At London Reconnections we don’t want to appear to always be bringing bad or disappointing news but events over the past year meant that we rarely have good things to write about. It makes a welcome change, therefore, to update readers with the success of the implementation of the first stage of Automatic Train Operation (ATO) of the Sub-surface Railway (SSR) and to look forward beyond it.

A much needed upgrade

New lines are always more exciting than upgrades to existing lines. On the Underground, the primary objective of any upgrade is nearly always to provide a capacity increase. This is certainly one of the main objectives of the SSR upgrade which is formally referred to as Four Lines Modernisation (4LM) and encompasses the Metropolitan, District, Hammersmith & City and Circle lines (also known as the Sub-surface Railway or SSR).

A major part of the upgrade of the Sub-surface Railway is the replacement of rolling stock. This phase has already been completed. Much needed as the new rolling stock was, it didn’t do that much to increase capacity. The significantly longer trains (an extra carriage length) on the Circle and Hammersmith & City lines certainly help. The extra standing space inside the carriages also helps – albeit at the price of a reduced number of seats per carriage.

The signalling factor

Another objective of the upgrade is to replace the old, unreliable signalling. But it makes no sense these days to replace signalling on a like-for-like basis. Whilst you are replacing the signalling you might as well improve it, where there are quantifiable benefits in doing so.

To get the full benefit of the substantial increase in capacity on the SSR you really need automatic train operation. Normally there is the additional cost of the extra trains required but as the fleet size of the SSR stock (S Stock) was calculated assuming ATO would be introduced, there is no need to buy additional rolling stock in order to increase capacity. The stock is already there sitting in the depots.

No one said it was easy

ATO on the SSR is not easy. Although it has been applied to other lines, these have a limited number of junctions. Providing efficient, capacity-increasing ATO on a complex network with a central ring and many flat junctions off it (some in close proximity to each other) is a challenge. One for which there is probably no precedent anywhere in the world.

The implementation has also been made harder (but more beneficial) by using radio waves for all train-track communication. This dispenses with the additional wiring laid between the rails – as featured on the Northern and Jubilee lines.

A much delayed upgrade

The history of ATO on the SSR has not been good, as long term readers will recall. The original tender for ATO was issued under the government-mandated Public-Private Partnership (PPP) scheme. On this, London Underground had no say. The collapse of PPP led to London Underground cancelling it whilst they reassessed the situation and decided what they wanted from such a scheme.

The second ATO contract went to Bombardier to install what was believed to be a mature reliable product. It soon became apparent, as is often the case, that the conditions on the London Underground were more demanding than in many other apparently similar situations in the world. It then became apparent that it was going to be a struggle for Bombardier to get its product to work in London.

Frustrated, the TfL board cancelled the contract, but not before around £100 million had been spent on it – money that it turned out was largely wasted as the next supplier had to start from the beginning again.

It does seem that a lot of the problems with the Bombardier contract were down to an unfortunate period in Bombardier’s ATO development. At the time, Bombardier were buying both various fledgling and also some more mature ATO systems but neither consolidating them nor deciding which ones to support in future. As a result they had a multitude of different systems all touted by their Canadian sales team who were probably not up to speed on fully understanding the capabilities and limitations of each offering. There is, therefore, the suggestion that not only did they not fully understand London Underground’s requirements, but that they may not have put forward the best technical solution utilising the most appropriate Bombardier ATO offering.

Once the second contract was cancelled it was immediately obvious that there was a major problem. For all the bold talk from TfL about getting value for money with a competitive bid, one of only two potential contenders with a proven capability to deliver, Siemens, just wasn’t interested. Siemens were busy with other projects – ATO and ETCS on Thameslink and CBTC on Crossrail, for example. Realistically, this left only Thales. Thales had converted the Jubilee line to ATO, but not without major problems and, rather more successfully, had also converted the somewhat more complex Northern line to ATO.

The third contract

So Thales were the only game in town and they knew it. Whilst it could have been worse for London Underground, this meant that Thales weren’t going to be pushed aggressively on keeping costs down. Nor were they going to be tied down to a tight timetable. The TfL board were hoping that the 2018 target could still be met. Thales made it clear this was very much not the case. They won the bid but would only commit to a project completion of 2023.

Although we are led to understand there was considerable frustration in London Underground at the reluctance of Thales to commit to an earlier date, it is likely that it was unreasonable to expect otherwise. Thales had a huge work commitment in both Singapore and Hong Kong and it probably just wasn’t possible to promise earlier completion dates for London Underground. Whilst competition on cost might have been nice, at the same time the lack of pressure on Thales likely meant that the timetable they proposed ended up being far more realistic – and achievable – than is sometimes the case with contracts on this scale. A subject to which we will shortly return.

It is also probably worth noting that both Singapore and Hong Kong had train crashes caused by ATO – the one in Singapore was when the trains involved were in passenger service. This serves as a sober reminder, if one were necessary, that ATO systems really do need to be comprehensively tested before they are implemented.

A further potential problem was the aforementioned belief that the failure of the second contract was partly down to the suppliers at Bombardier (a Canadian company) being out of touch with what was required in London. Although Thales is a French company, its train automation division was based in Toronto in … Canada. No doubt Thales realised that a lot of effort would have to be put in to ensure that the previously perceived problem didn’t bedevil the revised offering.

The New York factor

When you have only one supplier capable of doing the job, you are usually not in a good position. In this particular case, however, it seems to have worked out well for London Underground. In fact Thales really wanted the contract and was anxious to co-operate fully and do a good job. That is not to say they didn’t make sure this was a profitable contract for them but in reality, they could have demanded a lot more. To understand why they didn’t you have to look across the pond to New York.

New York currently only has two ATO, lines and these are not complex ones to operate. With 14 lines in total, comprehensive adoption of ATO would produce enormous benefits. It is known that the head of New York’s transportation system, Andy Byford, is keen to extend its use. The New York Subway is quite complex. Like London Underground, some of it is even four-track and the lines are not fully segregated from each other.

Furthermore, for historical reasons dating back to the start of the 20th century when the Underground really developed, its power supply setup is almost identical to that of the New York Subway. So basically, if it works in London it will probably work in New York.

The New York Subway has many problems but the principal one is lack of capacity. In terms of asset status, the system is roughly equivalent to London Underground in the 1980s. As it was described in the pub at LR Drinks, imagine what could be achieved if you apply modern day ATO to a network akin to London Underground in the 1980s.

Another potentially attractive factor for Thales is New York is less than 600km away from Thales Rail Signalling Solutions at Toronto – practically on their doorstep by world standards.

Effectively, ATO on SSR in London is the shop window display to entice New York to provide a lot of work for Thales. In this particular instance, even more so than normal, the Underground and the Subway are truly sister systems. Something the current political leadership in New York would do well to note.

The stakes for Thales were already high but with the planned congestion charge to be introduced in New York “by 2021” the stakes have got higher still. One of the reasons for introducing the charge is to provide around $1 billion annually – most of it to be spent on the New York subway. For the first time ever, a realistic means of paying for ATO on the New York subway has been identified and looks like being implemented.

Thales – but really a triumvirate

Whilst Thales is the headline name, the task of implementing ATO on the SSR is really mainly down to three companies.

Most obviously, Thales itself. They have to provide and install the equipment on the track and in the central control office.

Less obvious, but also important, is Bombardier who originally supplied the trains (and lost that second contract). They have to modify the trains to work with the supplied Thales equipment. This is not a small task. It is not helped by the fact that one is limited by how many trains one can work on at one time – either due to factory capacity limitations or due to a certain number of trains being required to maintain the existing service level.

Also easily forgotten in the ATO upgrade is London Underground itself. Before Thales can install equipment on the Underground, London Underground has to make the space available and, sometimes, provide the buildings in which the equipment will be housed. They also have to re-write operating procedures, retrain staff such as drivers and controllers and also recruit and train the extra drivers needed to run a more intense service.

In addition to this, it is important to note that other organisations are also involved. These include the operators of the railway test track in Melton Mowbray, the railway regulator and even Crossrail, who must ensure the SSR upgrade does not affect Crossrail and vice versa.

The two challenges to Thales

For Thales the contract really boils down to two major tasks. One task is to physically install the equipment across the network. The other task is to get it working. Clearly the two are related. Equipment can’t work if it hasn’t been installed. But also, you don’t want to install equipment until you are confident it works – or at least the hardware works.

For Thales, or any signal automation supplier, there is clearly a balancing act to be found. The obvious thing to do is to install equipment on one part of the line and get it working, before continuing to the next stage. Unfortunately, if this was rigidly done, rollout would be very slow indeed. So there has to a vanguard phase on an implementation programme and a follow-up phase that subsequently implements ATO in separate sections of the SSR network.

Thales gets the 2018 curse

Along with many other railway projects in London, things didn’t quite go right for Thales in 2018. The first section of ATO was due to go live in May but was put back to June. This had to be called off at an early stage as trains were taking longer to travel through the migration area than expected. This was a considerable issue as, on leaving the migration area which was closed to passengers, the trains went into public service along the next section of track. With the weekend train service collapsing on the SSR, the planned introduction was aborted early on.

From then on, dates got planned and inevitably these planned dates leaked out to become known in railway circles. Then reports would come in that the planned implementation had been put back again. Various reasons were suggested for this. One of the plausible ones was that there was little point in rushing and, with not all the trains converted, there was a risk of disruption if a non-converted train was pathed over the ATO section of track. Another one was that weekend testing on future phases was still continuing and feedback from them could be used to increase the probability of a second attempt at introduction being successful.

Two decisions were made. An obvious one was that more testing was required before the first section could be cut over to ATO mode. The other big decision was that the pace of equipment installation would continue as planned, but that implementation section by section would be held back until all were satisfied the system was fully working.

Delaying actual implementation did have the advantage that it helped resolve the issue where one had to be careful only to send ATO-equipped trains into an ATO area. This would become even more critical early on as stage 2 (of a total of 14 stages for the complete SSR) required every single S8 (Metropolitan line) train to be ATO-fitted.

In fact, continuing to install the equipment had other potential advantages. A testing programme on the newly installed areas meant drivers still operated under ATO (but not in passenger service) which would help avoid having to retrain them through lack of continuing experience. It also meant that testing could continue ensuring that the future ATO areas could handle the capacity necessary for future plans.

The crucial test

So it was then that the date was fixed for implementation of the first section. Because it belatedly decided to phase-in the smallest length of line practical for the first phase, the originally-planned first phase got split in two, with the first section to go live given the title ‘Signal Migration Area 0.5’. This took place on the weekend of 16th/17th March 2019. Our report of this is online, but there are other reports available on the internet. All seem to agree it went well, but not perfectly, with a few little minor niggles to be fixed over the coming months.

There was probably a massive sigh of relief when everything worked. To be clear, it was expected to work and earlier, comprehensive testing had gone well a couple of weekends previously. Nevertheless there is no proof better than actually having a live implementation.

Benefits sooner than originally planned

The plan was always to introduce a new, more frequent timetable once the entire route of both the Circle and Hammersmith & City lines was complete. This was planned for April 2021. This improvement has now been brought forward and it is now intended to introduce a new timetable prior to full completion of the Hammersmith & City line, but after the Circle line is complete. The date given by the latest Programmes & Investment committee documents is March 2020. The Hammersmith & City line will by then be largely complete, but trains between Stepney Green and Barking will still be operated manually.

Never trust …

We feel a need to issue a caveat here. Roger Ford of Modern Railways has famously said that you should never trust a date based on the seasons. At London Reconnections we should say never trust a date quoted in Programmes & Investment committee documentation unless independently verified. Dates are often hopelessly optimistic and we are led to understand that March 2020 is an aspiration not a committed date. The desire is to implement this new timetable as soon as possible with 2021 still the ