Thermit Rail Welding: The Complete Guide to Aluminothermic Rail Joints

At 2 a.m., with the line closed and a 300-foot window before the first train, a mobile welding crew lights a crucible and 30 seconds later a one-kilo reaction burns at over 2,500°C, pouring molten steel into the gap between two rails. When the mold breaks away at dawn, the joint is one continuous piece of rail — no bolt, no bolt hole, no clickety-clack. This is thermit rail welding, and it quietly holds the world’s continuous welded rail network together.

What Is Thermit Rail Welding? A Straightforward Definition

Thermit rail welding is a method of joining two pieces of rail into a single continuous length using heat from a chemical reaction instead of electricity. A disposable crucible is mounted over the prepared rail gap, a measured portion of aluminum powder and iron oxide is ignited inside it, and roughly 30 seconds later the mixture produces molten iron that taps out and pours into a mold clamped around the rail ends. When the mold is stripped away and the surplus metal is trimmed, what remains is one continuous rail with no bolt, no bolt hole, and no fishplate. Put simply: it is a self-powered, field-portable pour that fuses track together wherever the line runs.

That definition answers the question most people arrive with, so let’s unpack it. The appeal of thermit welding is that it needs no mains power, no heavy plant, and no arc. The energy is already sitting in the powder. That single fact explains why crews light crucibles on remote mountain track, inside tunnels, and in three-hour night possessions far from the nearest substation.

The Many Names for the Same Process

The terminology confuses newcomers more than the chemistry does. The same process appears in standards, textbooks, and track diagrams as:

• Thermit rail welding or thermite welding — the railway spelling, from the trade name of the powder mixture.
• Aluminothermic welding (ATW) — the technical term preferred by UK and European standards, including the Network Rail standard for joining rails using the aluminothermic welding process.
• Exothermic welding — the broader category, which also covers cadweld-style joints for earthing conductors and cable terminations (see the overview of exothermic welding for the wider family).
• Goldschmidt welding — an older name that credits the inventor and still turns up in legacy documentation.

Where It Came From: The Goldschmidt Process

The chemistry traces back to one man. Hans Goldschmidt, a German chemist working in Berlin, patented the aluminothermic reduction process in Germany in the 1890s. His original goal was producing pure metals without carbon contamination, but the spectacular heat output quickly suggested a second use: welding without an external heat source at all. Aluminium’s appetite for oxygen is what drives it — the metal strips oxygen from iron oxide and releases the difference as heat.

Railways adopted the technique in the early 1900s, and the fundamental field procedure has barely changed since: align, preheat, mold, ignite, tap, trim, grind. Modern kits are simply more consistent, better specified, and far better documented.

Why Railroads Ditched Bolted Joints for Welded Rail

For more than a century, rails were laid in fixed lengths and stitched together with fishplates and bolts through holes drilled in the rail web. Every one of those holes is a stress raiser, every joint is a place where the rail ends batter and loosen, and every joint is a maintenance visit. Riders heard the result as the familiar clickety-clack; track crews felt it as endless bolt-tightening and joint replacement.

Continuous welded rail (CWR) removes the problem at its source. Fewer joints mean less maintenance, longer rail life, lower dynamic loads on the formation, and a smoother, quieter ride. It is now the default for mainline track worldwide, and US regulators publish a FRA continuous welded rail generic plan governing how that rail is installed, welded, and monitored. But continuous rail still has to be joined: at delivery ends, at switches, at repairs, and at every site where a cut has been made. That is precisely where thermit welding does its work.

The scale is easy to underestimate. Network Rail recorded roughly 30,000 aluminothermic welds for maintenance and 10,000–15,000 for renewals in the single year 2016/17, as reported in this overview of using the thermit welding technique on the rail infrastructure. In one network, in one year, that is tens of thousands of chemical pours keeping welded track continuous.

Where Thermit Welding Fits Among Rail-Joining Methods

Thermit is not the only way to join rail, and it is not always the first choice. It sits in a short list of established methods:

Method	Heat source	Power needed	Typical use
Thermit / aluminothermic	Chemical reaction	None	In-track maintenance, remote work, night possessions
Flash butt	Electric resistance + forging pressure	High, plant or mobile generator	New-build and heavy-haul mainline
Gas pressure	Oxy-fuel plus upset force	Gases only	Regional networks, some tram systems
Electroslag	Electric resistance in molten slag	High	Straightforward fixed-site joints

The pattern is obvious once you see it: the electric methods produce excellent welds but demand power and heavy equipment, while thermit trades a little metallurgical purity for total independence from the grid. An arc process such as aluminum MIG welding illustrates the opposite end of that spectrum — a machine, a supply, shielding gas, and a stable position. None of that exists at 2 a.m. on a single-track line in the middle of nowhere. That portability, not raw weld quality, is the reason thermit welding survived a century of competition and remains the everyday tool for keeping continuous welded rail in one piece.

How the Thermit Reaction Works: The Chemistry Explained

Strip away the molds, the crucible and the night shift, and thermit rail welding comes down to one elegant idea: a chemical reaction hot enough to manufacture its own steel. There is no arc, no mains power and no gas flame once it starts. A blended mixture of fine aluminum powder and iron oxide is ignited, and from that moment the reaction feeds itself until the charge is consumed. The aluminothermic family of exothermic welding processes all run on that same principle. Rail welding is simply the version that has to deliver an exact alloy and a rail-grade joint in roughly half a minute.

Aluminum as the Reducing Agent, Iron Oxide as the Oxidizer

The two main ingredients are deliberately ordinary. The oxidizer is iron oxide, usually mill scale, the flaky oxide layer collected from steel rolling mills and rich in Fe₂O₃. The reducer is aluminum, chosen because it has a far stronger affinity for oxygen than iron does. When the powders are sparked, aluminum tears oxygen away from the iron compound. What remains is elemental iron in liquid form and a new compound, aluminum oxide, as slag. In chemistry terms this is a classic redox reaction: aluminum is oxidized, iron is reduced.

That heat figure is the whole trick. The reaction needs no oxygen tank, transformer or trackside generator. A small ignition charge, historically a strip of magnesium ribbon, starts a local burn; the released heat raises neighbouring powder to its kindling point, which sustains the burn across the rest of the charge. Engineers call this a self-sustaining reaction: once lit, it runs to completion on its own. For a crew working a 2 a.m. possession in a tunnel or on a rural stretch with no power drop, that independence is not a chemical curiosity. It is the reason the process exists.

Superheat: Why Reaction Temperature and Melting Point Are Different Things

Here is the distinction that separates a successful weld from a cold, defective one. Rail steel begins to melt somewhere around 1,500°C, depending on its carbon and alloy content. The thermit reaction, by contrast, runs well above 2,500°C. The molten iron therefore arrives in the mold hundreds of degrees hotter than it needs to be simply to stay liquid. That excess, called superheat, does real work: it keeps the metal fluid long enough to fill every corner of the mold around the prepared rail ends, and it transfers enough heat into the rail sections themselves to fuse the boundary layer rather than merely butt against it. Without it, the poured steel would freeze on contact with cold rail and create the lack-of-fusion defects that ultrasonic testing later rejects. As the FRA’s technical review of thermite weld structure and properties notes, the joint that results is fundamentally a casting: one solid mass of steel that solidified in place between two rails.

Alloying Additions: Tuning the Poured Steel

Plain iron would be a poor match for modern rail, a carefully balanced high-carbon steel. So the thermit portion is not just aluminum and iron oxide. Suppliers blend in measured additions of carbon, manganese and silicon compounds: carbon sources such as anthracite, manganese as oxide or ferromanganese, and silica. As the reaction runs, these elements dissolve into the molten iron and steer its chemistry toward the target rail grade. Carbon builds the hardness and wear resistance the running surface needs, manganese adds strength and helps control sulfur, and silicon acts as a deoxidizer so the setting steel does not trap gas porosity in itself.

This is the same matching logic a welder applies when choosing alloy steel welding electrodes for a specific base metal: filler chemistry has to complement the material being joined, not fight it. The difference is that here the filler arrives as a powder and the adjustment happens inside a burning crucible rather than at the arc. Because of that, weld kits are supplied matched to both rail profile and rail grade. A kit formulated for standard carbon rail will not deliver the right chemistry on heat-treated or alloyed rail.

Slag, Tap-Off, and a Clean Single Use

Aluminum oxide, the reaction’s other product, is why the process is so tidy. It melts at roughly 2,050°C with a density near 4 g/cm³, while liquid steel sits closer to 7 g/cm³. The slag therefore floats. When the reaction finishes, the crucible’s tap hole is opened and the lighter alumina runs off first, or is held back, so only clean steel enters the mold. The separation is gravitational, not mechanical. Nothing needs cleaning up afterward: the spent crucible, sand mold and refractory linings are disposable and are stripped away with the joint, leaving no reusable equipment behind on site. What stays is a single monolithic section of rail where two ends used to be.

Step-by-Step: The Complete Aluminothermic Rail Welding Process

From a distance a thermit weld looks like one dramatic moment: a flash, a burst of sparks, molten metal pouring into a gap. In fact the pour lasts about half a minute in a job that takes a well-drilled crew most of an hour. Here is the field procedure, in the order a welding gang actually performs it.

Step 1: Cutting, Aligning, and Cleaning the Rail Ends

Work starts with the two rail ends about to become one. Each end is cut square with a rail saw or hydraulic cutter, because an angled end will not close evenly and gives a lopsided weld. The ends are pulled into alignment with a fixture or heavy-duty fishplates so the running surfaces and side faces match laterally and vertically within the procedure’s tolerances. Typical gaps sit around 25 mm (about one inch) for standard sections, though the exact figure comes from the procedure matched to the rail profile and grade, as set out in the UK’s standard for joining rails using the aluminothermic welding process.

Then comes the part crews rush: cleaning. Rust, mill scale, grease, paint and above all moisture must be stripped from both rail ends and the web and foot, grease burned off with a torch or solvent and scale brushed away. Any contaminant left behind is trapped in the solidifying steel and reappears as porosity or lack of fusion, the two defects that ruin thermit welds. If you have ever learned how to prep aluminum before welding, you know the principle: a weld is only as clean as the surface you pour onto.

Step 2: Preheating — the Make-or-Break Step

With the gap set and the ends clean, the crew preheats. An oxy-propane or oxy-acetylene torch is aimed into the gap until both rail ends reach the temperature specified in the procedure, checked with temperature-indicating crayons, a thermopaint stripe or an infrared pyrometer at the rail face.

Preheating does two jobs: it drives off moisture cleaning cannot reach, and it brings the cold rail up to a temperature where incoming molten steel fuses with it rather than freezing on contact. Cold ends give a cold-shut or lack-of-fusion defect; wet ends give a porous, gas-blistered weld, which is why no experienced welder shortens a preheat to save time in a possession. Our guide to air-acetylene torch basics covers the flame setup and heat control involved.

Step 3: Mold Assembly and Crucible Placement

Before the rail cools, the mold goes together around the gap. Most kits use two half-molds — pre-formed sand blocks or molds cast on site — clamped around the rail to form a cavity shaped like the finished weld, with a runner down to the joint and a riser above it. Every seam is packed shut with moulding sand, because molten steel will find any gap and run onto the ballast.

