Continuous Rail Welding: Processes, Safety & Track Performance

On a 1440-foot string of rail, a 50°F temperature swing would grow the steel by about 5⅝ inches if it were free to move — but in continuous welded rail it can’t move, so roughly 130,000 pounds of force build up inside each rail instead. That invisible force is what makes continuous rail welding one of the most demanding jobs in metalworking, and why a single bad weld or a mismanaged neutral temperature can kink a mainline track in the middle of a heat wave.

What Is Continuous Rail Welding — and Why It Replaced the Clickety-Clack Era

Close your eyes and picture an old mainline: short sticks of rail, each about 39 feet long, bolted together with fishplates at every seam. With every wheel crossing, a gap would be struck to produce that familiar clickety-clack rhythm. Each of those impacts battered the rail ends, fatigued the joint bars, and slowly worked the bolts loose. For decades, this was considered normal, until railroads realized the joint itself was the problem.

The solution was to fuse those ends together with heat. This practice is known as continuous rail welding, more commonly called continuous welded rail (CWR), or in some places, welded rail or ribbon rail. The Federal Railroad Administration (FRA) defines CWR as rail welded end to end into rail strings with no rail joints, and sets its threshold at any length of rail welded together exceeding 400 feet.

A Permanent Connection

In practical terms, continuous rail welding permanently joins rail ends using one of several fusion methods. According to the FRA’s Track Inspector Rail Defect Reference Manual, the three main processes used in the United States are aluminothermic welding, flash butt welding and gas pressure welding, with enclosed-arc welding used for certain field repairs. While each method differs in equipment, metallurgy, and failure patterns, they all accomplish the same goal: to create one continuous piece of steel for trains to roll over.

Bolted joints were, by far, the weakest and most maintenance heavy part of track. Removing them removed most of that workload. The FRA points to four clear payoffs for CWR:

• Elimination of joints: The failure-prone mechanical connection is gone entirely.
• Increased rail life: Rail ends are no longer cut short by joint bar and impact damage.
• Reduced track deterioration: Without the gap impacts, the ballast and ties are subjected to far less shock loading.
• Reduced maintenance costs: Section crews can focus on track upkeep beyond constant joint repairs.

Trading Joints For Forces

Welding out every joint didn’t erase the forces acting on the rail, it removed its ability to relieve them. Steel expands when hot and contracts when cold at roughly 0.0000065 inches per inch per degree Fahrenheit. As noted in the FRA’s continuous welded rail training material, a free 1,440-foot string of rail would grow about 5⅝ inches for a 50°F temperature increase. With bolted joints, that movement could occur in small increments at each gap. With CWR, the rail is restrained by sleepers, ballast and anchors, so that expansion and contraction is converted into immense internal thermal stress.

This stress is dictated heavily by the rail neutral temperature (RNT), the temperature at which the welded rail is laid, anchored, and experiences little to no longitudinal force. In the heat of summer, that restraint can push the track laterally and cause a sun kink. In bitter cold, it can pull the rail to its limit, leading to tensile breaks. For this reason, establishing, measuring, and restoring the correct RNT is just as critical as producing a sound weld.

The Cost Of A Hidden Joint

The biggest trade-off is inspectability. A bolted joint can be checked with a glance and a torque wrench. A welded joint is a metallurgical event, one that can conceal porosity, inclusions, lack of fusion, or an undesirable heat-affected zone within solid steel. Because of this, CWR requires its own standards, ultrasonic testing routines, and acceptance criteria, most notably AWS D15.2/D15.2M, the recommended practice for joining rails by thermite and electric flash welding.

Ultimately, continuous rail welding traded thousands of mechanical joints for fewer, but far more complex, metallurgical and thermal challenges per mile. It’s a trade that allowed for smoother rides, safer operation, and drastically reduced maintenance, but only when both the weld itself and the forces acting on it are properly understood.

A Short History of Rail Joining: From Bolted Joints to Ribbon Rail

For more than a century, the sound of American rail travel was a sound of impacts. Every wheel that crossed a bolted rail joint delivered a thump, a thump, a thump — the “clickety-clack” that songwriters turned into a lullaby but that track engineers understood as damage happening in real time. A joint was never just a connection. It was a gap, a loose bolt, a battered rail end, and a little bar of steel doing fatigue work that never stopped.

Continuous rail welding grew out of one blunt realization: the joint was the weakest, highest-maintenance piece of track on the railroad. Understanding why railroads gave it up is the fastest way to understand what continuous rail welding actually buys — and what it costs in return.

Why Bolted Joints Were the Weak Link

A conventional joint paired two rail ends with a pair of joint bars and four to six bolts. Under traffic, the rail ends deflect independently, and the steel at the very end of each rail gets hammered flat. That rail end batter grows until the running surface is no longer smooth, and every passing wheel adds another blow. Meanwhile the joint bars themselves flex with each axle, cracks nucleate at bolt holes, and the bolts themselves loosen or fatigue. Track gangs spent a startling share of their lives tamping, tightening and replacing joints rather than doing anything about the rail itself.

The impact loads were not merely noisy. Each blow accelerated ballast degradation, loosened cribs, and pumped the track geometry out of line, which in turn created more impact. It was a self-reinforcing maintenance spiral, and it was expensive.

The DOT Study That Put Numbers on the Problem

The classic U.S. Department of Transportation study of rail joints is still the reference point for how bad the old system really was. It found that at the time of the survey, continuous welded rail covered only about 15 percent of U.S. track, while roughly 80 million conventionally bolted joints remained in service. Eighty million places where a wheel could land on a gap. The same study compared joint performance across welding methods and reported that through 1970 flash welds showed about 5.8 failures per 100 track miles while gas-pressure welds showed 22.9 per 100 miles — and that thermite welds failed roughly 85 times as frequently as flash welds on a per-weld-year basis. You can review the underlying report material through the DOT rail joint study record.

