Flash Butt Welding: Joining Steel Rails With a Spark Shower

Clamp, flash, upset, trim: the four-beat cycle

A flash butt welder is essentially a giant resistance welder built around two heavy clamps. Each rail end is gripped in water-cooled copper electrode jaws that also serve as the current contacts; one clamp is fixed, the other rides on a hydraulic platen that advances and retracts with millimetre precision. A welding transformer steps mains power down to a secondary of only ≈5–10 V but tens of kiloamps, wired straight across the two jaws so the whole circuit closes through the rail-to-rail interface.

The cycle has four beats. Clamp: the rails are aligned and gripped. Flash: current is switched on and the moving rail creeps forward; each fleeting contact between the rough faces bursts into sparks, burning the ends back and raising them to a uniform near-molten heat over 30–180 s. Upset: the platen suddenly reverses into a fast, high-force thrust that forges the pasty ends together in a fraction of a second. Trim: while still red-hot and plastic, the squeezed-out collar of expelled metal is sheared off flush by a hydraulic knife, and the weld is later ground back to the rail profile.

Why the sparks fly: constriction resistance and exploding bridges

The spark shower is not decoration — it is the heating mechanism. When two steel faces touch, they meet only at a few microscopic high spots (asperities), not across the full ≈8,600 mm² section. All the current is forced through those tiny bridges. The resistance of a single circular contact of radius a is Holm's constriction resistance, R ≈ ρ ⁄ (2a), where ρ is resistivity. For steel (ρ ≈ 1.4×10⁻⁷ Ω·m) and a 50 µm spot, R ≈ 1.4 mΩ — small in isolation, but the current crammed through it is not.

Local current density J = I ⁄ A at such a spot reaches ~10⁵ A/mm², and the Joule power P = I²R dumped into a micro-volume is thousands of watts. The bridge cannot conduct that heat away fast enough, so it melts and then vaporizes in microseconds, exploding and flinging a molten droplet clear — one spark. The platen keeps advancing, new asperities touch, and the process repeats thousands of times a second. This flashing is self-regulating: it burns the faces flat, drives off surface oxide and mill scale, and leaves both ends coated in a thin, clean film of their own molten metal at forging temperature.

The upset: a forge weld hiding inside a spark shower

Here is the counter-intuitive part, and the most common misconception about FBW: the finished weld is not made of the melted metal you see sparking. That molten film is contaminated with oxides and must be got rid of. The bond is a forge weld — a solid-state joining of clean, plasticized parent metal.

That is the job of the upset. At the decisive instant the flashing current can be boosted and the platen is driven forward hard, consuming a further 5–8 mm of length in well under a second at an upset pressure of ≈50–90 MPa. On an 8,600 mm² rail head-web-foot section that is a forging force of F = pA ≈ 60 MPa × 8.6×10⁻³ m² ≈ 520 kN — the largest machines reach 700–1,000 kN. Speed is everything: the thrust squirts the oxide-laden molten skin out of the joint as flash before it can freeze, and presses fresh, atomically clean austenite faces together so grains recrystallize across the interface. Too slow, too cold, or too little upset and oxide is trapped as "flat spots" — the classic FBW defect and a ready-made fatigue-crack starter.

Running the numbers: volts, kiloamps and megajoules

Put numbers on the heat. To bring a 20 mm-deep slice of a 136 lb/yd rail (A ≈ 8,600 mm², so V ≈ 1.7×10⁻⁴ m³, mass ≈ 1.35 kg) from 20 °C to a forging temperature near 1,300 °C takes Q = mcΔT ≈ 1.35 × 500 × 1,280 ≈ 0.86 MJ of sensible heat alone. Add the wider heated zone, the latent heat and mass of the flashed-away metal, and the electrical losses, and a single rail weld consumes on the order of a few megajoules, delivered by a ~150–300 kVA transformer over a minute or two.

The consumables ledger: flashing burns off ~15–25 mm and the upset adds ~5–8 mm, so each weld eats ~25–35 mm — every joint makes the rail about an inch shorter, which planners must budget across the hundreds of welds in a long string. Total cycle time, clamp to trim, is roughly 1–4 minutes.

Metallurgy, defects and inspection

Rail is a near-eutectoid pearlitic steel — roughly 0.6–0.8 % C with ~1 % Mn — chosen for wear resistance, which is exactly what makes it awkward to weld. The narrow heat-affected zone (HAZ) either side of an FBW joint is austenitized and then quenched by the cold rail on either side; cool it too fast and it can form hard, brittle martensite. FBW's advantage over cast processes is a narrow, symmetric HAZ with no foreign filler — but the weld and its HAZ are still the metallurgical weak link.

So finished welds are controlled tightly: forced-air normalizing or programmed cooling to tame hardness, then hardness traverses, straightness checks (peak/dip typically within ≈0.3 mm under a 1 m straightedge), ultrasonic and sometimes magnetic-particle inspection, and slow-bend acceptance tests. Standards such as AREMA in North America and EN 14587 in Europe specify the flashing/upset schedule, the allowable defects (flat spots, penetrators, incomplete fusion, grey spots) and the test regime. Broken welds are a leading cause of rail defects, so this scrutiny is not bureaucratic — it is derailment prevention.

From welding plant to open track: CWR and thermal stress

Electric resistance butt welding dates to Elihu Thomson's work in the 1880s; flash welding matured in the early 20th century and became the backbone of continuous welded rail (CWR) from the 1950s–60s. Today stationary welding plants flash-weld standard 25 m rails into strings of ~100–400 m, which travel to site on special trains; self-propelled mobile welders (Schlatter, Plasser & Theurer's APT machines, Progress Rail) then make the field closure welds in the track itself, carrying their own transformer, hydraulics and shear.

Eliminating the bolted fishplate joint removed a notorious source of impact, wear and fatigue — the rhythmic "clickety-clack" — but it created a new problem FBW forces engineers to confront: a jointless rail cannot expand. A fully restrained rail develops thermal stress σ = EαΔT; with E ≈ 210 GPa and α ≈ 12×10⁻⁶ °C⁻¹, a 30 °C swing gives σ ≈ 76 MPa, or ~650 kN of axial force in the section. That is why CWR is installed at a rail neutral (stress-free) temperature — pull it too far and it can fracture in a winter cold snap; let it heat past neutral and it can buckle into a "sun kink." Flash butt welding is what makes CWR — and the compromise it demands — possible.

Flash butt welding vs. aluminothermic (thermite) welding for joining rails

Attribute	Flash Butt Welding	Thermite (Aluminothermic) Welding
Bond type	Solid-state forge weld (molten metal expelled)	Cast fusion weld (molten steel poured in)
Filler / heat source	No filler; I²R resistance (Joule) heating	Cast filler from Fe₂O₃ + Al reaction, ~2,000–2,500 °C
Heat-affected zone	Narrow, symmetric (~20–40 mm)	Wide (~50–100 mm), less uniform
Time per joint	~1–4 min	~30–45 min incl. preheat & cooling
Portability	Heavy machine + high power needed	Portable crucible & mould; minimal power
Weld quality / fatigue life	High, consistent; fewer defects	Lower; porosity & inclusions more likely
Typical use	Plant strings & mobile in-track welding	Field closures where a welder can't reach