Liquid Penetrant Testing in Railways: Complete Guide

Liquid penetrant testing (PT) is a surface NDT method that shows cracks and other discontinuities open to the surface of nonporous materials. US work is usually done to ASTM E165/E165M-23 or ASME Section V, Article 6, within a standard temperature range of 40–125 °F, and PT is chosen where magnetic particle testing (MT) can't be used.

That last point decides most railway applications. MT is faster and finds slightly subsurface flaws, so on ordinary carbon steel it usually wins. Liquid penetrant testing earns its place on materials that aren't ferromagnetic, such as cast austenitic manganese frogs, stainless steel and aluminum, and on structural welds where the contract or code allows PT as the surface method.

This guide is written for the people who buy and specify the work: weld QA managers, track and structures engineers, and maintenance supervisors. It covers where PT fits, which standards apply, the process variables a procedure must control, a timed fillet weld walk-through and the method's limits.

What is liquid penetrant testing, and what can it find?

Liquid penetrant testing draws a colored or fluorescent liquid into surface-breaking openings by capillary action, then pulls it back out with a developer so the opening shows as a visible indication. It finds only what is open to the surface.

ASTM's general-industry practice lists the kinds of discontinuity the method is meant for. In the scope of ASTM E165/E165M-23, penetrant testing is described as a method for discontinuities "open to the surface such as cracks, seams, laps, cold shuts, shrinkage, laminations, through leaks, or lack of fusion". The same scope applies it to in-process, final and maintenance examinations of nonporous metals, ferrous and nonferrous, and of some nonporous nonmetals.

Every PT system has the same five working parts. The surface has to be clean and dry so the opening isn't plugged. The penetrant has to stay on the surface long enough to enter the opening (the dwell time). The excess has to come off the surface without being washed out of the opening. The developer has to draw the trapped penetrant back out and give a contrasting background. Finally, the inspector has to look under the right light within a set time window.

Liquid penetrant testing tells you that a discontinuity is there, where it is on the surface and roughly how long it is. It doesn't tell you how deep it goes. ASTM E165 says the method indicates presence and location, and "to a limited extent, the nature and magnitude" of a discontinuity. Depth needs grinding, ultrasonic testing (UT) or another method.

Key takeaway: PT is a surface-only method for nonporous material; it gives location and length, not depth.

Where does liquid penetrant testing fit in railway and structural work?

In US rail and structural work, liquid penetrant testing is mainly used on non-magnetic materials and on welds where PT is allowed as the surface method. Most carbon steel rail components are examined by MT, UT or visual inspection instead.

The table summarizes where PT is normally the right call and where another method usually leads. The confidence column shows how firm each statement is; a Level III should confirm the hedged rows against the governing specification.

Austenitic manganese frogs and crossings

Cast austenitic manganese steel (AMS) frogs and crossing diamonds are the clearest railway case for liquid penetrant testing. AMS is used because it work-hardens under traffic, but it is essentially non-magnetic, so a yoke can't magnetize it and MT doesn't apply. That leaves PT as the practical surface method once a crack is suspected or a repair is ground out.

UT is not a simple substitute on the casting. An FRA research report on special trackwork found that AMS "is highly attenuative" and "coarse-grained, nonhomogeneous, and anisotropic", and it describes current in-service practice in one line:

"Revenue service examinations of these special trackwork components currently use only visual inspections." — FRA, Feasibility of Special Trackwork Inspection Using Ultrasonics, RR 22-25 (Wilson and Poudel, September 2022)

The same report notes that visible cracks are ground out until they are gone or until a depth limit is reached, and it calls that process "highly subjective". Liquid penetrant testing on the ground-out groove is one way to show that the crack has actually been removed before the frog is repaired by welding or returned to service. Whether an owner requires PT at that step is set by the owner's frog maintenance procedure, so check it rather than assuming. For the wider inspection picture, see our guide to railroad switch and crossing inspection.

Stainless steel and aluminum

Austenitic stainless steel and aluminum alloys can't be magnetized, so PT is the standard surface method on them. On rail property they turn up in passenger car bodies and underframes, equipment enclosures, platform and station steelwork, handrails and piping.

Two cautions apply to these materials. Penetrant materials can contain sulfur and halogens that harm some alloys, and ASME Section V Article 6 limits sulfur for nickel-base alloys and chlorine plus fluorine for austenitic stainless steel and titanium (paragraph T-641; check the limit in the edition your contract names). Aluminum welds are fabricated to AWS D1.2/D1.2M, whose current edition is D1.2/D1.2M:2026, not to D1.1. Make sure the PT acceptance criteria in the contract come from the code that governs the weld.

