test results. Other makers sell hi-rail and rail-bound systems with similar descriptions, and their published speeds are rated maximums, not working averages.

When a contractor proposes any system, the questions are the same:

  • Is the operator qualified under 213.238 on that specific equipment?
  • If the test is continuous, who interprets the data and where is that person's 213.240(d)(2) qualification record?
  • How is the system set up on reference rail, and how often is the setup checked?
  • Does the output give the six 213.241(c) items and the untested track?

For what sits inside such a system, see ultrasonic rail flaw detection system parts and checks. For what test speed does to detection, see high-speed ultrasonic rail inspection. For the wider range of cars that measure geometry, gage restraint and imaging, see the role of railway inspection cars.

Key takeaway: a product name says nothing about compliance; the operator, the analyst and the record do.

What did FRA do in 2024 to 2026?

FRA's 2024 to 2026 actions touched geometry, track surface, training and roadway worker safety, but none amended the rail flaw sections. On Cornell LII's copy of the CFR, checked in October 2026, the latest amendments are 2014 for 213.237 and 213.238 and 2020 for 213.240 and 213.241.

DateActionStatusEffect on rail UT
Oct 24, 2024NPRM, Track Geometry Measurement System (TGMS) Inspections (FRA-2024-0032), proposing a new 213.236Proposal; comment period extended to Jan 7, 2025; no final rule found as of October 2026None directly. It would require TGMS runs 3 times in 365 days on busier Class 1–5 track
Apr 28, 2026Final rule repealing the 213.63 runoff limit (91 FR 22733)Effective 30 days after publicationNone. Runoff is a surface geometry limit
May 15, 2026Part 243 training rule amendment (91 FR 27849; FRA-2020-0017)Effective July 14, 2026Indirect. Contractor staff who inspect track under Parts 213 and 214 are covered by Part 243 on top of 213.238
Aug 31, 2026Part 214 rule Repealing Outdated Railroad Workplace Safety Requirements (91 FR 55749)Effective Sept 30, 2026None to testing; it trims roadway worker rules (adds 214.119, removes and reserves 214.331 and 214.333)

The Part 243 amendment is the one a test contractor should read. Under 243.3 and 243.5, a contractor's employees who inspect track "in the application of parts 213 and 214" are safety-related railroad employees, and 243.1(d) says Part 243 adds to other training rules rather than replacing them. FRA says the 2026 amendment codifies its guidance, drops some submissions, lets each railroad and contractor decide who conducts periodic oversight and delays refresher requirements for small entities by one year.

The Part 214 rule matters for test crews only because every on-track test still needs roadway worker protection; it didn't create a testing requirement. The geometry NPRM is worth watching because it would make automated geometry runs a Part 213 duty for larger railroads. If it becomes final, it still won't change 213.237.

Key takeaway: as of October 2026, the last change to the federal rail flaw rules was the 2020 continuous testing rule.

What hasn't changed: the limits of rail UT

Ultrasound still can't see through every rail condition, and Part 213 still names no test method. The word "ultrasonic" appears nowhere in Part 213; 213.237 requires "internal rail inspections" with equipment capable of "detecting defects between joint bars". FRA's 2014 preamble describes ultrasonic as "the primary technology used, and induction technology is currently used as a complementary system."

These limits are the same in 2026 as in 2014:

  • Surface masking. FRA's 2014 preamble refers to "the limitations of current technology to detect internal rail flaws beneath surface conditions and in the base flange area". Shells, head checks and spalls block the beam from what lies beneath, which hides detail fractures in particular. In TTCI's 1999 benchmark (TD-99-006), detection of transverse defects under shells was 62% against 77% with no surface anomaly, on only 7 flaws under shells.
  • The base. Vehicle testing covers the head, the web and the base directly under the web, not the base flanges. TD-16-050's prototype had the same limit.
  • Special trackwork. Cast manganese frogs scatter sound too strongly for routine UT (RR 22-25).
  • Worn rail. Fixed-angle probes on curve-worn rail can inspect the wrong volume (TD-16-030).
  • Thresholds and calibration. ORD-26/07 showed POD moves with the detection threshold and with each operator's calibration.
  • Speed. No public TTCI or FRA study measures POD of conventional contact rail UT as a function of test speed.
  • No published step change. No public US study since 2014 reports a field-measured POD improvement for revenue rail testing. Claims of "over 95% accuracy" or "40% lower maintenance cost" have no traceable source and shouldn't be repeated.

