Rail accidents must be put in perspective

As a result of the spate of four accidents on the railways across the world in July – Canada, France, Spain and Switzerland – I was interviewed several times about safety on the railways by both foreign and domestic media.

The first question was invariably some version of ‘are the railways safe to travel on?’

It was an easy question to bat away. The fact that these accidents happened in quick succession was just a strange coincidence. They were all different in nature – a runaway tanker train, a broken fishplate, an overspeed at a curve and a collision – and there was no pattern, apart from the fact that there were fatalities at each one.

This was, though, happenstance, just as there have been several long periods when there have been no accidents at all. I pointed out on the radio several times that some 3,300 people are killed on the roads daily. Yes, daily, a quite staggering figure amounting eight full jumbo jets falling out of the sky, and adding up to 1.2 million deaths per year.

However, while rail is undoubtedly much safer than road transport, there are still issues to address arising out of these accidents, in particular the Spanish one at Santiago de Compostela on July 24 which had such disastrous results and appears so eminently preventable.

The media rushed to suggest the driver was to blame but actually the cause appears to have been the result of a much more complex process caused by the failure to understand the risks the particular lay-out of the route posed. The accident happened 2.5 miles after the end of a lengthy fast section where the train had been travelling at around 140mph. The curve at which the accident occurred had a speed limit of 50 miles per hour and therefore the train had to be slowed down quickly. Moreover, there was change of signalling technology from the ETCS (European Train Control System) Level 1 system that would not allow any overspeed to the standard Spanish Asfa system.

On exiting the ETCS, the driver would have been alerted that that all automatic driving modes were disabled, and that manual driving mode was activated. However, ETCS for some reason was not operating as this type of class 730 set operate exclusively on Asfa on that route and therefore that warning was never issued.

Moreover, unlike on the UK network, there are no specific warnings to alert drivers to the possibility of overspeed at curves. In the UK, this was introduced as a result of several derailments on the curve at Morpeth.  Worse, the driver was apparently fielding a phone call from his own guard just before the accident, with a conversation over what platform would be used. In the UK such conversations would be banned as, quite rightly, they are a distraction

The driver, who may well have made mistakes too, was consequently left exposed in a way that should not happen in today’s railway.  The most fundamental lesson of 175 years of rail accidents is that safety systems should be devised around the notion that one single mistake by a railway worker should not lead to a disaster. Of course this type of accident cannot be designed out entirely but the risk should be reduced. Precisely the opposite seems to have happened in the accident at Santiago.

There are resonances, as has been pointed out by the excellent Zelo Street blog with the 1997 Southall disaster caused by a driver failing to observe two signals. It was suggested initially that driver Larry Harrison had been fiddling with his bag and that was why he missed the two signals.

Manslaughter charges were initially being considered, but they were soon dropped when it emerged that not only was the Automatic Train Control system that was being trialled on the Great Western and Chiltern lines not working, but neither was the basic Automatic Warning System either. Consequently Harrison was left with no protection from technical devices.

There is an added point to make here. Drivers in today’s environment come increasingly to rely on the fact that they are backed up by a safety system. So when there is none, drivers are more likely to commit errors than when they are used to working on their own.

We are in a transitional phase where the risk of a single mistake by a driver leading to an accident is gradually being phased out. It has been, though, a long phase. A primitive predecessor of AWS was first introduced on, ironically, the Great Western in 1905 but the system was not widely used until the 1970s.

The lessons of Southall were not learnt quickly enough as the 1999 Ladbroke Grove accident, just a few miles away, happened before the accident report had been published. This time there was a quick response with the speeding up of the implementation of the Train Protection and Warning System which had already been discussed. This was just the right sort of kit because, while it does not always stop a train which has passed a signal at danger before the overlap, it will slow the train down automatically and was also able to be introduced across the system quickly and relatively cheaply – a bargain at £500m.

Partly as a result of the introduction of TPWS, but also thanks to other improved safety features and, it must be said, luck, there has been just one fatal train crash in the past 11 years on Britain’s rail network, Grayrigg in 2007 caused by poor track maintenance like the recent French disaster. That disaster, too, is reminiscent of Potters Bar, too, as apparently the fishplate had three out of four bolts missing, an apparent failure of basic maintenance.

There is no room for complacency. To maintain the long term trend of improved safety, drivers have to receive more protection and systems have to be implemented to ensure that a single mistake does not lead to tragedy. It is terrible that the Spanish authorities made such an elementary mistake.

Scroll to Top