The crucible, holding the correct thermit portion for that rail section and steel grade, sits above the mold with its tapping outlet aligned over the runner. The gas equipment and consumables used here — cylinders, hoses, regulators, torches — are checked before the pour, not during it.

Step 4: Ignition, Reaction, and Tapping

Ignition comes from a starting rod or pellet. The aluminothermic charge lights and completes in roughly 20 to 30 seconds, a self-sustaining reaction needing no external power. As RSS Infrastructure’s summary of the process describes, the lighter slag floats on top while the molten steel taps automatically through the crucible bottom into the mold cavity.

Step 5: Retention, Mold Stripping, and Hot Shearing

The weld then sits undisturbed for the specified retention time, usually several minutes, while the metal solidifies and the joint gains strength; rushing this stage risks cracking the still-mushy metal. The molds and sand are broken away, and the excess metal — the upset or riser above the running surface — is sheared off with a hydraulic trimming shear while the weld is still hot: easier than cutting cold, and with less left for the grinder.

Step 6: Profile Grinding and Handover

The last step is grinding. A rail profile grinder blends the head, web and foot back to the original cross-section, checked against a profile template. The joint should be indistinguishable from the rail either side, with no dip, hump or undercut to hammer passing wheels. Once it passes the template check, the weld is logged and scheduled for ultrasonic testing before the line reopens.

Stage	What happens	Indicative timing
Cut and align	Square ends, gap set, alignment checked	Longest single stage
Clean	Degrease, brush, strip moisture and paint	Quick, never optional
Preheat	Oxy-fuel torch to specified rail-end temperature	Several minutes
Mold and crucible	Half-molds clamped and sand-sealed	A few minutes
Reaction and tap	Ignition, molten steel fills the cavity	20–30 seconds
Retention	Weld solidifies undisturbed	Several minutes, per procedure
Strip, shear, grind	Mold off, upset sheared hot, profile ground	Finishing stage

Each stage depends on the one before it: clean ends only matter if the preheat is right, and a perfect preheat is wasted if the mold leaks. That chain is why procedure compliance, not speed, is what a good welding gang is measured on.

Equipment, Materials, and Consumables Explained

One of the quiet advantages of thermit rail welding is how little infrastructure the job actually needs. There is no generator, no plant, no rail-mounted flash butt rig waiting on a flatcar. A welding gang can arrive in a pickup, unload, and be joining rail within an hour. But “simple” does not mean “basic” — the kit is highly specified, and almost every item in it is a single-use consumable matched to the exact rail being welded.

The Thermit Portion: The Heart of the Kit

The welding portion is the whole reaction in a sealed container: a pre-weighed charge of aluminum and iron oxide plus the alloying additions that determine the chemistry of the finished weld metal. Portions are supplied as sealed units in sizes that correspond to rail mass — a kit for 60 kg/m rail is physically different from one for 54 kg/m or 46 kg/m rail, and a kit for standard carbon rail is different again from one formulated for heat-treated or alloy grades.

This matching matters more than most newcomers expect. The alloying additions — carbon, manganese, silicon — are dosed to land the deposit in the right composition range for the parent rail. A portion specified for one grade can put the weld metal outside the specification for another, and the error will not be visible to the eye. As the UK standard for joining rails by the aluminothermic process makes clear, selecting the correct welding portion for the rail profile and steel grade is a documented procedural requirement, not a shop-floor judgement call. If you are buying consumables, our Alloy Steel Welding Electrodes: Selection, Use & Safety guide covers the same “match the filler to the parent metal” logic in more familiar stick-welding terms.

Crucibles, Molds, and Sealing Sand

The crucible holds the portion until ignition, then holds the superheated steel while the reaction runs and the tap hole releases the metal. Site crucibles are single-use, typically refractory-lined steel or bonded units, with a replaceable tap hole plug and a thermite starter cap. Because they are discarded after each weld, there is no cleaning, no residual-metal carryover, and no cross-contamination between grades — a large part of the process’s reputation for site cleanliness.

Molds come in two families: sand-bonded assemblies built around the rail ends, and pre-formed units clipped into place. Either way, two mold halves sit on either side of the gap, capped by a riser that feeds molten steel down into the joint cavity. They are sealed against the rail with molding sand or a kneadable luting compound so nothing escapes under the head of molten metal. The sealing is where leaks start: a poorly packed mold can spray steel into the ballast, ruin the weld, and end the possession.

Preheating, Alignment, and Finishing Gear

Preheating is the make-or-break step, and its equipment is correspondingly substantial. A typical rig combines an oxy-fuel or propane-oxygen torch with regulated gas cylinders sized for a full night’s welding — a single 47 kg LPG bottle and oxygen set will not always cover a multi-weld shift, so gangs plan cylinder changes between welds. If you are setting up a gas package for the first time, the Air Gas Welding Supply: Gases, Cylinders & Safety Guide explains regulator choice, cylinder handling, and flashback prevention in detail.

Alignment fixtures are equally important. A hydraulic or screw-type gap-setting frame grips both rail ends and controls gap, line, and level simultaneously, while gauges verify gap width and the required end offset. Hold the alignment wrong and you inherit a kink that grinding cannot fix. After the mold is stripped, a hydraulic trimming shear slices off the surplus upset metal while the weld is still hot — this must happen inside a narrow temperature window — before a profile grinder restores the running surface and side profile to the rail section. The finished head is checked against a template.

Expect roughly 30 minutes of preheat, 30 seconds of reaction, 4 to 5 minutes of retention, then trimming and grinding. As industry write-ups of the technique describe, the consumption of single-use items per weld is significant: crucible, mold, ignition device, sealing sand, and the portion itself. That is central to the economics — cheap per-weld consumables in exchange for almost no capital plant.

PPE and Personal Kit

Finally, the welder. Molten steel at 2,500°C radiates hard, so the standard issue is a leathers suit, aluminised face shield rated for radiant heat, gauntlets, and fire-resistant leggings and sleeves. Add a hard hat, safety boots, hearing protection for the grinding, and respiratory protection for fume and mold-binder smoke. Add a gas monitor for tunnel or enclosed work, and a fire blanket or extinguisher for ballast that catches the occasional spark. FRA technical material on thermit weld structure and properties treats the process as a controlled metallurgical procedure — and the PPE list is what keeps that procedure safe to run in a track pit at 3 a.m.

Advantages of Thermit Rail Welding

It is worth asking why thermit welding has survived for more than a century when flash butt welding makes metallurgically superior joints. The answer is not sentiment: aluminothermic welding works anywhere, with almost nothing, and still delivers continuous rail.

Portability: A Welding Plant That Fits in a Truck

The biggest advantage is that thermit welding needs no external electricity and no heavy plant: no generator set, no flash butt tractor, no power supply to negotiate. A small gang with a pickup carries everything: single-use crucibles, disposable molds, a welding portion matched to the rail section, a preheating torch and an alignment fixture.

Rail runs across remote branch lines, through cuttings, over bridges and inside tunnels where a mobile flash butt train is logistically impossible or uneconomic. Thermit welding scales down to the job rather than forcing the job to scale up to the plant. The preheating torch runs on propane and oxygen cylinders, stocked by any air gas welding supply outfit, so field re-supply is routine.

Speed Inside Tight Night Possessions

Rail work happens inside possessions: short overnight windows measured in hours, after the last train and before the first, and thermit welding fits that reality. A typical joint takes roughly 20 to 30 minutes of reaction, retention and hot trimming, followed by grinding to profile, and an experienced gang can finish several welds in a shift.

That turnaround keeps service disruption low. When a defect forces an emergency replacement at short notice, thermit welding is often the fastest way back to full speed. As industry practice descriptions note, it remains a standard technique for joining and replacing rail in live infrastructure because a compact crew can deploy it quickly.

A Low-Capital Entry Compared With Flash Butt

Capital cost is where the economics become decisive. A mobile flash butt tractor and its support train represent a very large investment that only pays off on high-volume work: new-build mainline, heavy haul corridors or mass re-rail programmes with kilometres of weld in a straight line. Occasional welds cannot justify that outlay.

A thermit outfit costs a fraction of that and travels on a single vehicle. Consumables are bought per weld, so cost tracks workload rather than capital. That low barrier is why small contractors and regional railways keep rail welding capability in-house, and why thermit remains the most common method for in-track joining during re-rail and defect replacement. It also appears in national CWR guidance such as the FRA’s generic continuous welded rail plan, which assumes joining work will routinely be done on the track itself.

The Monolithic Joint: Better Ride, Less Maintenance

A bolted fishplate joint is two rail ends held by four bolts through drilled holes. Those holes are stress raisers, the joint flexes under every axle, and the gap produces the familiar click-clack and its impact loads, which degrade the rail end and loosen bolts.

A thermit weld removes all of that: no bolt holes, no joint bars, no fasteners to tighten and no rail-end batter from a discontinuity in the running surface. The result is a monolithic joint that behaves like the rail it connects.

Aspect	Bolted fishplate joint	Thermit welded joint
Rail continuity	Interrupted at every joint	Continuous single rail
Bolt holes	Four per joint, crack starters	None
Impact loads	Heavy at each wheel passage	Eliminated
Routine maintenance	Bolting, alignment	Inspection only
Ride quality	Joint noise and vibration	Smooth running surface

Fewer joints also means fewer places for defects to develop, lowering whole-life maintenance cost even though each weld costs more than a bolted joint.

A Century-Long Track Record Worldwide

Aluminothermic welding grew out of the Goldschmidt process developed in Germany in the 1890s and has joined rail ever since. It has been used on every continent, on lines from urban transit to heavy haul, and its weld designs are codified in national standards. Few industrial processes survive a century of scrutiny; this one became the default answer for joining rail in place. Anyone who works with alloy steel welding electrodes knows the discipline: chemistry matched to the parent rail grade, so the joint is engineered rather than improvised.

Taken together, portability, possession-friendly speed, low capital cost, a joint without bolted-joint maintenance and a century of field evidence make thermit welding the default for most in-track joining. The choice is rarely about the better weld in a laboratory; it is about which weld you can actually make, at 2 a.m.

Limitations, Risks, and Common Weld Defects

Thermit rail welding earns its reputation honestly: it joins rail anywhere, at night, without mains power. But a thermit weld is a casting poured into a live track structure, and cast metal does not behave like the rolled rail it connects. Understand those limitations and you can plan around them; ignore them and you buy a dipped joint, a rejected weld, or a crack that UT finds six months later.

A Cast Structure Inside a Wrought Rail

Rail steel is continuously rolled and heat-treated into a fine, uniform pearlitic structure with a directional grain flow. The metal in a thermit weld never sees that processing. It solidifies in place inside the mold, so its structure is as-cast: coarse columnar grains that grow inward from the fusion boundary, dendritic solidification toward the centerline, chemical segregation of carbon and manganese, and non-metallic inclusions trapped as the metal freezes. Most of those inclusions are alumina — the oxide by-product of the reaction itself — plus slag and mold particles entrained if the pour is turbulent or the rail ends were not clean.