Any of these can turn a stretch of track into one where a valid search couldn't be made. When that happens, 213.237(g) and (h) apply, and the record must show it under 213.241(c)(6).

What should a track owner do about these changes?

Treat the changes as a compliance checklist: confirm the segment targets, the operator and analyst qualifications, the records and the test type for each run. These steps apply whether the owner tests in-house or hires a contractor; under 213.5 the owner stays responsible for compliance.

  1. Know each segment's target. List every rail inspection segment, its class, passenger and hazmat status and its 213.237(a) rate. Calculate the rate within 45 days of each 12-month period, and test more often than the 370-day or 30 MGT ceiling when failures rise.
  2. Ask for 213.238 records. For each operator, get the training program, the 160-hour experience record, the written authorization and the equipment types covered.
  3. Name the test type in the contract. Say whether each run is stop-and-verify or continuous, and whether a continuous run counts for 213.237. If continuous, name who interprets the data and how fast suspects reach the verification crew.
  4. Specify the record. Require the six 213.241(c) items, untested track with mileposts and, for continuous runs, the 213.241(f) timestamps. Ask how electronic files meet 213.241(j).
  5. Plan for untested track. Agree in advance who re-tests invalid-search stretches and how fast, so the owner isn't left choosing between a 25 mph restriction and replacing rail.
  6. Check Part 243. Ask a contractor for the 243.101(e) statement about its training program, and keep it on file.
  7. Discount unsourced claims. Ask any vendor for the report number behind a detection figure, the defect population and the threshold used.
  8. Watch the geometry NPRM. If FRA finalizes 213.236, larger railroads will need scheduled automated geometry runs; this is separate from rail flaw testing.

For how often each class of track must be tested, see how often must rail be tested.

How EMA Industries applies this

EMA Industries performs ultrasonic rail flaw testing for transit agencies, short lines and industrial track owners, and is based in Southern California, mobilizing nationwide. Most of what changed since 2014 shows up in the deliverable: the report should name the qualified operator and the equipment, record the segment limits, mark every stretch where a valid search couldn't be made, and give the type, size and location of each defect so the owner can act under 213.113. To scope a testing program or check a specification against 213.237–213.241, see our ultrasonic rail testing services or contact us.

Related reading: Internal rail defect detection methods · Continuous vs stop-and-verify rail testing · 49 CFR 213.237 requirements · False calls in rail UT

Frequently asked questions

What was the biggest change to US rail flaw testing rules since 2014?

The 2014 rule, which made 213.237 performance-based. Each segment now has a service failure target of 0.10, 0.09 or 0.08 per mile per year, with 370 days or 30 MGT as the maximum interval on Class 4–5 track. The same rule created the 213.238 qualified operator and the valid search definition.

FRA added 49 CFR 213.240 in its October 7, 2020 final rule, effective that day. Before then, railroads tested continuously only under 213.17 waivers, the first in 2009, because 213.113(b) required verification within 4 hours.

Does AI now decide whether a rail indication is a defect?

It doesn't. Software can flag and sort indications, but under 213.240(e)(4) a suspect isn't a defect until a person qualified under 213.238 has field-verified it. On continuous tests the data must also be interpreted by a separately qualified analyst.

Is phased array now used for routine rail testing?

Not as standard practice. The main public US work, TTCI's 2016 phased array prototype, was tested on the FAST loop at up to 20 mph. Phased array in rail remains a verification and research tool, used for tasks such as thermite weld characterization.

Did FRA change the rail testing rules in 2026?

Not the rail flaw sections. FRA amended Part 243 training rules (effective July 14, 2026) and Part 214 roadway worker rules (effective September 30, 2026), and repealed the 213.63 runoff limit. The 2024 geometry proposal (213.236) had no final rule as of October 2026.

Sources

← All articles