Structural welds under AWS D1.1 and ASME

Structural steel welds are where EMA Industries is most often asked for liquid penetrant testing. AWS D1.1/D1.1M:2025 treats MT and PT as surface methods the contract can call for in addition to visual inspection. CWI study guides for the 2025 edition place MT and PT in Clause 8 and give ASTM E165 as the PT practice and ASTM E709 as the MT practice; confirm both against the edition in your contract.

How a D1.1 PT indication is judged is a common source of argument. Inspectors on the AWS forum report that welds examined by MT or PT are evaluated against the visual acceptance criteria, and that D1.1 judges the actual discontinuity once the developer is wiped off, not the size of the bleed-out. That reading is widely held but comes from secondary sources, so treat it as CONFIRM. Cracks are unacceptable at any size under the D1.1 visual criteria. For the full D1.1 picture, see our article on AWS D1.1 weld inspection criteria.

For pressure equipment and piping, ASME BPVC Section V Article 6 sets how PT is done, and the construction code (Section VIII, B31.3 and others) sets what is acceptable. Section V itself contains no acceptance criteria.

Thermite welds in track

Liquid penetrant testing is not a routine test for thermite (aluminothermic) rail welds in North America. A 2021 TTCI technology digest, Phased Array Ultrasonic Beam Modeling for Rail Thermite Welds (TD21-027), states the current position:

"Existing practices for flaw detection and characterization in thermite rail welds after installation are limited to visual inspection, which does not reveal purely internal defects."

Penetrant has been used on thermite welds in research and special investigations. A 1973 paper by J. Banks in the British Journal of Non-Destructive Testing describes dye penetrant and magnetic methods used to examine the running face of thermite-welded rails for surface cracking. Rail steel is ferromagnetic, though, so MT is usually the better surface method on a rail weld, and the defects that break thermite welds are mainly internal or at the web and base. Where an owner wants a surface method on a thermite weld collar or a ground running surface, write it into the weld specification. See aluminothermic weld inspection for the methods that normally apply.

Wheels and axles: mostly UT and MT

The original version of this article said that axles and wheelsets are typical PT parts. In North American freight practice that isn't the normal pattern. Wheels and axles are covered by AAR specifications (M-101 for axles, M-107/M-208 for wheels), and shop inspection relies mainly on UT and MT.

The Casselton, North Dakota derailment of December 30, 2013 shows why. NTSB investigators found a void inside a reused axle. NTSB presentation material describes the secondhand axle as having been requalified with magnetic particle testing, and the AAR moved to require more thorough testing of secondhand axles after an urgent NTSB recommendation. An internal void can't be found by any surface method, PT included.

Federal freight car rules don't name a method at all. Under 49 CFR 215.105, an axle is defective if it "has a crack or is broken". PT can still be useful on a non-magnetic wheelset component or to check a local grind-out, but don't assume it is the inspection method for a carbon steel axle.

Key takeaway: PT leads on manganese, stainless and aluminum; on carbon steel rail parts MT and UT usually lead.

Which standards govern liquid penetrant testing in the US?

Liquid penetrant testing in the US is normally performed to ASTM E165/E165M or ASTM E1417/E1417M, or to ASME Section V Article 6 for code work. None of these sets acceptance criteria; those come from the product code or the contract.

Each standard below was checked on the publisher's store or catalog page in October 2026. ASTM states on its store pages that its standards may not be used with AI tools, so only titles, status and short scope quotes are given here.

Note that the method-specific practices aren't one block of numbers. E1208, E1209, E1210, E1219 and E1220 are the method-specific practices in the E1208 to E1220 range; the designations between them are other ASTM standards, and the visible water-washable practice is E1418.

ASTM E165/E165M and E1417/E1417M

E165 is the broad general-industry practice, and it is the one AWS D1.1 cites. ASTM E165 is explicit that it doesn't decide what passes. Its scope says that after indications are found, "there must be a separate code, standard, or a specific agreement" to define what is acceptable. Contracts that call up only "PT per E165" without an acceptance standard leave the inspector nothing to accept or reject against.

ASTM E1417/E1417M sets minimum requirements for the penetrant process and says it "must be supplemented by a detailed procedure". The 2021 revision changed several process checks. Quality Magazine (Kilty and Geis, April 2022) and Magnaflux report that LED UV-A lamps are now exempt from the 10-minute warm-up still required of other UV-A sources, and that the UV lamp integrity check moved from weekly to daily.

