Return to Archive -by date - by topic - 2000 Archive.
Call it ‘head checking' or ‘gauge corner cracking' – something baffling is happening at the wheel/rail interface
When I give a talk on railways in general, I invariably take along eight 5p coins. When these are arranged on a table in a Bo-Bo formation I tell the audience that this represents the really clever part of a Class 91 locomotive. Well, it's different, and I happen to believe it is true.
When a steel wheel under a load of five tonnes or more sits on a steel rail, the two surfaces deflect at the point of contact. The resulting ‘contact patch' is about the area of a 5 p coin, though not the same shape.
This tiny area where steel meets steel provides four key functions. It supports the vehicle, keeps it running straight, steers it round curves, in the case of the class 91 each patch transmits about 800hp to drive the train forward and brakes the wheel at up to 12% of the wheel load. All at speeds of up to 300mile/h and more.
When you consider the hundreds of millions of abortive expenditure over the last quarter of a century trying to achieve the same results with computers and electro-magnets you realise that our little contact patch is something special. Musing on this thought, that pioneer of railway dynamics Alan Wickens once said to me ‘nature is rarely so kind'.
My old physics teacher said that God was obviously a physicist because He ensured that water was densest at 4 degrees above freezing. As a result as a pond cooled, the ice formed from the surface down, ensuring life survived at the bottom.
Father Ford the Railway Padre hypothesises that at some time during creation the thought occurred that man would eventually need a fast and reliable form of surface transport and the laws of physics and the metallurgy of steel were adjusted to make the contact patch possible.
Of course, other than Railtrack's Group Standards and Her Majesty's Railway Inspectorate, railway engineers have free will and over the years more and more has been demanded of the contact patch. Axle loads have increased, speeds have risen while bogie dynamics have improved greatly.
So, while the contact patch has to work much harder, it enjoys what ought to be more stable conditions. In particularly bogies track true: but for high speeds this means stiffer suspension in yaw which could increase contact patch forces on high speed curves .
Contributing to the ability of the contact patch to meet these demands have been improvements in rail steels, both the material and its production.
Continuous casting, where the steel is tapped from the furnace rather than being poured into moulds, minimises inclusions such as slag or hydrogen in the metal. This means that there is a greatly reduced risk of an imperfection in the rail rolled from con-cast steel. And the rail itself, it has become more versatile.
Engineering invariably involves trade offs. Steel rails need two contrasting characteristics.
Obviously the steel must not be brittle to minimise the risk of breakage under multi-tonne axles loads. But equally, the steel has to be hard enough to resist wear. Hardness is measured by Brinell Hardness Number (BHN) and rails are available in the range 220-380 BHN. But harder steel is more brittle.
At one time, the highest hardnesses were only possible with wear resistant steels which included chromium and vanadium. The downside was that these steels were both brittle and difficult to weld.
But today conventional carbon steel rail can match the former wear resistant grades by subtly varying the composition and subjecting it to various heat treatments. Even better the rail can be hardened on where needed.
First came head hardened rail. This takes rolled rail with an original hardness of 280BHN, reheats the head and then cools it under controlled conditions to give a hardness of 360-370BHN on the running surface, the hardness reducing to around 300BHM as you get to the top of the web. So, hard head, tough web.
Production of Mill Heat Treated (MHT) rail, introduced in the late 1980s, is more sophisticated. As the hot rail emerges from the rolling mill, it is cooled by water sprays under computer control. This results in a harder rail, 375-380BHN across the running surface and gauge corners, with the maximum hardness also extending deeper into the head while retaining that softer, tougher web.
Which is where the contact patch meets free will.
These processes have made wear resistant grade steel available at prices only a little above those of other grades. So max head hardness all round then and say goodbye to rail wear? Err, no.
Remember the contact patch under a Mk 4 coach is carrying five tonnes. That is a lot of pressure and under this pressure you get plastic flow – the steel behaves like pastry under a rolling pin.
If the steel is too soft, the steel flows round the top edge of the rail, the gauge corner, and there is massive wear. So max hardness after all? No!
Particularly with modern bogies the wheels are all running over the same narrow strip of rail. If the rail is too hard, resulting in little or no plastic flow or wear, the same steel surface is being repeatedly deformed.
And what happens if you deform a piece of metal repeatedly? It fatigues and cracks. And it has been known for a long time that if your rail is too hard for the duty cycle you will get fatigue cracking.
An early example was on the first Japanese Shinkansen line where head checks, as surface fatigue is generally known, began to appear after some 14 years operation.
At this point we need to differentiate between rail breaks as opposed to fatigue cracks. Broken rails are generally caused by isolated defects, either the result of an inclusion or other manufacturing fault or, more generally, by mistreatment of the rail.
Many broken rails start as a ‘squat'. In lay terms this is a ‘bruise' on the railhead, caused by the impact of passing trains with wheel flats or localised heating as a result of wheelspin. Squats are easily spotted and instructions for appropriate remedial action are well established.
