Making London Underground ‘driverless’ is financially unjustifiable, TfL report states

Passenger safety at the 'platform train interface' is paramount, and TfL's report argues that a train assistant could not possibly manage the problem by looking out of a train door.

Although Prime Minister Boris Johnson has stated that making London Underground ‘driverless’ should be a condition for government increasing its funding of Transport for London, a leaked internal TfL document has shown that this is financially unjustifiable and could cost up to £7 billion to implement, with a negative return.

The report, in the form of a slide presentation leaked to the ASLEF train drivers’ union and seen in full by Rail Insider, is dated August 2020 and examines the issues and constraints for introducing driverless operation on the LU network. In particular, it considers the key elements of costs and benefits, highlights the financial and economic case for driverless and identifies practical next steps for driverless on LU.

The report uses the UITP (international passenger transport union) definition of the four Grades of Automation (GoA):

The four Grade of Automation (GoA) levels considered in the TfL report.

In its summary, the presentation concludes:

  • Driverless operation requires high up-front capital spending on infrastructure, as in the most recent line conversion in Paris;
  • Whilst some operational savings are achievable, overall, the case is not financially positive given the high capital costs;
  • Significant customer benefits are possible, most notably from improved safety, accessibility between train and platform, the ambience of the platform environment and some reliability improvements (which could be delivered with or without trains being driverless);
  • When delivered as part of an integrated line and train system upgrade, driverless operation offers reasonable value for money for these passenger benefits, although the value-for-money case is much weaker than that for moving from GOA1 to GOA2, which always delivers the majority of journey time and capacity benefits in an integrated line and train systems upgrade;
  • Value for money is significantly weakened if driverless conversion is undertaken via a retro-fit of existing trains and infrastructure, due to increased costs, risks and shorter remaining life of key assets;
  • Value for money and affordability of driverless conversion should be considered as opportunities arise alongside planned line upgrade and asset renewal programmes through the 2020s-2040s – the driverless case is unlikely to be strong enough to justify prioritisation ahead of planned capacity enhancements on the LU network;
  • Whilst there are marginal positive benefit-to-cost ratios, the financial payback is negative.
Although DLR is already driverless, the trains and platforms are shorter and usually less crowded, so the train assistant can see potential problems more easily.

Platform Train Interface (PTI)

Proponents of adopting a driverless system often refer to the Docklands Light Railway (DLR). These trains have no driver, but there is a manager (mobile attendant) on every train who both closes the doors when the train is ready to depart and can drive it, using a concealed control panel, in emergencies.

Why not, the argument goes, do the same on the London Underground?

The TfL report considers this option. It notes that the network was originally designed and built in the mid-1980s as a Light Rail system with driverless (GOA3) capability. The straight, level access platforms and low ridership allowed door control and safe train despatch by a mobile attendant from any train door position.

However, over the last 30 years, Docklands development and rising demand has driven progressive network expansion and higher frequency services but with ‘grandfather rights’ to extend operations using the original GOA3 operating model.

At peak times, and on more crowded parts of the system, attendants performs their duties from the front control panel (effectively in GOA2-mode) to manage the increased PTI safety risks now being experienced. Even so, the busiest 10 station-to-station links on DLR have an average of 18,247 weekday boarders, compared with 66,325 on LU – or around 28% of LU levels.

In further contrast, the London Underground network was originally designed and built in the 100 years between 1860 and 1960. It has a wide range of platforms topography, with many being neither straight nor level. The restricted space, high crowding levels and safety hazards from large steps and gaps between train and platform present significant risks to customers using the system.

Some Underground stations, such as this one at Bank, have a considerable gap between the train and the platform edge.

All LU lines operate with a driver in the front cab (in GOA1 or GOA2 mode) who manages the PTI risk, supported by an array of CCTV images of the whole train in the platform to enable door closure and observe safe train departure. CCTV images of the PTI are displayed on monitor screens located in a secure cab environment, free from distraction by customers.

The report concludes that, in driverless mode, a mobile attendant, located within the train, would not be able to safely observe all parts of the platform-train interface from the train door positions under normal loading conditions and would also face significant distraction during this safety-critical task.