How the Transition Actually Happened

CWR installation in the United States began in the 1930s, initially combining oxy-acetylene, thermite, gas-pressure and electric flash welding depending on where the work was done and who was doing it. Early practice was as much art as procedure, and for decades welded rail and jointed rail coexisted on the same subdivision.

Two very different welding venues emerged:

Setting	What happens	Typical output
Plant (factory) welding	Rails are welded indoors under controlled conditions into long strings, then shipped by special rail cars	Strings of roughly 200–800 m, quality tightly monitored
Field welding	Welds are made on the roadbed to close the gaps between delivered strings, or to repair defects	Flash butt trains or portable thermite crews

Global practice followed the same logic. Nippon Steel reports that mill rails are typically produced in 25 m lengths and welded in plant into strings of 200–800 m, often 1,000 m or longer before they ever reach a track crew. A modern mainline therefore arrives at the jobsite in pieces measured in hundreds of meters, not in 39-foot sticks.

For field crews joining those strings, the practical techniques are covered in detail elsewhere on this site, including torch-based joining basics and the electrode selection knowledge that carries over to alloy steel welding electrodes.

What CWR Changed About Maintenance Economics

The FRA summarizes the payoff plainly: eliminating joints increases rail life, reduces track deterioration and lowers maintenance costs. Fewer joints means fewer impact loads, longer ballast life, smoother riding track and fewer slow orders. But the failure modes did not disappear — they moved. Instead of battering and joint bar fatigue, railroads now contend with weld quality variation, rail breaks away from joints, and thermal stress problems that jointed track never had to face. Welds and rail breaks, not joints, became the inspection priority, and the FRA’s rail manual documents that shift.

In short: continuous rail welding traded a known, visible, high-frequency problem (the joint) for a less frequent but far more consequential one (thermal stress and weld integrity). Everything that follows in this article is about managing that trade deliberately rather than by luck.

The Four Main Rail Welding Processes Explained

With continuous welded rail now the standard across most modern mainlines, the way rail ends are joined matters more than ever. While the goal is always the same – creating a metallurgically sound, continuous joint – not all rail welding methods are created equal. Across the world, four primary processes are used to achieve this, each with its own strengths, limitations, and ideal applications. As noted by Nippon Steel, the four worldwide rail welding methods are flash-butt, gas-pressure, thermite, and enclosed-arc welding.

1) Flash Butt Welding: The Forged Joint

Flash butt welding is widely considered the gold standard for rail joining, especially when quality and long-term reliability are the top priorities. Rather than melting filler material or casting steel into a gap, this is an automated forging process. The two rail ends are brought into close proximity, and an electrical current is passed through them to generate resistance heat. As the ends begin to soften, small contacts between them create a series of “flashes” – rapid arcing across the interface that burns away surface oxides and impurities.

Once sufficient flashing has occurred, the machine applies a high upset pressure, forging the rails together under heat. This solid-state-like union creates a joint with strength almost equal to the parent metal. According to the FRA Rail Manual, flash butt welding produces a narrower heat-affected zone than thermite welding and a more consistent hardness through the finished weld, normally resulting in fewer weld failures. These benefits are why flash butt welding is the dominant choice in fixed plant facilities, and increasingly common with mobile flash-butt trains working directly on the track.

2) Aluminothermic (Thermite) Welding: The Portable Casting Process

When portability and low capital investment are critical, aluminothermic welding – more commonly known as thermite welding – takes the lead. This process relies on an exothermic reaction between finely divided iron oxide and aluminum. Once ignited in a crucible, the reaction produces superheated molten steel and a layer of slag. The molten steel is then tapped into a single-use sand mold that has been clamped around the prepared rail gap, filling it by gravity casting.

While incredibly versatile for field work, thermite welding is a casting process rather than a forged one. As the FRA notes, this makes it more susceptible to casting-related defects such as porosity, inclusions, and lack of fusion. Proper preparation is non-negotiable, and even with strict adherence to procedure, these inherent metallurgical differences explain why thermite welds have historically seen higher failure rates than their flash butt counterparts. Still, for remote locations or emergency repairs, its ability to be deployed with minimal heavy equipment keeps it in widespread use.

3) Gas Pressure Welding: Precision Through Skill

Gas pressure welding takes a different approach – using a gas flame to heat the butted ends of the rails to a plastic (but not fully molten) state, then applying axial pressure to forge them together. Much like flash butt, this creates a forged joint rather than a cast one. However, the key difference is the heat source: instead of electrical resistance, it’s an open gas flame, making the process highly dependent on the operator’s skill and consistency.

This method has seen significant adoption in Japan, where tight quality control and experienced crews have allowed it to perform reliably. While not as prevalent in North America as flash butt or thermite, gas pressure welding remains a proven option in regions where its workflow fits local practices.

4) Enclosed-Arc Welding: The Field Repair Option

Enclosed-arc welding is the least common of the four for joining full rail strings, but it still has its place in the rail welding toolbox. Typically used for certain field repairs, this arc-based process relies on protected electrodes to deposit weld metal in a controlled manner. Its use is far more limited than the other methods, and it’s generally reserved for specific repair scenarios rather than mass production of continuous welded rail.