Nothing forges or rolls the joint, so nothing breaks up those dendrites or closes porosity. The result is lower ductility, a lower fatigue endurance limit, and a pronounced weakness at the weld centerline where the two solidification fronts meet. Above that sits a heat-affected zone in the rail ends, where preheat and pour have softened or coarsened the original rail microstructure, creating a hardness transition exactly where wheel loads concentrate. Filler and weld-metal chemistry are tuned to match the rail grade — the same alloy-matching logic you would apply when choosing Alloy Steel Welding Electrodes: Selection, Use & Safety — but matching chemistry cannot fully undo a cast grain structure.

Weld Battering and the Fatigue Gap

Battering is the visible symptom of that metallurgical difference. Under repeated heavy wheel loads, the slightly softer cast metal at the joint deforms plastically and flows, producing a local dip in the rail head. The dip then amplifies dynamic impact loads, which accelerates the deformation, and the joint slowly works itself into a dipped weld that riders feel and impact forces punish.

The numbers are uncomfortable. A statistical study drawing on more than 2,000 welds and 20 years of service data found weld battering rates roughly 25 times higher for thermit welds than for flash butt welds. The Federal Railroad Administration’s research into the structure and properties of thermit welds likewise reports higher failure rates than for electric flash butt welds, with correspondingly shorter expected service life. Independent work on improving rail weld reliability under heavy-haul conditions reaches the same conclusion: the cast joint is the weak link where axle loads are highest.

The Three Defects You Will See Most Often

Across inspection data and failure investigations, three defect types dominate, and they account for the bulk of thermit weld rejections:

Defect	What it looks like	Typical root cause
Porosity	Gas pores or shrinkage cavities in the weld head or web, found visually or as rounded indications on UT	Insufficient preheat, damp sand or mold paste, wet consumables, trapped gas from contaminated ends
Slag inclusions	Alumina or mold material trapped in the weld metal	Turbulent pouring, mold erosion, dirty rail ends, slag not fully floated out before solidification
Lack of fusion	A linear lack of bonding along the rail end face — the most damaging fatigue initiator of the three	Cold rail end, short preheat, wrong gap, preheat torch aimed incorrectly

Two further problems deserve mention. Transverse shrinkage cracks form across the rail head as the weld cools, because continuous welded rail restrains the joint and the contraction has nowhere to go; centerline hot tears are a cousin of the same mechanism. And misalignment or underfill — a step or low profile at the joint — creates an impact load that turns an otherwise sound weld into a battering candidate.

Preheat, Moisture, and Gap Control Decide the Outcome

Almost every serious defect traces back to one of three process variables:

• Preheating is the make-or-break step. Too short or too cool, and the molten steel hits a cold, possibly damp rail end and freezes instantly instead of bonding. Poor preheat is the single most common cause of both lack of fusion and porosity.
• Moisture and contamination are silent killers. Mold sand, sealing paste, and stored molds must be bone dry; rail ends must be free of grease, rust scale, paint, sealant, and cutting-torch oxide. The clean-metal discipline in Aluminum Cleaner for Welding: How to Prep Aluminum Right applies here in steel form: preparation quality sets the ceiling for weld quality.
• Gap and alignment matter more than most crews expect. Too wide and the weld fills with segregated, shrinkage-prone metal; too narrow and fusion suffers. Out-of-position rails force the mold into an overstressed geometry that leaks or cracks.

Mold sealing is equally unforgiving. Sand packed around a poorly seated mold lets molten steel run out, producing cold laps and incomplete fills — and, in the worst case, a fire in the ballast.

Weather, Workload, and the Human Factor

Thermit welding is unusually sensitive to workmanship because so much happens inside an opaque mold that cannot be reworked once tapped. Wind and rain strip heat from a preheated rail end within seconds, night temperatures change the required preheat time, and possession clocks push crews to compress steps the procedure says are fixed. Deviations in preheat time, gap, or mold sealing translate almost linearly into rejection rates, which is why procedure discipline, not equipment, separates a 1% rejection gang from a 15% one. Diagnostics rely on ultrasonic testing to catch what the eye cannot, and the detection of these internal flaws is well covered in UT examination of rail welds at the European Conference on Non-Destructive Testing.

The candid takeaway: a thermit weld is only as good as the crew, the conditions, and the discipline behind it. Where those are controlled, joints last decades. Where they are not, you get a cast structure with known metallurgical limits and a defect sitting invisibly inside it.

Thermit Welding vs Flash Butt Welding vs Other Rail Welding Methods

Thermit welding is not the only way to join rail, nor always the best one. It is one of four mainstream processes, each with different physics, a different cost profile, and a different place on the network. Comparing them lets an engineer specify the right weld for the right stretch of track.

Cast Joint vs Forged Joint: The Difference That Explains Almost Everything

The single most important distinction in rail welding is this: a thermit weld is a cast joint, and a flash butt weld is a forged one.

In a thermit weld, superheated steel from the aluminothermic reaction is poured into a mold around the rail gap and solidifies there. The result is a casting with a columnar, dendritic grain structure growing inward from the chill faces of the rail ends. It works: decades of service history, strength far beyond a bolted joint. But a casting never matches the fatigue performance of metal that is heated and mechanically upset, because forging breaks up coarse grain, closes porosity, and leaves a recrystallized fine-grained structure. That is what happens in flash butt welding: the rail ends are brought together, an arc flashes material away while heating them plastic, and they are then rammed together under pressure. There is no cast zone at all — the joint is pressure-bonded steel.