But if they cannot be rectified in time, squats initiate cracks in the rail which are known as ‘taches ovales' because of their oval shape. These cracks then propagate down through the head, usually at an angle of around 45 degrees, under the forces of passing wheels. Eventually, the crack continues down through the web and you have a broken rail.
This is rarely fatal because the rail is held securely by the sleepers on both sides of the break. If fortune smiles, the rail is track circuited and the break shows up as an occupied section, setting the previous signal at red.
Should you be on an electrified line with single rail track circuit, an alert driver may spot the break and report it. At the worst, the train runs over the break, the driver feels the distinctive shock, and telephones the signalbox.
Whether you call it head checking (on the rail head) or gauge corner cracking, which starts at the rounded profile where the running surface blends into the vertical side of the head, the process is different and, potentially, very nasty .
You can understand why the gauge corner suffers when you realise that on a high speed curve, with a permitted cant deficiency of 6inches compared with the normal 4.25inches, the contact patches need help to react against centrifugal force. The axle moves across and on the high rail additional contact, still on a small contact patch, is made between the root of the flange, the curve where the tread blends into the flange, and the gauge corner. Same very high pressure on the same strip of hard steel.
Now, where the tache ovale is a bruise, a head check is a hairline fatigue crack. It is hard to detect visually and requires the ultrasonic test unit which is pushed along the head of the rail, to be tweaked. Apart from which the 070RTS manual ultrasonic tester has two probes which run on the centre of the rail head leaving the gauge corner exposed.
Head checking starts as vertical crack in the rail surface not more than 2-3mm deep. It then starts to grow along the surface. In the case of a gauge corner crack it runs vertically down the side of the head and along the running surface against the direction of travel at an angle of around 45 degrees.
At this stage it is not an immediate risk. The danger arises when the shallow horizontal crack suddenly turns downward and initiates a tache ovale. This, it is believed, is what happened at Hatfield.
When head checking began to emerge as a problem in the 1990s, Railtrack worked with Corus (British Steel as was) and Serco to try an understand what was happening. Fatigue cracks, after all, are not dangerous if you manage them properly. Readers of a nervous disposition should skip rapidly to the next paragraph. Aircraft fly safely with fatigue cracks in the airframe because they are known and the length and rate of growth is monitored.
This work showed that the length of the crack indicated the likelihood of its initiating a break. If you plotted crack growth against length on a graph, from 20mm on the increase in risk appeared as a straight line.
So the urgency of remedial action was categorised as shown in the table. The ‘severe' category also included any rail where the ultrasonic test equipment could not get a reading, for example on the gauge face, or where the surface had started spalling and the probes could not get good contact.
And here, as at Clapham, we see a failure in process damming the whole railway industry. According to informed sources, the rail at Hatfield was ultrasonic tested in April this year. The high rail was classified as unreadable which, (see table below) required an immediate 20mile/h TSR.
According to informed Sources the surface of the broken rail at Hatfield had sections of spalling, where a thin layer of the surface had broken away, nearly 40mm long and 3mm deep. The rail had also broken in up to a dozen places
Somehow the follow-up action was not taken with tragic consequences. But, if this report is confirmed it means that the most severe form of gauge corner cracking withstood a further six months of high intensity use which suggests that Railtrack's system for managing head checking incorporated adequate margins of safety. This further suggests that the subsequent programme of inspections and speed restrictions may have an over-reaction – assuming the inspection and reporting system could be trusted.
| Max length (mm) | Description | Action |
| 0-9 | Light | visual or u/s inspection |
| 10-19 | Moderate | visual or u/s inspection |
| 20-29 | Heavy | TSR based on risk analysis plus u/s monitoring until replacement |
| Above 30 | Severe | 20mile/h TSR and replace rail |
Railtrack's policy for managing head checking, which the HSE accepted, was to inspect light or moderate cracks, where the risk of a break was low, visually or ultrasonically depending on the traffic level.
Such shallow cracks were also removed by grinding the rail, in effect artificially wearing away the fatigued surface.
With ‘Heavy' cracks there is a higher risk of a crack turning down and initiating a broken rail. Thus a risk assessment would be made and an appropriate Temporary Speed Restriction imposed until ultrasonic inspection could be made and remedial work carried out. Finally, a ‘Severe' crack, required an immediate 20mile/h Temporary Speed Restriction, once again until ultrasonic inspection could establish the condition of the rail.
Another problem with head checking is that, unlike squats, multiple cracks over lengths of track measuring tens of yards long are a regular feature. In this case the track is classified on the basis of the longest crack.
Multiple cracks are what makes head checking so nasty. If the rail breaks and there is another fatigue crack between the broken end and the adjacent sleeper, the passage of trains on what is effectively a cantilever, will exacerbate the second crack and cause it to propagate rapidly.