Comparing the Four Processes

With so many factors at play – from cost to metallurgy – it’s helpful to see how these methods stack up against one another. The table below breaks down their key characteristics based on industry practice:

Process	Type	Capital Intensity	Typical Application	Relative Quality/Reliability
Flash Butt Welding	Automated forging (resistance heat + upset)	High (especially mobile flash-butt trains)	Plant strings & on-track by mobile units	Highest – narrow HAZ, consistent hardness
Aluminothermic (Thermite)	Casting (exothermic reaction)	Low	Field welds & repairs (highly portable)	Good when done correctly, but prone to porosity/inclusions
Gas Pressure Welding	Forging (gas flame heat + pressure)	Moderate	Field/plant use, common in Japan	High if performed by skilled operators
Enclosed-Arc Welding	Arc deposition	Moderate-Low	Limited field repairs	Niche – not typical for full string welding

Regardless of which method is chosen, proper procedure and acceptance criteria are essential. In the United States, AWS D15.2/D15.2M serves as the governing recommended practice, covering minimum standards for welding rail, including both thermite welding and electric flash welding guidelines. It’s this combination of proven processes and strict standards that keeps continuous welded rail safe under the immense thermal forces it must withstand.

For crews looking to broaden their understanding of welding fundamentals that tie into these procedures, resources like All Position Welding Rod Guide: Types, Uses & Tips can help build a stronger foundation in how different processes and consumables behave in real-world conditions.

Step by Step: How a Field Thermite Weld Is Made on the Roadbed

A thermite weld on the open roadbed is essentially a miniature steel foundry built between two rail ends, usually inside a track window measured in hours. The chemistry is unforgiving and the sequence is fixed, which is why crews work from written procedures instead of memory. The walkthrough below follows the Goldschmidt Thermit operating manual and the recommended practice in AWS D15.2/D15.2M, the governing US document for joining rails by thermite and electric flash welding.

The Nine-Step Field Sequence

1. Cut the rail and set the gap. Rail ends are cut square with a rail saw or profiling torch, then spaced to roughly one inch — the standard “inch-gap” procedure. A straightedge checks horizontal and vertical alignment so the finished weld doesn’t sit proud or dipped.
2. Preheat the rail ends. A propane or air-acetylene torch heats the rail faces and the mold until the specified preheat is reached. This burns off moisture and lets the molten steel wet the rail faces cleanly; skipping or shortchanging preheat is the fastest route to a lack-of-fusion defect.
3. Assemble the mold and crucible. A single-use refractory sand mold is clamped around the gap with a runner and a riser. The crucible sits above it, charged with the Thermit portion and closed at the base by a fusible plug.
4. Ignite the Thermit portion. An ignition mixture starts the exothermic reaction between iron oxide and aluminum, which runs well above 2,000 °C (3,600 °F). The iron oxide is reduced to molten steel while the alumina by-product becomes a floating layer of slag. If you have never seen an aluminum-based reaction up close, the chemistry is closer to what our aluminum flux welding guide describes than to anything a MIG arc does — except far hotter and entirely self-sustaining.
5. Tap the plug and fill. The fusible plug melts and molten steel streams down the runner, across the gap and up into the riser, pushing air ahead of it and washing against the preheated rail ends. Slag trails behind and collects in the riser.
6. Cool and strip. After the specified hold time the weld solidifies, and the mold, runner box and crucible stand are broken away.
7. Shear the upset metal. Excess steel squeezed out along the head and rail sides is trimmed with a hydraulic shear while the weld is still hot. Left to cool, it turns into a long grinding job.
8. Rough and final grinding. Rough grinding removes the shear stub; final grinding restores the rail head profile and smooth running surface.
9. Ultrasonic testing. The weld is visually inspected, dimensionally checked and ultrasonically tested before acceptance. Anything showing porosity, inclusions or incomplete fusion is cut out and re-welded.

Field Conditions That Change the Procedure

Outdoors, weather writes part of the rulebook. The operating manual recommends thermite welding at rail temperatures of 40 °F and above, with additional special procedures required between 0 °F and 40 °F. Below that, moisture, thermal shock and slow solidification make the weld metallurgy unpredictable.

Field condition	Requirement
Rail temperature 40 °F or higher	Standard procedure
Rail temperature 0–40 °F	Special low-temperature procedures required
Wet Thermit welding powder	Never used — violent reaction risk and serious thermal burns
New weld near an existing Thermit weld	Minimum 14 ft spacing; 4 ft only if the alternative is a rail joint

Two rules deserve emphasis. Wet Thermit material is never used: when ignited it reacts violently, and the molten product causes deep thermal burns, as the manufacturer safety data and field manuals both warn. And a new Thermit weld must not sit closer than 14 feet to an existing Thermit weld, with a 4-foot minimum allowed only when the alternative would be installing a bolted rail joint. Closely spaced brittle welds concentrate hard, brittle sections in one short panel, and such clusters have contributed to derailments — a point reinforced in FRA track guidance on weld placement and quality.

Done right, a field thermite weld is nearly invisible: a smooth head profile, a weld that ultrasonics clear, and a joint that behaves like the rail on either side of it. Done wrong — cold rail, damp molds, tight spacing, no post-weld UT — it becomes the weakest point in the string, and the failure shows up later as a dipped weld, a winter break or a summer buckle.

Rail Neutral Temperature: The Hidden Physics Behind Continuous Welded Rail

Every length of continuous welded rail is a giant temperature gauge with no dial. Before a train ever crosses it, the rail holds a stored force set by one number: its rail neutral temperature (RNT). RNT is the stress-free temperature at which the rail carries no longitudinal thermal force at all. At 95°F neutral, a rail sitting at 95°F is relaxed. Ten degrees colder, it is in tension. Ten degrees warmer, it is in compression. Nothing looks different, but the internal load has changed by roughly 25,000 pounds.

That difference decides whether a track survives a heat wave or buckles in it, which is why the FRA’s rail manual treats neutral temperature management as a core safety requirement, not a construction detail.

