Earlier this year, Paul Darlington put together an article that dealt with some of the major accidents that have happened in recent history and how these have had a major impact on infrastructure and operating procedures.
Around the time that the article was published, the industry had a ‘bad week’. It involved a passenger train that was routed into the side of a work train at Waterloo, a freight train destroying the track near Ely, and a train hitting the buffers at King’s Cross.
Although all these specific incidents will be examined in detail and formal reports published elsewhere, here at the Rail Engineer we thought it might be useful if some of the general failure mechanisms of the rail/wheel interface were explained to the non-track or rolling stock fraternity.
Fantasy vs reality
There is a scene in the vintage 1953 film ‘The Titfield Thunderbolt’ where a tank engine is derailed and then proceeds to run down a road, through an advertising hoarding, then on through a wood before colliding with a tree.
Some may believe that this could be possible – the film-makers obviously thought so. For those involved in any aspects of practical railway engineering, this is understood to be complete fantasy – great fun, but utter nonsense. Sorry, but it really isn’t what happens when a train leaves the track.
The reality can be shocking. Even a low speed derailment will cause a disproportionate amount of destruction. Anything involving an element of speed will leave a lasting impression on the observer. Great fun it is not. Derailments are to be avoided.
It probably doesn’t give much cause for comfort, but the mechanism of a train travelling – and staying – on the track is largely still a gloriously crude Victorian (even pre-Victorian) mechanism. Essentially we have two strips of steel laid at a fixed distance from each other. The strips of steel are supported on a medium – timber/concrete (even plastic now) – which sits on terra firma. The wheels are made of steel, have running surfaces that are also a fixed distance from each other, courtesy of a rigid axle making up a wheelset. Helpfully, each of the running surfaces of this wheelset has a flange.
That’s it. Despite nearly 200 years, there is nothing else to augment this rail/wheel arrangement – nothing to stop the wheels bouncing up and over the strips of steel (which we will now call rails). The whole arrangement relies on gravity. It’s gloriously primitive and, despite the current preoccupation with additional safeguards and precautions, is still very simple and largely unaltered. If the railways hadn’t been invented when they were and if they hadn’t proved their effectiveness and their safety over generations, it would be inconceivable that they could be invented today. Who would believe they would work?
But, without any additional comfort mechanisms, it is important that the basic rail/wheel interface is managed rigorously as there is little scope for mistakes.
Of course it didn’t always work, which is why in those intervening 200 years we have invented some very clever rail steel and equally clever steels for the wheels and axle along with cunning rail and wheel profiles. These are so cunning that the flanges almost never touch the rail inside edges. But they are not so cunning that the flanges can be left off. That’s just a stage too far.
Defining a derailment
A simple search of the internet will come up with a definition of a derailment, and most definitions talk about a train accidentally leaving the track. Railway engineers have a gallows sense of humour, which they have had to adopt to deal with the enormity of these events.
‘Leaving the track’ is often referred to as ‘on the floor’, ‘on the dirt’, ‘on the deck’ or simply ‘one off’. In almost all cases, a derailment does indeed involve a train on the floor, but strictly a derailment happens before the train hits the deck.
Once a wheel has left the comfort of its rail surface it has derailed.
So, what will cause a wheel to leave a rail? Of course, there are a variety of reasons, broadly categorised as track defects, vehicle defects and outside forces.
These include a failure of the track itself or rail geometry that is outside of safe limits.
Sleepers, baseplates/chairs and fastenings all form part of the track structure. Any single failure of one of these components is generally unlikely to cause the track to collapse. There is enough redundancy in the system to cope with a couple of adjacent baseplates failing, for example, or for a few missing fastenings.
If track is not inspected/maintained, then there is a real chance of collapse leading, usually, to track spread – an inability of the track to keep the rails to the correct gauge. This leads to one – or both wheels – forcing the rails apart with a consequential rapid descent to the ballast.
The instant that the wheel running surface drops below the rail head is the point of derailment.
The normal sequence of events thereafter is for the wheelset to then unzip the rest of the track allowing the rest of the train to drop into the hole. This is often because the rest of the track generally is in an equally ropey condition and thus cannot withstand the forces involved.