One theory is that this mechanism explains why the Hatfield derailment occurred in mid train, rather than when the locomotive hit the broken rail. The first break may have occurred earlier, indeed there are suggestions that as many as 12 trains may have passed over the break before the derailment.
‘These rolling contact fatigue defects, which often appear on the high rail gauge corner and can be missed by ultrasonic inspection, pose a special safety risk in that they generally appear in groups at a given site. If one defect leads to a rail break, the increased dynamic forces produced at the break can cause nearby similar defects also to break, allowing the rail effectively to disintegrate'. Report by TTCI for the Office of the Rail Regulator Dated October 25 2000 |
After Hatfield, Railtrack's empiric classification came under fire from the HSE for not being having a scientific basis. Meanwhile, Railtrack was checking track and finding more and more sites requiring ultrasonic testing. With finite resources of trained manpower and ultrasonic test equipment it was a long job.
As more severe cracks were found more TSRs had to be applied. Meanwhile the HSE was concerned about the rate of progress and the empirical nature of the classification. Its response was to impose a deadline of Midnight on Sunday 5 November for ‘heavy' head checking cracks to be ultrasonically tested. Any outstanding sites would have a mandatory 20mile/h TSR.
According to Chief Inspecting Officer Vic Coleman, the HSE had been challenging Railtrack to ‘understand the scientific basis of their own logic'. He says that Railtrack offered to apply 20mph TSRs on the outstanding sites from midnight on Sunday, ‘and we said that's reasonable'.
This effectively extend the mandatory TSR to the ‘ heavy ' category of head checking, increasing the number of TSRs to around 500 . And it came into effect just as some Train Operating Companies were ready to introduce emergency timetables developed to cope with known TSRs after much burning of midnight oil.
This work was rendered nugatory because the list of speed restriction on which the timetable was based was too optimistic and trains could not keep even to the new timings.
So, total chaos and more to come, because no one knows what has caused the rise in head checking. It happens on straight track, on curves and on switches and crossings. It happens on high speed lines and on commuter routes. It happens on other railways, notably Deutsche Bahn And it is so bad that Midland Zone is having to lift rail it laid during the 1999 broken rail crisis on the WCML.
Were we perhaps too greedy by going for the MHT rail – which at one time seemed the cause? No, it has suddenly appeared in 24 year old rail.
Is it because modern bogies are stiffer in yaw for stability at higher speeds? That appeals to me, but it's happening on Southern Zone on routes where Class 455's are the dominant traction and rolling stock.
Is it anything to do with the change of gauge from 1432mm to 1435mm? Or has every operator started turning wheeltreads to a different profile, or aren't they turning often enough ? Or has the British Rail worn wheel profile been overtaken by bogie technology?
What we do know is that two years ago there was about 100 miles of track affected by head checking, most of it concentrated on the WCML. Now the total length is ‘many times that figure' according to informed sources.
Meanwhile, there are some nasty straws in the wind. One of the relatively recent – well last 15 years or so - changes, was British Rail's decision to adopt Exceptional Curving Speeds (ECS) based on 150mm (6in) of cant deficiency through curves in place of the previous 110mm (4.25in).
One thing Railtrack does know is that where you have ECS and high speed the rate of propagation of head checking is faster. This is understandable since both lateral and vertical forces on the rail increase with cant deficiency, increasing the stresses in the tread and root flange contact patches.
Thus for the immediate future ECS have been withdrawn for new track layouts and existing curves with more than 110mm cant deficiency will have additional inspections. What this means for Virgin's tilting trains which will, of course, curve at substantially higher cant deficiencies than 150mm doesn't bear thinking about. Indeed given the total shambles out there it probably isn't worth thinking about.
Something has gone wrong with that marvellous contact patch which has served the railways so well – or, more probably, some things because there is likely to be a concatenation of minor causes.
Railtrack has set up a ‘task force' headed by my old chum Prof Rod Smith of Imperial College who not only knows about railways but is a metal fatigue guru. He is supported by another old chum Brian Clementson now of Virgin Trains but better known as Chris Green's Traction and Rolling Stock engineer at Network South East. Rod Muttram, Director Safety & Standards Directorate and Andrew McNaughton, Great Western Zone director but and engineer by training, completes the team, together with a representative from DB.
Technical muscle will be provided by consultants TTCI, the consultancy arm of the Association of American Railroads, and ARUP. The eclipse British Rail Research, which once led the world on the dynamics of the wheel/rail interface, is complete.
Meanwhile, just as we went to press TTCI's assessment of Railtrack's rail failures commissioned the the Office of the Rail Regulator was delivered. It failed to foresee the current epidemic but did recommended that in view of the special safety risk posed by head checking Railtrack should ‘actively pursue' a research programme into how defects form so that mitigation techniques can be prepared.