The Math: How Heat Turns Into Force

Thermal movement in steel is linear and predictable. The FRA’s CWR training puts it in one line:

Rail movement (inches) = rail length (inches) × temperature difference (°F) × 0.0000065

The constant 0.0000065 is the coefficient of thermal expansion for steel — about six and a half millionths of an inch per inch per degree Fahrenheit. Run a 100-foot rail (3,936 inches) through a 50°F swing and it wants to grow by about 1.28 inches. That number is trivial on a fabrication table, where the metal is free to move. Worth comparing: aluminum expands at roughly double that rate, which is why heat distortion dominates conversations about aluminum MIG welding and any restrained aluminum joint.

In welded rail, the metal cannot move. Sleepers, ballast and welded ends hold it in place, so the would-be expansion becomes compressive force instead. A constrained 136 lb rail develops approximately 130,000 pounds of longitudinal force for a 50°F rise. Push the rail temperature to 140°F and compressive stress can exceed 12,000 psi, with loads above 150,000 pounds in a single rail, according to the FRA’s rail neutral temperature testing.

Rule of thumb inspectors actually use: every 1°F above neutral adds about 2,500 pounds of compressive force per rail. Twenty degrees above neutral is a 50,000-pound extra push that no one can see.

Rail Temperature Is Not Air Temperature

This is where field crews get caught out. Dark rail in direct sunlight can run about 30°F hotter than the ambient air, reaching 140–160°F depending on geography, ballast color, cloud cover and time of day. A pleasant 105°F afternoon can therefore be a 135°F rail event. Rail temperature must be measured on the rail itself — shaded web or base, per the railroad’s approved method — not read off a truck thermometer.

Why the Target Sits in the Middle of the Range

If heat is the enemy, why not set neutral temperature as high as possible? Because cold is the other enemy. A very high RNT leaves the rail in strong tension on winter nights, and tension breaks rail — those are the pull-aparts and joint failures that open gaps in January. A very low RNT leaves too much compression in August and the track sun-kinks. So RNT is deliberately targeted near the middle of the local seasonal rail temperature range, a compromise between buckling in heat and rail breaks in cold.

Older FRA testing placed the target between 75 and 95°F, while modern US installations are laid in the 90–110°F range, adjusted for local climate and the subdivision’s design buckling strength. The correct value is a documented desired RNT for each subdivision, not a crew’s judgment call on the day.

Rail Breaks Quietly Reset the Neutral Temperature

Here is the failure that surprises people. When rail breaks, or when a defective weld is cut out, the rail on each side relieves its stress by pulling or pushing toward a new equilibrium. The effective neutral temperature of the remaining string drops — sometimes dramatically — leaving the track far more buckling-prone than its records suggest. FRA training material shows how far RNT can fall after a break.

This is why regulation treats readjustment as mandatory rather than optional. Under 49 CFR 213.118 and 213.119, rail that has broken or been cut must be returned to the subdivision’s desired rail neutral temperature within specified time and temperature limits, following the FRA/RSAC generic CWR plan that every CWR track owner must maintain. Destressing by stretching or by the pull-apart method, followed by verification of the resulting neutral temperature, is the standard remedy.

Practically, that means any welding program is really two programs: joining rail, and proving the thermal state afterward. The weld itself is only as safe as the neutral temperature it was made at, and as the section on dimensional and distortion control in aluminum wheel repair reminds welders, geometry you cannot see is geometry you still have to manage. Specify the target RNT, measure it, adjust it, and record it — because the force inside the rail will not announce itself until the track moves.

Sun Kinks and Pull-Aparts: Failure Modes of Continuous Welded Rail

Continuous welded rail trades one set of problems for another. Welding rail into long strings eliminates the bolted joints that used to crack, loosen and pump under traffic, but it also removes the expansion gaps that once let the steel breathe. From that point on, every string is a constrained beam storing energy with each degree of temperature change, and it can fail in only two directions: it can bow sideways under heat, or tear itself apart under cold. Track forces call the first failure a sun kink and the second a pull-apart. Both are temperature-driven, both are predictable in principle, and both are unforgiving in practice.

Failure mode	Trigger	What the track does	First line of defense
Sun kink (buckle)	Heat; rail temperature above roughly 140–145°F on weak track	Sudden lateral shift, wavy or kinked alignment	Ballast and shoulder condition, speed restrictions, 49 CFR 213.119(g) inspection
Pull-apart	Cold snap; low or drifting rail neutral temperature	Gap opens or rail breaks at a weld or defect	Weld quality, ultrasonic testing, RNT verification and readjustment

Sun Kinks: When Heat Bows the Track

A sun kink is a lateral buckle. As rail temperature climbs, a constrained string generates longitudinal compression that the sleepers, fasteners and ballast shoulder are supposed to resist. When the compressive load exceeds the track’s lateral resistance, the rail shifts suddenly, sometimes several inches in a single movement. Trains rarely tolerate that well.

In FRA Safety Advisory 2012-03, the Federal Railroad Administration warned about a series of derailments preliminarily caused by rail buckling under extreme heat and reminded owners of the inspection requirements for buckling-prone CWR under 49 CFR 213.119(g). The advisory placed special emphasis on buckling near bridges, where a shifted track is compounded by the span beneath it.

The threshold to watch is roughly 140–145°F of rail temperature, at which point buckling risk rises sharply on track with weak lateral resistance. The earliest warning sign is usually not a full buckle at all: a wavy or kinky rail condition indicates micro-buckling, an incipient buckle condition in which the rail has started to move laterally but has not fully displaced. A crew seeing wave in tangent track during a heat wave is looking at track that is already in the process of failing.