As a result, a physical barrier system, such as platform edge doors (PEDs), or the full-height platform screen doors already used on the Jubilee line extension at stations such as Westminster, is considered a prerequisite for safe PTI management in driverless operation on LU, guaranteeing the separation of customers from moving trains in the absence of human vigilance and intervention.

Discussions between TfL and rail regulator the Office of Rail and Road (ORR) have established that, while the ORR has no objection in principle to having driverless trains on the Underground, there must be demonstrably no more risk than today to the travelling public and, ideally, an overall reduction in risk following its adoption.

The ORR has indicated that either sophisticated PTI object-detection equipment or some type of PEDs would be essential to properly manage platform–train interface risks. It was also thought that the potential unreliability of the former would mean that PEDS would be the preferred solution, but that this would be a decision for TfL to make.

Stations on the Jubilee line extension (east of Green Park) already have full-height platform screen doors.

Financial consequences

The TfL report considers the cost of conversion – platform edge doors, new signalling and control systems, retrofitting trains and stations – as well as the higher cost of maintenance (more equipment to maintain, more safety-critical systems).

It then factors in cost savings from having less-qualified staff (attendants vs drivers) – “The train attendant’s primary customer service and information duties will not demand the high levels of training and safety-critical licencing of the current Train Operator role. Hence a salary cost reduction could be realised” – but the overall number of staff would not change significantly as there would still be one person on every train.

Some trains would still need drivers and to be operated under GoA2. On the Bakerloo line, GoA2 operation would continue North of Queen’s Park, where Network Rail track and platforms are shared with TfL London Overground trains. On the open sections and interoperable areas of the Piccadilly line, where PEDs and level access cannot be achieved, there would need to be a mode change between GOA2 and GOA4 operation for the services over the Metropolitan line between South Harrow and Uxbridge.

On the sub-surface railway (SSR – the Circle, District, Metropolitan and Hammersmith & City lines), there are also open areas and interoperation with National Rail and London Overground, which would require GoA2 to be retained.

As a result of all of these factors, the report concludes that the network-wide incremental CAPEX (capital expenditure) cost for a conversion to GOA4 operation would be over £7 billion, and this investment would yield only around a 5% OPEX (operational expenditure) reduction, relative to the upfront cost of conversion.

So the Net Financial Effect (NFE) of an LU network-wide conversion would be negative, showing a net cost of around £5.6 billion after accounting for revenue and OPEX savings.

Indeed, considered separately, all LU lines have a negative NFE. Therefore, none of the GOA4 conversions would cover their costs over their expected asset life.

Union view

Mick Whelan, ASLEF.

Train drivers’ union ASLEF was quick to pick up on the financial figures contained in the report. “We know that passengers don’t want driverless trains,” general secretary Mick Whelan said.

“Whenever they are asked, they always answer with a resounding ‘No!’ And now we know it is not economically feasible, either.”

Finn Brennan, ASLEF’s organiser on London Underground, added: “Leaked internal Transport for London documents demonstrate that it would cost an additional £7 billion, on top of the money needed to upgrade existing lines, to make Underground trains ‘driverless’ and that TfL has concluded that there is, given the evidence, no economic case for doing so. In fact, the cost of conversion would make TfL’s current financial crisis and the long-term funding issues facing TfL much, much worse.

Finn Brennan, ASLEF.

“Prime Minister Boris Johnson has said that introducing driverless trains should be a condition of a funding settlement for TfL. But these documents show that, not only is there no business case for this, but it would make TfL’s financial position much worse.

“When Boris Johnson was Mayor of London, his failed vanity projects, like the Garden Bridge and those Boris buses, cost Londoners almost £1 billion. But this is dwarfed by the vast cost of his ideological obsession with driverless trains, a project that every transport expert, as well as TfL senior managers, have concluded makes no financial sense.

“‘The presentation shows that, given the constraints of the Underground’s infrastructure, there would still need to be a train operator on every train for the safety and security of passengers, and that any cost savings are offset by the increased maintenance cost.

“ASLEF has always pointed out that driverless trains on London Underground are a politically-driven fantasy. These documents show that we are right. And if the government tries to force TfL to waste huge sums on this pointless exercise, it would suck resources away from projects that could have real positive benefits for passenger safety and bankrupt the entire Tube network.”

Be the first to comment on "Making London Underground ‘driverless’ is financially unjustifiable, TfL report states"

Leave a comment

Your email address will not be published.


*