Rail welding today

Most of the uses Viall and Hart described for thermit have passed to other processes. The one that has not is joining rails in the field. Rails are long, heavy and fixed in place, often in remote country, the work happens on the track itself, often far from any power supply, and a weld must be made in the gap between trains. A process that brings its own heat in a crucible, needs no electricity and joins the whole rail section in one pour still suits that job, alongside flash butt welding in depots and mobile plants.

This chapter looks at why rails are welded at all, the forces a continuous welded rail carries, how rail welds are qualified and controlled in Australia, and how the quarry audited the welds on its siding after the failures described earlier.

From jointed track to continuous rail

Rails were traditionally joined with bolted fishplates, leaving a small gap at every joint. Every wheel strikes every joint, which batters the rail ends, loosens the bolts, damages the wheels and needs constant maintenance. Welding the rails into continuous lengths removes the joints.

Viall wrote in 1921 that oxy-acetylene and electric welding had not proved practical for welding trolley rails, the street tramway rails of the day, and that the so-called welded joints made by those methods consisted merely of plates welded to the rails, so the joint was never really eliminated. Thermit, by contrast, had proved efficient and economical, and thousands of thermit-welded rail joints had been in service for years around the world. TWI notes that exothermic rail welding has been used in the United States since at least 1935.

Hart recorded a cautionary example from the same era. In Los Angeles, several hundred miles of track with cast-welded joints, made by pouring molten iron around the rail ends, were being replaced as unsatisfactory, because barely one joint in ten had actually fused with the rail. The poured metal had simply cast around the rail ends without melting into them. The lesson, which Chapters 2 and 6 put into numbers, is that a weld is only a weld if the faces melt.

Thermit and flash butt welding

Rails are joined by two main processes. Flash butt welding, described in Resistance Welding by the Numbers, forges the rail ends together with no melted metal left in the joint; it is done in fixed welding depots, where rails are joined into long strings, and by mobile flash butt welders on track. Aluminothermic welding is used where a mobile flash butt welder is not available or practical: for joining the long strings in track, for repairs and replacements, at turnouts and crossings, and on lines where access is difficult.

TWI describes the modern rail process: the rail ends are preheated until orange so that the steel is not chilled as it flows into the gap; the thermit mixture includes high-carbon alloying metal that melts into the weld; the reaction is started with a spark and a booster or a hot flame; and the process creates a lot of smoke and needs protective equipment.

Figure 1Flash butt and aluminothermic rail welds

	Flash butt	Aluminothermic
Heat source	electrical resistance and flashing	chemical reaction in a crucible
Where done	depot or mobile welding plant	in the field, with portable equipment
Joint	forged, no cast metal left in the joint	cast weld metal with a collar
Material consumed	rail length lost to flashing and upset	thermit portion, mould, crucible
Main quality risks	flat spots, upset control	lack of fusion, slag, porosity, wrong portion

A rail weld in outline

The steps of an aluminothermic rail weld follow the fusion method that Viall described in 1921, in modern form. In outline, and not as a procedure:

1. The rail ends are cut square, cleaned, and aligned in line and level, with a gap set for the portion and mould being used and, commonly, a slight rise across the gap to allow for the weld settling as it shrinks.
2. A pre-formed mould, made for the rail section, is clamped around the gap and sealed to the rail.
3. The rail ends and mould are preheated with gas burners to the temperature and for the time the procedure sets, which TWI describes as until the rail ends are orange.
4. The crucible, loaded with the portion matched to the rail grade, is placed over the mould, and the portion is ignited with the specified igniter.
5. When the reaction has finished, the crucible taps, often automatically, and the iron fills the mould from the bottom while the slag stays above or runs to a slag pan.
6. After a set time, the mould is stripped, the excess metal over the head is trimmed while hot, and when the weld has cooled the head is ground to the rail profile.
7. The weld is inspected and recorded.

Every step has a counterpart in the earlier chapters: the gap and shrinkage in Chapter 7, the mould and riser in Chapter 5, the preheat in Chapter 6, the portion in Chapters 4 and 8, and the timing and superheat in Chapter 2.

Why rail welds are watched closely

Even a sound aluminothermic weld differs from the rail around it. Its weld metal has a cast structure rather than the rolled structure of the rail, and its collar creates a change of section at the rail foot, where bending stresses from passing wheels are high. Rails bend under every wheel, so every weld in a busy line sees millions of load cycles, as well as the thermal force on cold nights. Any defect in the weld, especially at the foot, is a potential fatigue crack.

Rail infrastructure managers therefore inspect welds periodically, often by ultrasonic testing from the rail head, alongside their inspection of the rail itself, and they record and analyse weld failures to find patterns. A weld that fails is a source of information about the process, the crews and the materials, as the quarry’s audit shows on a small scale.

Repairing the railhead

AS 1085.20’s scope also covers repairing railheads, by arc or aluminothermic welding. Railheads are damaged by wheel burns, where slipping wheels melt and harden the surface, by battered ends at old joints and by surface defects that grow in service. Repairs build the head back up with weld metal of suitable hardness and grind it to profile. These repairs are also qualified procedures, matched to the rail grade, because a repair much softer or harder than the rail causes the same problems as a mismatched weld (Chapter 8).

The forces in a continuous rail

A long welded rail fastened to sleepers cannot change length when its temperature changes, so the change becomes stress. The fully restrained thermal stress is about E × α × ΔT, where ΔT is the difference between the rail’s temperature and its stress-free temperature, the temperature at which it carries no thermal force.