ASTM E433 reference photographs

ASTM E433-71(2023) is a set of reference photographs of indication types. Its scope says the users must specify the maximum length and number of indications, and that "No attempt has been made to establish limits of acceptability". It helps an inspector name what is seen; it doesn't accept or reject anything.

ASME Section V, Article 6

Article 6 is the PT method used under the ASME construction codes. Its main requirements are written as T-6xx paragraphs: a written procedure (T-621), control of contaminants (T-641), surface preparation (T-642), the standard temperature range (T-652) and qualification at other temperatures (T-653), the dwell times in Table T-672, excess-penetrant removal (T-673), developing (T-675) and interpretation (T-676). A change in an essential variable, such as a shorter penetrant dwell, requires the procedure to be requalified (Table T-621). Paragraph numbers here follow the editions that public study guides use; confirm them in the edition your contract names.

ISO 3452-1

ISO 3452-1:2021 covers the general principles of penetrant testing for metals and, with conditions, other materials. Its scope says it "is not intended to be used for acceptance criteria"; ISO 23277 gives acceptance levels for welds. US rail and structural work seldom calls it up, but European rolling stock suppliers may.

Key takeaway: E165, E1417 and ASME V Article 6 say how to test; acceptance always comes from another document.

What are penetrant Types I and II and Methods A to D?

Penetrants are classified by Type (how the indication is seen) and Method (how the excess is removed). The scheme comes from SAE AMS 2644 and is used by ASTM E1417 and most US procedures.

Field work on welds and trackwork is almost always Type II, Method C, the familiar three-can kit of cleaner, red penetrant and white developer. It needs no water, power or darkness, which suits a bridge girder or a turnout. The price of that convenience is sensitivity: a visible dye shows less contrast than a fluorescent one, and solvent removal is easy to overdo.

Fluorescent (Type I) systems are more sensitive, but they need a darkened booth or enclosure and a calibrated UV-A lamp. That suits a shop inspecting machined parts. It is rarely practical on track.

Higher sensitivity also has a cost. A Level 4 fluorescent penetrant on an as-welded surface can leave so much background that real indications are hard to see, which is why procedures match the sensitivity level to the surface finish.

Key takeaway: Most railway field PT is Type II, Method C; Type I fluorescent belongs in the shop.

What dwell times, temperatures and light levels does a PT procedure need?

A liquid penetrant testing procedure has to fix the temperature range, the penetrant dwell, the removal method, the developing time, the interpretation window and the light level. ASME V Article 6 and ASTM E1417 set minimums; the written procedure sets the actual values.

The figures below are what ASME-based procedures and study guides give for the current editions. ASME text is paywalled, so these are Med confidence. Have the Level III confirm each one against the edition in your contract.

The table shows why an old procedure can't just be reused. A procedure written to the 2001 edition, with a 50 °F floor and 50 fc of white light, doesn't meet the current light requirement even though its temperature range is now stricter than needed.

The temperature range matters most on railway field work. Steel left outdoors on a winter morning can easily sit below 40 °F, and penetrant is too viscous to enter a tight crack when cold. Either warm the part into range and confirm with a contact thermometer, or use a procedure qualified for low temperature. Summer is the other extreme: a girder in direct sun can exceed 125 °F, and penetrant dries on the surface before it can enter the opening.

Dwell and developing times are minimums, not targets. Procedures often set a penetrant dwell of 10 minutes or more for weld work because fatigue cracks are tight. The interpretation window also has an upper end: after about an hour the indications bleed and spread so far that their size and shape are no longer reliable.

Key takeaway: Check temperature and light with instruments every shift; they are the variables most often skipped in the field.

Worked example: PT of a fillet weld, step by step

This walk-through shows a Type II, Method C liquid penetrant test of one fillet weld, with clock times. The procedure values are typical for an ASME- or E165-based field procedure; your procedure governs.

The job: a 3/8 in. fillet weld, 18 in. long, joining a stiffener plate to a beam web in a rail maintenance shop building. The contract calls for visual inspection plus PT of this weld to AWS D1.1, using a written procedure based on ASTM E165. The weld has been visually inspected and accepted for profile. The steel and the penetrant cans read 68 °F (20 °C), inside the 40–125 °F range.

Check the arithmetic. The developer is dry at 0:27. The window opens 10 minutes later at 0:37 and closes 60 minutes after drying, at 1:27. Evaluation finishes at 0:50, well inside the window, and the whole test of one weld takes about 55 minutes of clock time, most of it waiting. A technician working several welds would stagger them so that one weld's dwell overlaps another's evaluation.

