Posted 13th July 2020 | No Comments

Monday essay: look, no driver

Boris Johnson risked re-igniting a storm when he advocated driverless trains on the London Underground while on a visit to Siemens’ future train-building works at Goole. Rail unions are not in favour, but recently-designed new trains for the Underground include provision for automatic and even driverless working. Sim Harris describes how the foundations for this technology were being laid more than a century ago.

THE RMT does not support the idea of driverless trains on London Underground. On 6 July, following Boris Johnson’s remarks at Goole, the RMT’s senior assistant general secretary Mick Lynch said: ‘It is outrageous for Boris Johnson to wheel out the dangerous nonsense of driverless trains as a condition of the emergency Covid-19 funding of Transport for London. This is the sort of cheap political stunt that was a hallmark of his time as Mayor and we would have hoped he would have grown up by now.’

Until the start of the twentieth century train drivers had relied on visible signals, but their trains were not physically controlled in any way by the signalling equipment, which began as policemen on duty by the line, armed with hour glasses to enforce ‘time interval working’, and was developed into complicated systems of semaphore signals, complete with block working and mainly mechanical interlocking to prevent conflicting routes being set up.

The main electrical components of these systems were the block instruments and the telegraph, while track circuits began to be introduced at a few particularly hazardous locations, after a pioneer installation in 1864 on the London, Chatham & Dover Railway had been devised by Robert Sykes, who was an employee of the company's telegraph department and went on to make a name for himself as a signalling engineer, particularly after he had invented Sykes 'lock and block', which interlocks signals and points.

In 1872, American electrical and mechanical engineer William Robinson invented 'fail safe' track circuits which were able to detect hardware failures such as broken wiring or rails, and these became the accepted standard on both sides of the Atlantic.

A literal link between signal and train had to wait until 1906, when the Great Western Railway equipped locomotives on its Henley branch with underside shoes and matching tapered steel bars in the four-foot, which were placed at distant signals. The bars carried an electric current when the signal was clear, and this current, conducted through the locomotive shoe, rang a bell in the cab. If the signal was at caution the bar remained electrically dead but still pushed the locomotive shoe slightly upwards, and this time a steam whistle blew in the cab. An improved version accompanied the whistle with an automatic brake application, achieved by a valve connected to the vacuum pipe. If the current was off for any reason the equipment therefore gave a warning as the default, making it fail safe. The system was particularly valuable in fog.

This system, known as ATC (Automatic Train Control) or ‘audible signals’ and now more accurately described as an Automatic Warning System, spread quickly to other GWR lines. A contactless alternative, using magnets, was later introduced by the LMS.

This was the state of play at nationalisation in 1948, but the Railway Executive took several years to evolve a compromise design which included the best features of both, and this would become British Railways’ AWS.

A successful prototype was unveiled in the summer of 1952, but in October came the horrific double collision at Harrow & Wealdstone. The crew of an up express had failed to respond to a yellow colour light distant followed by two semaphore home signals at danger and ran their heavy train into the rear of a stationary local train, which was crowded with commuters bound for Euston. Seconds after this collision, a down express collided with the wreckage. Altogether 112 people lost their lives. The cause of the oversight by the crew of the up express was never known, because they were both killed.

This exceptionally destructive accident, the worst ever experienced in peacetime, was followed by a programme of AWS installation.

Transferring the task of driving to an automatic system followed, and the first successful example of ATO, or Automatic Train Operation, was introduced on the new Victoria Line of London Underground in 1968. Under ATO there is still a driver in the cab, but in normal working the driver only closes the doors and gives the signal to start. The train then ‘drives’ itself until it comes to a stand at the next station. The driver can take over fully if necessary, using standby conventional controls.

ATO is also now in use on the Thameslink core section under central London, overlaid on ETCS Level 2. The consistent driving style ATO offers makes the best use of capacity and allows very short headways of 'metro' frequency.

Three more London Underground lines have been equipped with ATO since 1968, and the Underground’s new sub-surface and tube stock includes provision for ATO and possibly something still more advanced. This is driverless operation, in which there is no longer a cab at all.

The Docklands Light Railway, opened in 1987, was designed for driverless operation from the beginning. However, there is still a member of staff on every train, who closes the doors and is also qualified to take over if the computerised system fails, by unlocking a control panel.

The last step is to remove even the final member of staff, and this has been dubbed UTO – Unattended Train Operation. You will have travelled on a type of UTO train in Britain if you have used the people movers at airports like London Heathrow or Birmingham, but some other countries, including many in the far east, have adopted full UTO on their newer metro lines. With UTO, ‘look, no driver’ becomes ‘look, no train staff’.

The system, although efficient, may not be suitable for the Underground in London, whose tunnels were not designed with escape routes. The only way to evacuate a train is to take the passengers along the track to the next station, and that needs staff on hand. ATO, however, does appear to be the way forward, and London Underground is planning to convert more lines, including the sub-surface network, although a major resignalling project on these lines needs to be completed first.

The distinction between driving trains and signalling systems has become blurred. Lineside signals may have a limited future – there is no point in showing indications to a driverless train and not very much on systems with ATO, particularly if ETCS Level 2 is also present to provide in-cab information, as it does on Thameslink. (There are still signals in the Thameslink core at the moment, because driver training for ETCS/ATO working has been delayed by the pandemic and some trains must be driven conventionally for the time being.)

Finally, what about heavy rail more generally? The future is less clear, but French state operator SNCF revealed in 2018 that it is developing driverless long distance, commuter and freight trains in partnership with the IRT Railenium research institute at the University Polytechnic Hauts-de-France in Valenciennes, the French rail safety authority EPSF and the information security agency ANSSI, with a target date of running the first automated trains in 2023.

In the meantime, we might also draw some tentative conclusions from a 1,500km automatic rail freight network in Pilbara, Western Australia, which connects more than a dozen iron ore mines with terminals at several ports. There are no drivers on board any of the 240-wagon double- or triple-headed trains except for the last kilometre or two into the port terminals. Owner Rio Tinto has described its automated railway as ‘the world’s largest robot’.

For more on this topic see the forthcoming feature in the August print edition of Railnews, RN282, which will be published on 30 July. The new edition and some previous issues can be obtained by calling 01438 281200 from UK numbers or +44 1438 281200 internationally, and selecting Option 2.