There can be rare cases where the track ahead of the derailment is not in poor condition and is robust enough to resist the unzipping process. The wheel set may be persuaded to climb back onto the good track and for the train to continue its journey as if nothing has happened. Nevertheless there has been a derailment despite the track healing behind itself and there being little or no damage.
So, how can such an extraordinary event be detected? It comes down to wheel edge marks. As the wheel drops down the running edge, it cuts into the edge making a very distinct gouge – it’s the point of derailment. Similarly, if the wheel is able to climb back onto the running surface there will be a gouge at this point also.
If a rail fails, then there may or may not be a derailment. A simple rail break initially is an open joint. If a large section of rail is dislodged, then that’s when a number of forces take over. The pounding of wheels over the gap will cause a major geometry track defect which we’ll cover in a minute. Initially, wheels may ‘jump’ the gap, but this may not last for long before sequential rail/component failure occurs.
A rail failure that prompted the industry to take a long and hard look at the whole issue of incipient rail defects was the Hither Green disaster in 1967. Here, a bolt-hole crack caused a triangular portion of rail end to become dislodged. Instead of being kicked out of the way by passing trains, the fragment rotated and formed a ramp that propelled the following wheels off the track.
Track is rarely consistently billiard-table flat. In fact, on the transition from straight to canted track, or from one portion of canted track to another, there is a deliberate – a designed – change of crosslevel over distance. There are, of course, safe limits for this change in crosslevel. Vehicles will tolerate such variations. But there are limits!
If the change is too rapid, then it is possible for the wheel encountering the lower rail to be momentarily less loaded. In an extreme case, it can be seen that the wheel tread might leave the rail surface. Problems really arise if, at the same time as the wheel is unloaded, the whole vehicle yaws slightly from side to side. This can lead to the flange of the unloaded wheel being at an angle to the running edge such that it can start to climb up the running edge. (This is termed, unsurprisingly, flange climb.)
Once flange climbing starts then it takes very little imagination to understand that it will soon reach the running surface of the rail. From thence it can run on and indeed over the running surface and drop off the wrong side. At the same time, the wheel at the other end of the axle has fallen off its rail surface and the derailment is complete. Only in extraordinarily rare cases is there any hope of recovery.
The derailed wheelset – even if it’s the only set to come off – is busy destroying fastenings and sleepers. The rest of the train may not derail and it is possible for several miles of track to be severely damaged by just one wheelset. If the wheelset encounters a crossover or a set of points then destruction and further derailment is very likely.
The point of derailment is again shown by a mark where the flange rides up the running edge. There’s usually a scar on the running surface caused by the relatively sharp edge of the flange running on the top and then there’s a scar on the opposite rail where the sharp edge of the other wheel drops down.
Flange climbing isn’t only confined to track twist faults. Two particular problems emerged 20-30 years ago. One involved an issue with freight trains travelling over track that, in most respects, did not have twist faults that were out of tolerance. The track could not be described as ‘good’ but individual faults appeared to be relatively minor. The problem lay in the frequency of the faults and their spacing.
This was a classic interaction between vehicle type, speed and track fault frequency. What happens is that a resonance is set up in the vehicles, which come off after lurching from side to side and bouncing up and down. At a particular point in the agitated vehicle movements, a flange is able to climb up the running edge and the train is then off the road. The cyclic top issue had been discovered and it continues to bring trains off to this day.
The other issue concerned flange climb over switch blades. Often, this involved suburban electrical multiple units and caused significant disruption to services even though the derailments were generally low speed. A new regime of testing switch blades was initiated using special handheld gauges so that wear rates could be detected and remedial works undertaken in time.
Switch blades themselves can cause problems if they fit up incorrectly and allow a flange to run between the closed switch rail and the stock rail. This is known as splitting the points (again unsurprisingly) and is not always caused through badly adjusted points. An undetected run-through in the trailing direction can leave a gap wide enough for the next train in the facing direction to split the points.