The US Federal Railroad Administration has documented the structure of thermit welds because their cast nature creates a softer, weaker heat-affected zone that behaves differently under load (FRA: Structure and Properties of Thermite Welds). Under heavy-haul tonnage, studies consistently report lower reliability for aluminothermic welds than for flash butt welds, which is why high-tonnage networks increasingly prioritize flash butt on mainline rail (ICRI: Improving the Reliability of Rail Welds Under Heavy Haul Conditions).

Head-to-Head Comparison of the Four Methods

Method	How it joins	Portability	Power demand	Relative weld quality	Typical use
Thermit (aluminothermic)	Cast: molten steel poured into a mold	Excellent; hand-carried kit	None; gas only for preheat	Good; cast structure limits fatigue life	In-track maintenance, remote lines, tunnels
Flash butt (stationary/mobile)	Forged: electric heating plus upsetting pressure	Plant excellent; mobile by machine size	High: hundreds of kVA	Highest; fine-grained forged structure	New CWR strings, heavy-haul and high-speed mainline
Gas-pressure	Forged: oxy-fuel heating plus upsetting	Good; gas cylinders required	Low (fuel gas only)	Good to very good	Selected networks, notably Japan and China
Electroslag	Cast/forged hybrid: remelted poured joint	Moderate	Moderate	Reported comparable to flash butt	Trialed alternative for in-track work

Why Flash Butt Welding Is the Quality Benchmark

Flash butt welding produces the highest-quality rail weld available. Because the interface metal never fully melts and is instead forged in a plastic state, the joint inherits the parent rail’s chemistry with no cast structure or coarse columnar zone. Bending strength, fatigue resistance, and impact toughness beat thermit equivalents, which is why plant-based flash butt has been the standard for factory-welding long CWR strings since the mid-20th century.

Mobile flash butt units have closed much of the old gap in convenience: self-contained machines travel to the worksite, align the rails, weld, and shear the upset in a single setup. The catch is logistics. They weigh tonnes, need a stable platform on the ballast, and draw substantial power, so they still need a generator set, flatcar, or road-rail vehicle. That is a reasonable price for a mainline possession with full plant access; it is not one for a siding in the mountains.

Where Thermit Welding Still Wins

Thermit welding has survived a century of competition for a simple reason: it is the only process two people can carry to the rail and execute with no external power. The consumables — crucible, portion, molds, sand — stay inert until ignition, with no generator, no cable run, and no flash-butt machine to position. That makes it the default for:

• In-track maintenance welding and break repairs during short possessions
• Tunnels, cuttings, bridges, and access-restricted locations
• Remote and rural networks where hauling heavy plant is uneconomic
• Emergency work after a rail fracture, where deployment speed matters more than metallurgy

The process is also covered by established network standards that treat it as a fully qualified method rather than a fallback, including the UK aluminothermic joining standard (SEMFW E214: Joining Rails Using the Aluminothermic Welding Process) and the FRA’s continuous welded rail framework (FRA CWR Generic Plan). Thermit also sits within the wider aluminum-driven industrial processes covered in our guide to aluminum flux welding, though rail work is far more demanding.

Gas-Pressure and Electroslag Welding: The Middle Ground

Gas-pressure welding heats the rail ends with oxy-fuel torches and then upsets them, giving a forged joint without flash butt’s power appetite. Japan built much of its early high-speed program on the method. Electroslag welding takes a different route: a consumable electrode remelts under a slag pool into the gap, producing a joint research describes as approaching flash butt quality at a cost nearer to thermit (IOP Materials Science and Technology; electroslag and exothermic process comparisons). Neither has displaced the big two, but both remain live options where networks hold the procedures and qualified operators.

Neither method is “better” in the abstract. The right weld is the one your network can qualify, your crew can execute consistently, and your maintenance budget can sustain for the next forty years.

Quality Control, Inspection, and Safety on Site

A thermit weld can be ground to a smooth profile, look identical to the parent rail, and still hide a slag pocket in the web. Aluminothermic welding casts sound metal inside a disposable mold, usually in the dark, in the rain, and on a clock, so quality control is never a single check. It is a layered system: competent people, a qualified procedure, correct consumables, visual inspection, ultrasonic testing, and safety discipline running alongside every step.

Visual and Profile Checks: The First Gate

Inspection starts when the mold comes off and the hot shear finishes its cut. The weld head is profiled to match the running surface of the parent rail, then checked physically with a template or rail profile gauge held across the joint. The inspector is looking for:

• Profile mismatch — underfill, overfill, or a head sitting proud or low relative to the adjacent rail, which creates the impact loading that destroys welds early.
• Surface-breaking defects — blowholes, porosity, and small shrinkage cracks on the crown, fillet, or web after grinding.
• Correct weld geometry — sound sidewall fusion lines and a consistent reinforcement profile.
• No residual mold material — sand or refractory fragments stuck in the throat or against the rail base.

Visual checks are fast, cheap, and mandatory, but they only see what is on the surface. Head finish tolerances are set network by network, and a weld that fails the profile gauge is re-profiled or rejected on the spot.

Ultrasonic Testing: Looking Inside the Weld

Ultrasonic testing (UT) is the standard non-destructive method for internal rail weld inspection, and it is where the real verdict is delivered. A transducer sends high-frequency sound into the rail, internal discontinuities reflect it back, and the technician reads the echoes. UT finds the flaws grinding cannot:

• Porosity and gas cavities left by casting and solidification of the weld metal.
• Slag inclusions — trapped reaction products and mold refractory in the fusion zone.
• Lack of fusion between weld metal and the rail end faces, particularly at the sidewalls or base.
• Transverse and longitudinal shrinkage cracks forming during cooling.