Pull-Aparts: The Cold Mirror Image

Pull-apart work the other way. When temperature drops, the rail tries to contract and the track structure holds it in tension. If a defect exists — a thermite weld with a small inclusion, a fatigue crack, a rail head checking — the tension concentrates there until the rail opens a gap or breaks clean. Rail-break-caused derailments encompass over one-third of all track-failure-caused derailments, about 50 derailments per year on US mainline tracks, per FRA accident statistics cited in the agency’s CWR training material. Cold weather does not create bad welds; it finds them.

What Makes a Section Prone to Buckle

• Weak lateral resistance. Recent tamping loosens the ballast before it reconsolidates, and fouled or badly drained ballast never regains full resistance. Both leave the track least able to hold the rail in line exactly when heat loads build.
• Line and alignment defects. Twists, dips and poor geometry give the compressive force a place to initiate movement.
• Low or drifting RNT. Rail installed below its target neutral temperature, or a neutral temperature that has fallen after rail breaks were cut out, shifts the whole stress balance toward buckling.
• Dynamic vehicle loads. FRA/TSC research found that the dynamic buckling temperature for tangent track under long cars can be significantly lower than the static buckling temperature, meaning a train can trigger a buckle at a rail temperature the static numbers said was safe.

Mitigation: Speed Restrictions and Targeted Inspection

The most direct operational control is speed. Some CWR procedures require a temporary 40 mph restriction when ambient temperature changes by 40°F or more within a 12-hour period, reducing dynamic load while the track is settling into a new stress state. Inspection follows the same logic: buckling-prone locations identified under the road’s CWR plan get walked and measured preferentially during heat events, with priority on bridges, structures, recently tamped territory and sections with a history of movement.

The underlying point is that buckling and pull-aparts are not independent accidents. They are the two ends of the same neutral-temperature problem: manage the stress-free temperature and the track structure, and both failure modes stay rare. Ignore either, and the weather will eventually run the experiment for you.

Weld Quality, Inspection and Acceptance: Grinding, Geometry and Ultrasonics

Making the weld is only half the job. A rail weld can be metallurgically perfect and still fail in service if it sits half a millimetre proud of the running surface, because every wheel that passes over a high spot delivers an impact load the rail was never designed to take. So the second half of any continuous rail welding program is a disciplined routine of grinding, measuring and ultrasonic scanning, judged against written limits rather than the welder’s eye.

Finished Geometry: The Tolerances Welders Actually Work To

The most cited geometric limits come from the ARTC Weld Quality Management Manual, which specifies finished weld vertical alignment with a peak of +0.0 to +0.3 mm measured over a 1 m reference, with an absolute maximum peak of 0.5 mm anywhere along the weld. In plain language, a finished weld should sit flat to very slightly proud under a one-metre straightedge; anything reaching half a millimetre is already at the outer limit of acceptance. Horizontal alignment and profile are controlled the same way: a weld that is vertically fine but laterally kinked will still steer the wheelset and accelerate joint degradation.

Those numbers shape the grinding sequence more than most crews realise. ARTC guidance is explicit on two points:

• After the excess weld metal is cut off, rough grinding must not cut closer than 1 mm to the rail head surface. Leave stock; you can always take more off, but you cannot put it back.
• The earliest time for the final grind is one hour after the semi-finished grind. The weld and the surrounding rail are still contracting thermally. Grind to profile while the metal is still moving and the weld will sink into a dip as it cools, producing exactly the defect the tolerance was meant to prevent.

Quality check	ARTC acceptance limit	Typical cause of failure
Vertical alignment peak over 1 m	+0.0 to +0.3 mm (absolute max 0.5 mm)	Over-grinding or grinding too early
Stock left after rough grind	Minimum 1 mm to rail head	Over-removal at cut-off
Wait before final grind	At least 1 hour after semi-finished grind	Schedule pressure on the gang
Surface condition	No visible burns, gouges or undercut	Aggressive angle-grinder work

Dipped and peaked welds are the two defects ARTC describes in detail, and both are overwhelmingly caused by grinding error rather than by the welding process itself. A dipped weld comes from taking too much metal out of the centre of the band; a peaked weld comes from leaving a crown in the middle. Once in service, a dip grows quickly: wheel loads batter the soft, ground surface, the batter collects water and debris, and the defect propagates inward until it shows up as a transverse indication on an ultrasonic scan.

Ultrasonic Testing: The Real Acceptance Gate

Geometry is checked with gauges; soundness is checked with sound. The RDSO manual for flash butt welding of rails sets the pattern used in most modern programs: every single weld must pass visual inspection, dimensional tolerance checks and ultrasonic testing (USFD) before it is released for traffic, while sample welds undergo transverse bending tests and metallurgical examination to confirm the heat-affected zone and fusion line are sound.

The United States applies a quantitative bar. Under the FRA rail manual, a weld is classified as defective when it contains discontinuities or pockets exceeding 5 percent of the rail head area individually, or 10 percent in aggregate, arising from incomplete penetration, lack of fusion or related causes. A single large lack-of-fusion pocket can therefore fail a weld alone, while several smaller indications add up to the same verdict. Anything at or above that threshold means the weld is cut out and remade.

The Limits of Ultrasonics (and What Is Being Done About Them)

Ultrasonic testing is not infallible. Weld profiles that are still out of tolerance, rough or oxidised surface conditions, and rail head shelling all scatter or absorb the beam and can mask a genuine transverse defect. That is a practical argument for grinding to tolerance before scanning, not after: an unfinished weld is harder to inspect as well as harder to run over.

To close that gap, TTCI and the FRA have researched phased-array ultrasonic testing (PAUT) for better transverse defect characterization. Reported results showed rail head defects could be located to within about ±0.5 inch of a known datum, comfortably inside the ±1 inch accuracy required to plan a rail head repair weld. Guided-wave and automated scanning work continues alongside it, with the same goal of removing operator guesswork.