The finding: at 0:40 a red line appears at the upper toe, about 1/4 in. long. The inspector wipes it lightly with a cleaner-dampened swab and re-develops it. The bleed-out comes back from the same thin line, which measures 3/16 in. long under a magnifier. It follows the toe, which is consistent with a toe crack, not a surface pore.

The disposition: under the D1.1 visual criteria a crack is unacceptable at any size, so the length doesn't decide this one (Med, CONFIRM in your edition). The weld is rejected and repaired: the crack is ground out, PT is repeated on the excavation to show the crack is gone, the groove is rewelded, and the repair gets visual inspection and PT again. If the same indication had been a rounded pore on ASME pressure work, the construction code would decide; ASME Section VIII Division 1 Mandatory Appendix 8, for example, treats relevant rounded indications over 3/16 in. as unacceptable (study-guide figure, CONFIRM).

What would change the result. If the steel had read 34 °F at step 1, the test couldn't go ahead under a standard-temperature procedure. If the developer had been applied in a heavy, wet coat, a fine toe crack could have been masked. If the remover had been sprayed straight onto the weld at step 5, the penetrant could have been flushed out of the crack and the weld wrongly passed. Each of these is a procedure variable, not a judgment call.

Key takeaway: PT of a single weld takes close to an hour, and most failures come from temperature, removal and timing, not from the dye.

How does PT compare with MT, UT and VT on railway parts?

Liquid penetrant testing is the slowest of the common surface methods on carbon steel but the only one of them that works on non-magnetic materials. MT is usually preferred on ferromagnetic welds, and UT is needed for anything below the surface.

Visual inspection always comes first, because PT can only examine a surface that is already acceptable for profile and cleanliness; see visual testing (VT). On carbon steel, MT finds surface and slightly subsurface cracks and needs no dwell time; see magnetic particle testing.

For a full side-by-side, read our comparison of non-destructive testing methods and how ultrasonic testing compares with other NDT methods. The short version: choose PT when the material can't be magnetized or the contract names it, and add UT when the question is what lies beneath the surface.

Key takeaway: VT first, MT on carbon steel, PT where MT can't go, and UT for depth.

What are the limits of liquid penetrant testing?

Liquid penetrant testing can't find anything that isn't open to the surface, and it gives no depth. Its results also depend heavily on cleaning, removal and timing.

• Surface-breaking flaws only. A subsurface crack, an internal void (as in the Casselton axle) or lack of fusion buried in the weld won't show. A fatigue crack still closed under paint or corrosion product won't show either.
• No depth information. PT shows length and location. A 3/16 in. toe crack could be 0.01 in. or 0.25 in. deep; only grinding or UT tells you.
• Coatings and contamination block the opening. Paint, mill scale, rust, grease and blasting debris have to come off first. Mechanical cleaning such as grinding or shot blasting can smear metal over a crack opening, especially on soft metals such as aluminum.
• Porous and rough surfaces. Porous material holds penetrant everywhere. Rough as-cast surfaces, such as a manganese frog casting or a weld with heavy ripples, leave background that can hide a fine indication.
• Temperature limits. Outside the procedure's range, results aren't valid without a qualified procedure, a real constraint on outdoor rail work in winter and summer.
• Over-removal and over-development. Too much remover pulls penetrant out of shallow flaws; too much developer masks them. Neither leaves any sign that it happened.
• Material compatibility. Sulfur and halogens in penetrant materials can harm nickel alloys, austenitic stainless and titanium, so the materials must be certified for low contaminants where the code requires it.
• Interpretation depends on the inspector. PT personnel are qualified by their employer under ASNT SNT-TC-1A or ANSI/ASNT CP-189; a Level II interprets, and acceptance is decided against the governing code.
• No acceptance criteria of its own. E165, E1417 and ASME V Article 6 say how to test. If the contract names no acceptance standard, the test has nothing to accept or reject against.

Key takeaway: A clean PT result means no open surface flaw at the time of test, nothing more.

How to specify PT in a purchase order or RFP

A PT specification should name the method standard, the acceptance standard, the Type and Method, and the extent of examination. Missing any of these is the most common reason a PT report can't be used.