This is an illustration. With E about 200 GPa and α about 11.5 × 10⁻⁶ per °C, each degree of difference produces about 2.3 MPa. A 50 kg/m rail has a cross-section of about 50 ÷ 7,850 m² ≈ 6,370 mm². On a cold night 40 °C below its stress-free temperature, the rail carries a tensile stress of about 200,000 × 11.5 × 10⁻⁶ × 40 ≈ 92 MPa and a tensile force of about 92 × 6,370 ≈ 586 kN, roughly the weight of sixty tonnes, pulling on every weld in the string. On a hot day above the stress-free temperature, the force is compressive and the track can buckle sideways if it is not held.

Figure 2Thermal force in a fully restrained 50 kg/m rail

F = E × α × ΔT × A

E

modulus of elasticity, about 200 GPa

α

coefficient of expansion, about 11.5 × 10⁻⁶ per °C

ΔT

rail temperature below stress-free temperature, 40 °C

A

rail cross-section, about 6,370 mm²

1. stress ≈ 200,000 × 11.5 × 10⁻⁶ × 40 ≈ 92 MPa
2. force ≈ 92 × 6,370 ≈ 586,000 N ≈ 586 kN

That force is why rail welds fail mostly in cold weather, and why a fusion defect that has survived summer can grow and break in winter. It is also why rail infrastructure managers set rules for the rail temperature at which rails are welded and fastened, use rail tensors to stretch rails to the correct length when the rail temperature is below the target stress-free temperature, and record the stress-free temperature for each length of track. Those rules, and the shrinkage allowance at each weld (Chapter 7), are part of the infrastructure manager’s standards, not something the welder decides.

How rail welds are qualified in Australia

In Australia, welding of steel rail is covered by AS 1085.20, Railway track material, Part 20: Welding of steel rail, published through the Rail Industry Safety and Standards Board. Its scope includes the qualification of flash butt and aluminothermic welds in steel rail and the repair of railheads. Rail infrastructure managers adopt it and add their own requirements for procedures, welder accreditation, inspection, records and the management of rail stress.

The practical effect is that a rail weld is not just a weld. It is made by an accredited welder, with an approved portion and mould matched to the rail section and grade, to a qualified procedure, under the infrastructure manager’s rules for rail temperature and stress, and it is inspected and recorded so that its history can be traced.

Inspecting a rail weld

A completed rail weld is checked in several ways:

• Visual inspection of the collar, the ground running surface and the rail foot, for cracks, porosity, fins and incomplete filling.
• Geometry, with a straightedge over the weld, because a weld that is high or low, or out of line, causes impact loads; welders often set the rail ends with a slight rise across the gap, so that the weld finishes level after its shrinkage.
• Ultrasonic testing, where the infrastructure manager requires it, to look for lack of fusion, shrinkage cavities and inclusions inside the weld.
• Records of the welder, procedure, portion batch, rail grade, rail temperature, preheat and timing.

The quarry’s siding audit

This is an illustration. After the failed weld in Chapter 2 and the dipped weld in Chapter 8, the quarry engineer audited the siding. It had 46 aluminothermic welds. The audit found:

Figure 3Findings of the quarry's siding weld audit

• Avoid: Welds counted and located46, marked on a track plan
• Avoid: Records found38 of 46 had complete weld records
• Avoid: Visual and straightedge checkall 46; one more weld with a slight dip found
• Avoid: Ultrasonic testingall 46, by the rail contractor; no further fusion defects found
• Do: ? Missing records8 welds; replaced or monitored at closer intervals
• Do: ? Contract requirementsrail grade, portion, rail temperature and stress-free temperature now required in records

Two defective welds out of 46, about 4%, both traced to process errors recorded at the time, told the engineer that the problem was control, not the process. A rate like that is not a statistic to extrapolate from a sample of 46; it is a prompt to look at each weld, which is why the audit tested every one of them rather than a sample. The eight welds with no records were the real gap: without a record, nobody could say which portion, procedure or rail temperature had been used, so their risk could only be judged by inspection. The quarry also confirmed with its rail safety adviser how the siding sat within the rail safety framework, since that determined who had to be accredited and what standards applied; whatever the answer, the quarry adopted the infrastructure standards its rail contractor worked to.

A failure: the winter break

This is an illustration. The weld that failed in Chapter 2 broke on a cold morning in July. The rail contractor’s report estimated that the rail was about 30 °C below its recorded stress-free temperature at the time, which corresponds to a tensile stress of about 200,000 × 11.5 × 10⁻⁶ × 30 ≈ 69 MPa and a force of about 69 × 6,370 ≈ 440 kN in the 50 kg/m rail. The unfused band at the weld face, there since the weld was made, had slowly grown into a fatigue crack under traffic, and on the cold morning the remaining sound section could not carry the thermal force plus the load of a passing train.

The break was found by the quarry’s train crew, who noticed the gap and stopped short. A clean break in a continuous rail opens a gap, because the thermal force pulls the two broken ends apart, and that gap is both the hazard and the warning. The rail was secured, the failed weld was cut out, a plug rail was installed with two new welds, and the rail stress was restored to the recorded stress-free temperature under the contractor’s procedure. The investigation’s conclusions fed the audit above.

Exercise: a heavier rail

The quarry considers relaying part of the siding with 60 kg/m rail. Calculate its approximate cross-section from its mass per metre, and the thermal stress and force it would carry 35 °C below its stress-free temperature, using the same E and α.

The cross-section is about 60 ÷ 7,850 m² ≈ 7,640 mm². The stress is about 200,000 × 11.5 × 10⁻⁶ × 35 ≈ 80.5 MPa, and the force about 80.5 × 7,640 ≈ 615 kN.

The heavier rail carries the same thermal stress for the same temperature difference, but a larger force, because its section is larger. Its welds must be made with portions and moulds designed for the 60 kg/m section, and its stress-free temperature set and recorded under the infrastructure manager’s rules when it is laid.