The takeaway: specify the geometry limits up front, insist on the waiting period before final grinding, and treat the ultrasonic scan as the release point, not an optional extra. Welds that skip any of those steps are the ones cut out later, at far greater cost than doing it right the first time.

Thermite vs Flash Butt Welds: Cost, Reliability and Real-World Performance

Ask a room of track engineers which weld belongs on a high-tonnage mainline and you will start an argument. Both processes deliver what continuous rail welding promises: an unbroken rail. The real question is which weld you pay for twice — once when it goes in, and again when it comes back out.

The Cost Equation: Cheap Welds Today, Expensive Welds Tomorrow

On paper, thermite looks like the bargain. The Transportation Research Board’s IDEA program, while evaluating electroslag field welding as a competing process, documented the North American market in its final report: more than 600,000 thermite welds are made in North America each year at an average cost of about $350 each, against roughly $500 for a flash butt weld.

The same report flags the catch. Between 2 and 15 percent of all new thermite welds contain unacceptable defects that require removal within the first year. Cutting out a bad weld is never a $350 operation. It means a night possession, rail cutting, a fresh mold and Thermit portion, more grinding, ultrasonic retesting, and usually a speed restriction until the geometry settles.

The gap starts in the physics. Thermite is an aluminum-powered exothermic reaction — the same aluminum-flux chemistry covered in our guide to aluminum flux welding, scaled up to yield a crucible of molten steel that solidifies in a single-use mold. It is casting. Flash butt welding is forging: electrical resistance heats the ends, flashing burns off irregularities, and massive upset pressure consolidates the plastic steel. The finished joint must still behave like the rail it joins — essentially an alloy steel electrode selection problem in miniature — but forged rather than allowed to freeze from liquid. That one difference drives most of the performance gap.

Reliability: What the Failure Data Says

Classic comparative data compiled by DOT put the gap in stark numbers: thermite welds failed about 85 times as frequently as flash welds on a per-weld-year basis, as reported in archived DOT rail research. Flash welds, made under controlled plant conditions, set the benchmark; thermite field welds trailed it badly.

Modern research reaches the same destination with better instrumentation. Loidolt et al., publishing in Engineering Failure Analysis in 2025, measured impact behavior at welds and found wheel battering roughly 25 times worse at thermite welds than at flash butt welds. Their maintenance findings were just as pointed: thermite irregularities reappear after grinding, while flash butt welds typically need only one correction.

Battering compounds. An uneven weld takes a heavier dynamic hit, the hit loosens ballast and degrades alignment, and degraded geometry amplifies the next hit. The weld that was cheaper on day one becomes the weld that keeps appearing on geometry exception reports.

Head-to-Head at a Glance

Factor	Thermite (Aluminothermic)	Flash Butt
Process type	Casting — molten steel solidifies in a mold	Forging — heated ends upset under pressure
Capital investment	Low; portable molds, crucibles, ignition kits	High; plant rigs or mobile welding trains
Average weld cost	About $350 (North American average)	About $500
First-year defect removals	2–15% of new welds	Markedly lower
Relative failure rate	About 85× flash welds per weld-year	Baseline
Wheel battering	About 25× worse (Loidolt et al., 2025)	Baseline; one correction usually suffices
Typical defects	Porosity, slag inclusions, lack of fusion	Few; parameter-driven and recorded
Portability	Excellent — anywhere a crew can stand	Requires a welding train or plant setup
Best fit	Remote territory, mid-rail closures, emergency repairs	Heavy-haul mainlines, factory-made strings

Why Heavy Haul Is Shifting to Mobile Flash Butt

Nippon Steel reports that freight railroads are increasingly shifting from thermite to mobile flash butt welding in pursuit of higher weld quality, and notes that on a 200-meter welded rail string, flash butt accounts for 87.5 percent of applications in typical practice.

Mobile flash butt trains erase thermite’s old advantage. A welding train travels the subdivision and makes forge-quality joints on the roadbed at production rates, with every parameter logged — current, flashing time, upset distance — so quality assurance is a record, not a guess. The trade-off is access: the train needs open track and works poorly in congested territory, tight urban windows, or washed-out remote branches. Thermite still goes where a welding train cannot.

Honest Guidance: Choosing the Right Weld

Neither process is being regulated out of existence. Both are still judged against the same acceptance bar — weld profile tolerances, dimensional checks and ultrasonic criteria — so the choice is economic, not legal. Program managers running mixed infrastructure usually land on a sensible split: flash butt wherever a welding train can reach and tonnage is heavy, thermite for the rest, with extra scrutiny on the thermite population through first-year ultrasonic campaigns and prompt removal of welds showing lack of fusion or heavy porosity.

The trend line is clear on heavy-haul networks, but thermite is not obsolete. It is the field surgeon of continuous welded rail: portable, fast, good enough when the alternative is a bolted joint — as long as you budget honestly for the welds that will not survive their first year.

Safety, Standards and Best Practices for Rail Welding Crews

Rail welds rarely fail because someone could not run a reaction or drive a flash-butt head. They fail because a crew skipped a pre-check, welded to an out-of-date procedure, or released track without confirming where the neutral temperature ended up. Continuous rail welding concentrates extreme heat, molten steel and stored thermal force into a few feet of track, which is why the safety picture is built from three layers: the written standards, the hazards at the joint itself, and the operating rules that govern the roadbed before, during and after the weld.

The Standards Landscape

Different countries keep different rulebooks, but the intent is the same: define who may weld, how the weld is proven, and what happens when it fails inspection.

• United States. AWS D15.2/D15.2M is the governing recommended practice for joining rails by thermite welding (TW) and flash welding (FW). It covers thermite welding safety precautions and sets minimum performance specifications for new welded rail, so a weld either clears the acceptance bar or gets cut out and remade.
• Europe. The EN 14587 series covers flash butt welding of rails. EN 14587-3 (2026) requires weld parameters to be monitored and recorded for all welds, with visual and geometry checks during production and bend testing at least weekly, plus defined alignment tolerance classes. It is a continuous-verification model rather than a spot-check one.
• India. Indian Railways route the work through RDSO manuals for flash butt welding and ultrasonic testing. The flash butt welding manual requires rails to be free of internal defects by ultrasonics and rejects rails carrying cracks, heavy corrosion pits, or more than 2 mm of wear at the rail seat.
• Australia. ARTC maintains its own weld quality management system and work instructions, governing who welds, what records travel with each weld, and how finished geometry is signed off.

Crews crossing between systems should read the performance specifications rather than memorize procedure names. The metallurgy behind those pass/fail limits is the same whether the code is American, European or Australian, and the fundamentals are covered in our guide to alloy steel welding electrodes: selection, use and safety.

Hazards at the Joint

The thermit reaction is not a fire, it is a portable foundry. It produces molten steel that causes serious thermal burns on contact, and smoke that may irritate the eyes, skin and respiratory tract. The welding powder itself is classified for shipping as a flammable solid under UN3178, which governs how it is stored, documented and transported to the site.

The controls that keep crews intact are unglamorous and non-negotiable:

• Full PPE — face shield with side protection, leathers, gauntlets and safety footwear, because spatter travels further than anyone expects.
• Materials stored dry; damp thermit material is a rejected condition, never a judgment call.
• An exclusion zone around the crucible, mold and tap area, covering both workers and anyone merely walking the right-of-way.
• Ventilation awareness in cuts, underpasses and tunnels where smoke collects.

Operational Safety: CWR Plans, Spacing and Destressing

Process safety alone does not make a section of track safe. Under 49 CFR 213.118 and 213.119, railroads must operate under an approved continuous welded rail plan covering installation, maintenance and restoration. In practice that means:

• Weld spacing rules are respected, including the spacing limits for thermit welds discussed earlier.
• Destressing (neutral-temperature adjustment) procedures are followed whenever rail is laid, cut or released.
• Post-weld rail neutral temperature is verified before normal speeds resume, not assumed from the procedure sheet.
• Any rail break, defect cut-out or weld replacement triggers an RNT reassessment of the surrounding string.

Best-Practice Checklist for Welding Crews

Most preventable weld problems are caught before the reaction is ever ignited. The habits below separate routine work from rework:

• Test before you cut. Ultrasonically test rails pre-weld; the RDSO approach of rejecting internal defects before welding is the right default everywhere.
• Reject bad rail. Rails showing severe head checking, crushed heads, wheel burns, kinked or out-of-tolerance welds must not be used.
• Eliminate old thermit welds where practical. Every surviving legacy weld is a future candidate for removal rather than a new weld placed beside it.
• Grind within tolerance. Finish profiles to the geometry limits in your governing standard, and give semi-finished welds time to cool before final grinding.
• Record everything. Weld parameters, rail temperature, operator, location and UT result should travel together in the weld record.
• Verify neutral temperature. Confirm RNT after any cut, weld or destressing operation before releasing the track.

Treat that list as the last thirty seconds before a crew signs off a weld, and the majority of quality escapes, buckling risks and premature removals disappear long before a train ever crosses the joint.

Common Mistakes, Troubleshooting and the Future of Continuous Rail Welding

Ask a track engineer where continuous rail welding goes wrong and you will rarely hear that the metallurgy failed. More often it is a process decision made days earlier: a weld laid in cold rail, a neutral temperature nobody re-checked, a ground surface that looked fine on the day and dipped six months later. The weld itself was sound. The system around it was not. This closing section collects the mistakes that cause the most damage, a practical troubleshooting guide, and the technologies likely to change how rail is welded over the next decade.

The Five Mistakes That Cost the Most

1. Welding cold rail without special procedures. When rail temperature sits well below the design range and no low-temperature procedure is applied, the weld locks in a low neutral temperature. FRA training data indicates that where joint welding is deferred, installation rail neutral temperatures can end up 20–30°F below the Desired Rail Neutral Temperature, eating directly into the buckling margin you were counting on for summer.
2. Skipping RNT readjustment after cutting out a defect. When rail breaks or a section is removed for defect repair, the rail’s neutral temperature is substantially reduced and must be readjusted before warm weather arrives, as the FRA’s track safety standards guidance and Kish CWR training Module 2 both emphasize. A repaired joint that is never destressed is a buckle waiting for July.
3. Final grinding performed too early. ARTC’s weld work instruction documents the trap clearly: a weld ground to profile immediately after semi-finished grinding can pass inspection on the spot yet fall out of tolerance later as the head cools and settles, becoming a dipped weld. Waiting at least an hour between stages is cheap insurance.
4. Poorly spaced thermit welds. Placing a new thermit weld within 14 feet of an existing one (the narrow 4-foot exception aside) stacks brittle casting-type welds in a short stretch of track, a configuration that has contributed to derailments.
5. Skipping post-weld ultrasonics. A weld that is never tested is an assumption, not a verified joint. Undetected oxide entrapment or lack of fusion grows into a transverse defect under traffic.

