The problem of stress fatigue that can arise when two pieces of metal work against each other under load was known to the first engineers. But it took a serious rail accident in the UK, at Hatfield in 2000, to show that good management is the best preventive. Eric Russell reports.

When British Rail was split into innumerable small companies a decade ago, it dissipated a large amount of railway engineering knowledge away from the industry. One lesson that was lost was how best to cope with the specific type of stress fatigue known as rail gauge corner cracking.
This occurs on the inside corner of the rail and Professor Lewis Lesley, professor of transport science at Liverpool John Moores University, says is caused by wheels sliding laterally across the rail head.
Gauge corner cracking is a specific case of rail contact fatigue (RCF), which can also occur on the rail top, or head. A fundamental division can be made between defects whose life begins at the rail surface and those whose life begins at so-called inclusions under the surface. Gauge corner cracking is a form of surface-initiated defect.
RCF exists on a railway because stresses are enormous: a typical passenger train is carried on an area little larger than a single compact disc, while locomotive traction is now controlled so tightly that driving wheels are close to the point of slip for much of the time.
High loads give rise to particularly high stresses if the transverse profile of the rail does not properly conform to the wheels that run over it.
The transverse profile is critical, and particularly so if hard rail is used. This does not readily wear or deform plastically to accommodate non-conforming wheels.

Crack initiation

Cracks are initiated at the rail surface as a result of the combination of high normal loading and a high ratio of traction to normal load, which give rise to pronounced shear of the surface layers. The direction of the crack on the rail surface shows the direction of traction that gave rise to it.
For squats, which are associated with tangent track and gentle curves, the crack mouth lies approximately across the rail. The significant traction is therefore longitudinal.
Typical head checks, which occur on the high rail in curves on all types of railway are usually oriented at an angle to the running direction. A combination of lateral and longitudinal traction has been significant.
In a few cases, lateral traction has been the principal influence, and the crack mouth lies parallel to the axis of the rail.
For the type of gauge corner cracking which exists on high speed, high cant deficiency curves in Britain, the crack mouth lies almost across the rail. In this respect and in many others, this type of defect is essentially identical to a squat.
Once a crack has developed a little at the rail surface, further growth occurs mainly as a result of a mechanism of fluid entrapment, primarily water but also grease lubricants. The crack enters the rail in the direction of travel, with the surface layers sheared backwards by locomotive traction.
Once fluid enters the crack, enormous stresses are caused at the crack tip as the wheel rolls onto and seals the crack mouth. By this means, the crack develops eventually to a length of the order of 10mm and a depth of a few millimetres into the rail.

Crack direction

When the crack grows out of the relatively thin, heavily plastically deformed layer at the rail surface, it tends to change direction, to grow almost at right angles to the axis of the rail. Sometimes it may grow up to the rail surface or meet up with other cracks and spall out, causing little further damage.
Sometimes, the crack turns down into the rail. Whether the crack turns up or down is influenced by the local stresses in the rail and is still a subject of research. The crack then grows under the influence of bulk stresses: residual stresses from rolling the rail, thermal stresses and bending.
If the crack is not detected and removed, it grows until the rail can no longer carry the hogging moment either ahead of the wheels or caused by the impact from wheelflats. It then breaks.
It may be surmised that because bulk stresses are primarily longitudinal, those defects for which the crack mouth lies across the rail, are more critical than those for which the crack mouth is at angle to the rail.
RCF has become more prevalent not only because ever-greater loads are borne by the rails but also because wear has been greatly reduced, primarily as a result of more effective lubrication. A high rate of wear would remove the cracked material before the crack could progress into the rail.
It is known that manufacturing is not the cause of gauge corner cracking, some cracks are found in rail that is less than a year old and it has also occurred on straight as well as curved track.

Detection

Ultrasonic detection is currently the technology of choice for crack investigation. It is an established technology that is well understood, is cost effective and can function at track speeds with an accuracy of 5m.
Electromagnetic and guided wave radar systems are more expensive and offer too fine a resolution. They are also slower to produce results and these can be more difficult to interpret. The fine resolution can mean that impurities in the rail show up as cracking.
Residual stresses in the rail crown are known to have an important influence on the initiation and evolution of gauge corner cracking. In the past there has been no method available to monitor localised stress levels out on the track. This can be achieved through magnetic anisotropy, which analyses the interior of the rail.
The use of low-frequency Raleigh surface waves can detect gauge corner cracking in railway lines. The cracks absorb certain frequencies depending on the crack geometry and can be detected by comparing the transmitted and received waveforms.
Bombardier Transportation and TSC Inspection Systems are developing Alternating Current Field Measurement, an advanced non-destructive testing technique for detecting and sizing surface-breaking defects in metals. It can do this through dirt and coatings. The company says the technique has the potential for large cost savings by streamlining the entire inspection, repair and overhaul process.

Remedy

Grinding down the rail is the remedy for gauge corner cracking. It is a well established operation, but the lesson that has been learnt is one of timing. Little and often is the answer rather than leaving treatment for a long time, followed by a major grinding session.
The trick is to strike a balance between wear rate and crack growth so the crack geometry never achieves a potentially disastrous depth. So grinding is used selectively to increase artificially the wear rate of the rail.
Surface cracks need to be removed when they are a fraction of a millimetre deep, before they propagate into the rail. So grinding becomes a tool for preventive rail maintenance, not a means for correcting damage once it has occurred.
Besides reducing the occurrence of broken rails, a balanced programme of grinding increases rail life by many times.
On other European railways, grinding has been undertaken at a very much higher level than in Britain for most of the last 10 years. This has been to treat corrugation, which often occurs on curves. It now appears that this has alleviated RCF problems without the railway realising it.
To study the problem in the UK requires different companies to sit down together. Once British Rail was split into individual companies, the responsibility for tracks and rolling stock was also separated. Now gauge corner cracking has been highlighted as a growing danger, a body has had to be set up to bring the wheels and rails factions into the same camp.

Special organisation

This body is called the Wheel Rail Interface System Authority (WRISA), a cross-industry body to provide a forum for the discussion of cross-boundary issues affecting both wheel and rail.
Besides its own activities, it is part of the TRAINS project, which will investigate the way in which rails and wheels affect each other.
This project is headed by the University of Birmingham and Manchester Metropolitan University and includes leading industrial partners, organisations and academics.

The overall feeling from people in the industry is that rails should be brought into the field of condition monitoring. This technique was first applied to rotating machinery as a way to predict when failure might occur by analysing the trend of regular measurements of, principally, vibration. This enabled repairs to be carried out at a time that did not disrupt production.
Today, it is also used to determine when maintenance is due. Instead of stripping down machinery and replacing parts at a pre-set time, when such work may not be necessary, condition monitoring suggests when maintenance is needed and provides cost savings.
The principle also applies to static situations such as rails, where condition monitoring could predict when grinding is required. But it could well be that current research merely underlines what trained and experienced engineers would have done anyway.