Mention it not
And then there’s the cause that has no name – at least not a name that permanent way engineers wish to utter. Muttered under the breath it is sometimes referred to as a… misalignment or a… heat issue. Just cut the crap, we’re talking about a track buckle, plain and simple. Depending on the severity of a buckle, the results can range from a very rough ride to a rapid flange climb and derailment.
A buckle is caused by a build-up of compressive forces within the rail/s. Normally, the forces are managed by ensuring that continuously welded rail is fooled into thinking it was laid on a hot day, by having a surround of sufficient ballast and/or, in the case of jointed track, by having properly regulated rail joints. Buckles are not confined to hot weather and there can be rare cases of vertical buckles where rail joints ‘stand up to attention’. Vertical buckles rapidly convert to horizontal buckles as the raised portions of track fall over.
In almost all cases of derailment caused by track defects, it is possible to pinpoint the exact point of derailment and this gives a clue as to the cause. It is possible to take detailed track geometry readings upstream of the derailment as the track is undisturbed. Anything that is downstream is usually destroyed and in any case has little relevance to an investigation on cause.
There are naturally exceptions to the rule. If, after derailing, the vehicles come to a very abrupt halt, then a shock wave can travel back along the train – even back beyond the point of derailment. The shock can be sufficient to derail or cause track damage upstream of the actual point of derailment.
For many vehicle defects the upstream evidence can be vital. Marks on rails or through level crossings from errant components can be seen many miles ahead of the actual point of derailment, so giving a clue as to when the original failure occurred.
When looking at vehicle defects it’s useful to keep in mind the mechanisms involved with track defects. For example, flange climb – the very same mechanism – is involved if a vehicle has a stiff bogie. The bogie assembly is unable to follow the track geometry and so sets up an unnecessary angle of attack for the flange on the running edge. If the vehicle has unevenly set up suspension, then it will be overly sensitive to cyclic top type of defects, as it will if it is unevenly loaded.
Unevenly loaded freight vehicles, in combination with track geometry faults, can also result in flange climb. Containers can be problematic in this respect as any uneven loading is not visually obvious.
Buffer locking can happen as a result of a violent shunt or if a vehicle is somehow wrongly routed. It involves buffer assemblies at the ends of a pair of vehicles riding behind or on top of each other. The vehicles may be able to tolerate this whilst going round a curve but, when the track straightens out, the vehicles are then locked together and will not follow the track alignment. The result will be that one pair of wheels – at least – will be forced over the rail head and onto the floor.
Pretty obviously, if a component breaks on a vehicle, falls off and jams underneath then the result can be terminal.
Even with vehicles and track under control, it is still possible for the infrastructure to cause problems. Track geometry can be disturbed by failures of the supporting structures. Although very rare, it is possible for structural elements of bridges to fail. An example was the bridge near Stewarton in Scotland in 2009.
Perhaps less rare are failures of embankments, the causes of which can range from clay shrinkage, overloading and even instability due to rabbits. Rabbits can cause derailments – or at least the activities of many over the years. Warrens can riddle an embankment which, after a sustained period of rain, can collapse, causing just enough subsidence to affect the track with cyclic top.
Earlier, we summarised the various causes of derailments, the last category being ‘outside forces’ This is a catch-all term to include collisions with vehicles, overspeeding around curves, livestock, landslips and even ice in flangeways. All these effectively destroy the integrity of the track as a vehicle support system, lifting flanges over the running rail.
The list of derailment causes is long and, to the non-railway engineer – daunting. But it has to be kept in mind that, over the 200 years of railway evolution, all the causes have been recognised. There is very little left that will catch out the competent track or rolling stock engineer. It has all been seen before and strategies are in place to manage the risks.
As Paul Darlington stated, the main reason for dissecting railway accidents in great detail is to ensure that lessons are not forgotten. Hither Green, for example, was 50 years ago, but its legacy is still with us today.
This article was written by Grahame Taylor.
This is the second of a three-part series of articles on railway accidents, their causes and what lessons have been, or should be, learned. The first, From Blame to Better Understanding by Paul Darlington, appeared in issue 155 (September 2017). The third and last article, looking at wheel and rail profiles and their interdependence, will be published next year.
Read more: From Blame to Better Understanding