Troubleshooting: What You See, What It Means, What to Do

Defect	How you spot it	Typical remedial action
Dipped or peaked weld	3–10 m profile check; passenger or IM reports of a “thump”	Re-grind within tolerance if within limits; cut out and re-weld if the dip exceeds acceptance criteria
Kinked weld / angular misalignment	Straightedge or geometry car deviation at the weld	Correct by grinding only where metal allows; otherwise cut out and re-weld, with a speed restriction meanwhile
Slag entrainment or oxide entrapment	UT indications — irregular reflectors, often near the centerline or fusion line	Confirm with repositioned UT passes; if confirmed, cut out and re-weld
Porosity / lack of fusion	Clustered UT echoes exceeding FRA limits (5% of head area individually, 10% aggregate)	Remove the weld; apply speed restriction until replaced
Incipient buckle (wavy rail)	Visual: smooth lateral waves in the rail	Immediate speed restriction and destressing investigation — treat as a pre-failure

Where Continuous Rail Welding Is Heading

• Mobile flash butt trains displacing thermite on heavy haul. Railroads are steadily shifting field work to mobile flash butt units for tighter metallurgy and lower lifecycle cost, with flash butt already accounting for the large majority of applications on typical welded strings.
• Electroslag field welding as the middle path. TRB’s IDEA study targeted a field process positioned deliberately between thermite and flash butt: comparable to flash butt in quality, comparable to thermite in cost.
• Better inspection physics. FRA and TTCI research continues on phased-array and non-contact guided-wave ultrasonic systems designed to overcome the transverse-defect blind spots of conventional contact UT. Operator technique still matters too — our guide to all position welding rod work shows how access and body positioning shape weld and inspection quality in manual applications.
• Rail head repair welding. Cutting out a defective head section and welding in clean material disturbs the neutral temperature far less than removing a full plug — a meaningful advantage when RNT is already marginal.
• Smarter RNT management. Tools such as RNT-RESTORE give engineers a defensible way to recalculate neutral temperature after breaks and repairs rather than relying on memory and rule of thumb.

Conclusion: What to Verify Before Anyone Strikes an Arc

Continuous rail welding trades a thousand bolted joints for one long, thermally active structure. The joints are gone; their replacement obligations — neutral temperature control, weld spacing, grinding discipline and ultrasonic verification — remain for the life of the track. Before you specify or supervise a weld, run this checklist:

Done that way — with AWS D15.2 as the baseline practice and an up-to-date CWR plan under 49 CFR 213.118 and 213.119 — a continuous welded rail string delivers exactly what it promises: a smoother, longer-lived, cheaper-to-maintain track that simply has to be respected for the heat it carries.

Frequently Asked Questions

What is continuous rail welding?

Continuous rail welding is the process of permanently joining rail ends end-to-end to form long strings of rail without bolted joints, known as continuous welded rail (CWR). The FRA defines CWR as rail welded into strings exceeding 400 feet. Rails are joined by flash butt, aluminothermic (thermite), gas pressure or enclosed-arc welding, either in a fixed plant or on the track.

What are the main welding methods used for rail?

The main methods are flash butt welding (an automated electrical forging process considered superior for its narrow heat-affected zone), aluminothermic or thermite welding (an exothermic iron-oxide and aluminum reaction cast into molds, ideal for field work), gas pressure welding, and enclosed-arc welding. In the United States most rail welding uses thermite, flash butt and gas pressure welding.

Why does continuous welded rail buckle in hot weather?

CWR is constrained by fasteners and ballast, so temperature rises create compressive force instead of expansion. Rail temperature can exceed ambient air temperature by about 30°F and reach 140–160°F, generating compressive loads that can exceed 150,000 pounds per rail. If the track’s lateral resistance is weak, the rail shifts sideways suddenly — a ‘sun kink’ buckle — which FRA links to a series of heat-related derailments.

What is rail neutral temperature and why does it matter?

Rail neutral temperature (RNT) is the rail temperature at which the welded rail carries zero longitudinal thermal stress. Railroads install and anchor CWR near a target RNT, typically 90–110°F in the US, as a compromise between buckling in summer heat and rail breaks (pull-aparts) in winter cold. Losing RNT — through rail breaks, cuts or fastener creep — makes track buckling-prone and must be corrected per 49 CFR 213.118/119.

Are thermite welds or flash butt welds better?

Flash butt welds are metallurgically stronger because the process forges parent metal together, producing a narrower heat-affected zone and fewer failures; research found weld battering about 25 times worse on thermite welds. However, thermite welding is far cheaper and portable, costing about $350 versus $500 per weld, which is why it dominates field work even though 2–15 percent of new thermite welds may contain unacceptable first-year defects.

How are rail welds inspected for defects?

Every production weld passes visual inspection, dimensional/geometry checks against tolerances (for example ARTC limits of +0.0 to +0.3 mm peak over 1 m), and ultrasonic testing (USFD) for internal defects. FRA considers a weld defective if discontinuities exceed 5 percent of rail head area individually or 10 percent in aggregate. Phased-array ultrasonics are being adopted to better detect and size transverse defects.

How close can thermite welds be spaced in track?

Thermit welds should not be placed closer than 14 feet to an existing Thermit weld, because closely spaced brittle thermit welds have contributed to derailments. A limited exception allows placement no closer than 4 feet when the alternative would be installing a rail joint or cutting in extra rail. Existing thermit welds are eliminated where practical during rail replacement work.

What are the biggest safety hazards in rail welding?

Thermite welding involves molten steel that causes severe thermal burns, irritating smoke, and a flammable-solid shipping classification (UN3178), so crews use full PPE, dry stored materials and controlled exclusion zones. Electrical hazards, heavy rail handling, working within track curfews, and the structural risk of an unexpected track buckle during warm weather are the other major hazards addressed by AWS D15.2 and FRA regulations.

Why must rail be re-stressed after a weld or rail break?

Cutting rail for a defect or weld releases locked-in thermal force and substantially lowers the rail neutral temperature, sometimes leaving track effectively at a very low RNT. If that condition is not readjusted to the subdivision’s desired RNT before warm temperatures, the track can buckle. FRA/RSAC generic CWR procedures specify readjustment time and temperature tables for these situations.