1. Method standard and edition: for example "ASTM E165/E165M-23" or "ASME BPVC Section V, Article 6, 2025 edition".
2. Acceptance standard: the D1.1 or D1.2 clause, the ASME construction code, or the owner's specification. Don't leave it to the inspector.
3. Type, Method and sensitivity: Type II Method C for field work, or Type I with a stated sensitivity level for shop work.
4. Extent: which welds or areas, what percentage, and whether repairs get re-examined.
5. Personnel qualification: Level II for interpretation, under the contractor's written practice to SNT-TC-1A or CP-189.
6. Material certification: sulfur and halogen certification where stainless, nickel alloys or titanium are involved.
7. Report content: procedure number, materials and batch numbers, temperature, light reading, dwell and developing times, indications with location and length, and photographs.

How EMA Industries applies this

EMA Industries performs liquid penetrant testing as part of its AWS D1.1 structural weld inspection services, alongside UT and MT, and on railway components where MT can't be used, such as non-magnetic castings. Based in Southern California, EMA mobilizes nationwide and can also supply NDT Level III support to write or review PT procedures. To scope a job, see our ultrasonic and weld examination services or contact EMA Industries.

Frequently asked questions

What is the difference between Type I and Type II penetrant?

Type I penetrant is fluorescent and is viewed under UV-A light in a darkened area; it is the more sensitive family. Type II penetrant is a visible dye, usually red, viewed under white light against a white developer. Type II is the usual choice for field work because it needs no darkness or UV lamp.

How long should penetrant dwell time be?

The minimum comes from the governing standard and the written procedure, not a single universal figure. ASME Section V Article 6 sets minimums in Table T-672, and study guides commonly quote 5 to 10 minutes for steel welds at standard temperature. Many weld procedures set 10 minutes or more because fatigue cracks are tight.

Can liquid penetrant testing be used on manganese frogs?

PT is the practical surface method on cast austenitic manganese frogs because the steel is essentially non-magnetic and MT won't work. FRA research (RR 22-25, 2022) reports that in-service examination of these castings is visual only, so PT is mostly used to confirm that a crack has been ground out before repair. Check the owner's frog maintenance procedure for the requirement.

Does AWS D1.1 require liquid penetrant testing?

AWS D1.1 doesn't require PT on every weld. The contract documents decide whether MT or PT is done and on which welds, and D1.1 then gives the practice (ASTM E165 for PT) and the acceptance criteria. Cracks are unacceptable at any size under its visual criteria.

What temperature range is allowed for liquid penetrant testing?

Under current editions of ASME Section V Article 6, the standard range for the part and the penetrant is 40–125 °F (5–52 °C). Older editions used 50 °F as the lower limit. Testing outside the range needs a procedure qualified for that temperature.

Sources

• ASTM E165/E165M-23, Standard Practice for Liquid Penetrant Testing for General Industry, ASTM store page (Active), checked Oct 2026
• ASTM E1417/E1417M-21e1, Standard Practice for Liquid Penetrant Testing, ASTM store page (Active)
• ASTM E1208-21, E1209-18(2026), E1210-21, E1219-21, E1220-21 and E1418-21, method-specific PT practices, ASTM store pages (Active)
• ASTM E433-71(2023), Standard Reference Photographs for Liquid Penetrant Inspection, ASTM store page (Active)
• ASME BPVC Section V, Nondestructive Examination, ASME product page (2025 edition)
• AWS D1.1/D1.1M, Structural Welding Code – Steel, AWS
• AWS D1.2/D1.2M, Structural Welding Code – Aluminum, AWS
• ISO 3452-1:2021, Penetrant testing, Part 1: General principles, SIS catalog listing
• Wilson and Poudel, Feasibility of Special Trackwork Inspection Using Ultrasonics, FRA RR 22-25 (2022), ROSA P
• Poudel, Phased Array Ultrasonic Beam Modeling for Rail Thermite Welds, TTCI TD21-027 (2021), MxV Rail
• Banks, Magnetic Particle and Dye Penetrant Techniques as Aids to the NDT of Rails, British Journal of NDT 15(6), 1973, TRID abstract
• 49 CFR 215.105, Defective axle, eCFR (text read on Cornell LII)
• NTSB: Defective axle led to Casselton crude oil train derailment, Progressive Railroading, Feb 8, 2017
• ASTM E1417 key changes and updates, Kilty and Geis, Quality Magazine, Apr 2022; ASTM E1417 updates, Magnaflux (vendor)
• ASNT standards: SNT-TC-1A and ANSI/ASNT CP-189, ASNT
• ASME V Article 6 procedure values: open-exam-prep API 653 PT study guide and API 570 study guide (study guides, Med); 2001-edition values from a copy of ASME V 2001 Article 6