Rail welding research groups aluminothermic weld flaws into two broad families: deviations caused by human execution (poor preheating, wrong rail gap, badly sealed molds, contaminated consumables) and deviations inherent to the reaction and casting process itself (solidification porosity, shrinkage, inclusions). Studies use these categories to compare defect rates between processes, and the FRA has published its own data on the structural properties of thermit welds to support weld quality requirements on US track.

Timing matters too. Networks specify when testing happens relative to the weld and the reopening of the line, because readings change as the weld cools and as traffic works the joint. Some procedures require a first test before the possession closes, with a follow-up at a defined interval afterwards.

Procedural Qualification and Operator Certification

Nobody lights a crucible on a national rail network on instinct. In the UK, network standards require welders to hold recognised aluminothermic welding qualifications before working on the infrastructure, with the process defined in the national standard for joining rails using the aluminothermic welding process. Certification covers the welder, the procedure, and the specific rail grades and profiles being welded.

On the US side, the Federal Railroad Administration regulates track safety through 49 CFR Part 213, which governs continuous welded rail and weld quality requirements, while AWS D1.5, the Bridge Welding Code, is widely referenced for welding procedure and inspection practice. Whichever framework applies, the paperwork discipline is the same:

• Welder tickets and renewal dates recorded for every shift.
• Kit batch numbers logged and traceable to the rail grade and section welded.
• Preheat times, gap measurements, and weather conditions documented per weld.
• UT reports filed against the weld location.

If that feels familiar, it is the same procedural discipline that governs structural aluminium work, covered in our guide to aerospace welding standards and careers, and the same consumable control you need when storing and selecting alloy steel welding electrodes. Certificates are only worth the storage conditions and record-keeping behind them.

Safety: The Real Risk Profile of a Night Possession

The hazards are not subtle. A single reaction produces molten steel at well over 2,000°C, and the practical consequence is the one every crew learns quickly: sparks and molten metal can travel several metres from the mold. Around that sit four overlapping risks.

• Molten metal burns. The exclusion zone around the crucible is not decorative. Full leathers, face protection, and dry, non-flammable clothing are mandatory, and nobody stands at the mold’s tap end.
• Intense radiant heat. The preheat torch and the reaction throw serious infrared and convective heat, which is why preheating operators work in shifts and rotate position rather than holding one posture through the burn.
• Fume exposure. The reaction releases aluminium and iron oxide fumes plus vapours from rail coatings and contamination. Respiratory protection and ventilation matter, and in tunnels or cuttings fume concentration gets significantly worse. Cylinder handling adds its own rules.
• Working beside live traffic. Most thermit welding happens at night inside tight possessions next to open lines. Lookouts, protection arrangements, and a briefed evacuation plan are part of the job, and the crew’s escape route is planned before the crucible is positioned.

Fire risk closes the list: dry vegetation, wooden sleepers, oily ballast, and nearby signalling cables all need checking before ignition. The propane-oxygen preheat kit and its cylinders bring the same storage and handling discipline you would apply to any industrial gas setup, as set out in our guide to welding gas supplies and safety.

The honest summary is that thermit rail welding rewards discipline more than speed. Crews with the lowest rejection rates slow down for preheat, keep consumables dry and in date, measure the gap properly, and book the ultrasonic test before the possession ends — not after the line has been handed back to traffic.

Practical Tips, Common Mistakes, and What to Do Next

By now you understand how thermit rail welding works, what it costs, and where it falls short. What separates a weld that lasts thirty years from one that fails ultrasonic testing at dawn is rarely theory. It is discipline on the ground: the right kit, dry consumables, a controlled gap, and a crew that refuses to rush the preheat. Here is the whole guide compressed into field-level rules you can act on.

Match the Weld Kit to the Rail, Exactly

The most common avoidable failure in aluminothermic welding is chemistry mismatch. A thermit portion is not generic “hot metal.” It is a formulated alloy designed to produce weld metal matching the parent rail’s carbon, manganese, and silicon content, so hardness, tensile strength, and wear behavior line up with the rail you are joining. Using a kit specified for 90A/115RE rail on a heavier section, or a standard-carbon portion on premium heat-treated rail, gives you a joint that looks fine and underperforms silently. Confirm the portion code against the rail profile and steel grade before the crucible is ever set down. If your work touches alloy metallurgy more broadly, our guide to alloy steel welding electrodes explains the same selection logic in a different context.

Never Skip or Shorten the Preheat

If this article leaves you with one operational rule, make it this: preheat is not a formality, it is the process. The rail ends must be brought up to temperature with a propane-oxygen torch long enough to drive off moisture and ensure the molten steel does not freeze on contact with cold rail. Cold rail, rushed preheat, and damp conditions account for a disproportionate share of lack-of-fusion defects, porosity, and early failures. Treat the preheat timer as sacred and consumables as moisture-sensitive: sealed molds, unopened portions, and heated rail ends are the standard for a reason. Damp sand, damp rail ends, or a portion left in a humid van can cause porosity and, in bad cases, a violent steam release during the pour.

Control the Gap and the Alignment

Rail gap control is where geometry is won or lost. Too narrow and molten steel cannot flow properly into the mold cavity; too wide and you consume excessive weld metal and risk a sluggish fill with unstable chemistry. Typical gap values are printed on the portion’s instruction sheet for a reason — respect them rather than “eyeballing” what looks right. Alignment matters just as much: vertical and lateral offset between the two rail ends creates a stress riser and a grinding nightmare, and it can leave a dipped or lifted joint that hammers under traffic. Use alignment fixtures and gauges every single time, and check alignment again after the mold is sealed but before the crucible goes on, because sealing pressure can nudge the rail.

Plan for Thermal Expansion in Continuous Welded Rail

Thermit welds almost always end up as part of continuous welded rail, and CWR behaves like a giant expansion joint that cannot move. As temperature rises, the rail develops enormous compressive force, and a badly welded or badly stressed panel can buckle. The FRA’s generic CWR plan sets out the framework rails use: plan welds around a target stressing temperature, calculate the pull length needed for the prevailing rail temperature, and stress or cut the rail accordingly before it is anchored. Practically, that means checking rail temperature with a calibrated probe before welding, not guessing from the air temperature at 2 a.m.

Schedule the UT, Not Just the Weld

A weld is not finished when the grinder goes down. Ultrasonic testing is the only practical way to see inside the joint for porosity, slag inclusion, or lack of fusion, and it should be booked into the same plan as the welding itself rather than bolted on as an afterthought. Book the tester, the track access, and the acceptance criteria (often a percentage of the rail web area, per the applicable standard) into the possession plan up front. This matters commercially as much as technically: even a small increase in rejection rate can erase thermit’s cost advantage over flash butt on high-volume work, because each rejected weld means a re-cut, a re-weld, and another possession. The ICRI research on rail weld reliability under heavy haul conditions makes the same point from the maintenance side: weld integrity, not weld price, drives whole-life cost.

Choose Thermit or Flash Butt on the Right Criteria

Decide the process before the gang mobilizes. On high-speed lines and heavy-haul mainlines, best practice increasingly favors flash butt welding for building new strings, because its forged joint offers better fatigue resistance and a more consistent, machine-controlled result. Thermit welding retains its place where mobility, portability, and access win: in-track maintenance, remote terrain, tunnels, defect replacement, and any job where dragging a mobile flash butt plant into position is impractical. If you want the wider context behind these joining choices, our comparison of aluminum flux welding processes covers how flux and exothermic chemistry behave in related applications.

What to Do Next

1. Identify the exact rail profile and steel grade on site, and order matching thermit portions.
2. Write the preheat, gap, and alignment steps into a job sheet, with timers and tolerances, and audit them.
3. Confirm consumable storage is dry and sealed, with stock rotated.
4. Check rail temperature and plan stressing around the FRA CWR framework or your network’s equivalent.
5. Book ultrasonic testing before the possession, not after the failure.
6. Train and refresh the crew. Thermit quality lives or dies on whether the operator follows the sequence under time pressure.

Conclusion: Answering the Core Question

So what is thermit rail welding? It is an aluminothermic, self-sustaining redox reaction that produces superheated molten steel, poured into a molded gap between two rail ends to create a continuous, monolithic joint. It works because it needs no mains power, deploys quickly in a night possession, and joins rail in places no flash butt plant will ever reach. It carries real liabilities: a cast rather than forged microstructure, lower fatigue performance, greater sensitivity to workmanship, and defects like porosity and shrinkage cracking when preheat or moisture control slips. Used where it fits, guided by standards such as SEMFW E214 in the UK and AWS D1.5 in the US, and inspected properly, thermit welding remains an essential tool in the permanent way engineer’s kit. For more on surface and consumable preparation discipline that carries over into any precision welding work, see how to prep metal properly before welding.

Frequently Asked Questions

What is thermit rail welding?

Thermit rail welding, also called aluminothermic welding, joins two rail ends by igniting a mixture of aluminum powder and iron oxide in a crucible above the rail gap. The exothermic reaction produces molten steel at over 2,500°C, which is tapped into a sand mold around the rail ends and solidifies into a continuous joint.

How hot does a thermit weld get?

The aluminothermic reaction exceeds 2,500°C, well above the roughly 1,500°C melting point of rail steel. This superheat ensures the molten iron stays fully liquid long enough to fuse properly with the preheated rail ends before solidifying.

How long does one thermit weld take?

The reaction itself lasts only around 20–30 seconds, but a complete weld including rail preparation, preheating, mold assembly, pouring, cooling, hot trimming, and grinding typically occupies a significant part of an overnight possession window. Crews commonly complete several welds per shift.

Are thermit welds as strong as flash butt welds?

Generally not. Flash butt welds are forged pressure welds with a homogeneous structure close to rail steel, and research consistently shows lower failure rates and longer service life than thermit welds. Thermit welds are cast joints, so they are more prone to defects and to weld battering under heavy axle loads.

What are the most common thermit weld defects?

Porosity, slag inclusions, and lack of fusion are the three most frequent defects. They are usually caused by inadequate preheating, damp consumables or mold surfaces, contaminated rail ends, or an incorrect rail gap, and are typically detected by ultrasonic testing after welding.

Why is preheating so important in thermit rail welding?

Preheating drives moisture out of the rail ends and mold, brings the rail steel to the temperature needed for proper fusion, and prevents the molten steel from chilling too quickly. Skipping or shortening preheating is one of the leading causes of lack of fusion and porosity.

How are thermit rail welds inspected?

Completed welds are visually checked for profile and then inspected with ultrasonic testing, which detects internal porosity, slag inclusions, and lack of fusion. Welders and inspectors must hold recognized qualifications, and procedures are governed by standards such as AWS D1.5 and network-specific specifications.

Where is thermit welding still used today?

Thermit welding remains widely used for in-track work — re-rail projects, defect replacement, and joining rail in remote areas, tunnels, and bridges where flash butt plant or mains power is impractical. Many networks prefer flash butt welding for new mainline and heavy-haul CWR strings.

Is thermit welding dangerous?

It involves molten metal above 2,500°C, intense radiant heat, sparks, and fumes, so it demands full PPE, exclusion zones, fire precautions, and trained crews. Combined with night working beside live railway lines, strict safety procedures and certification are mandatory.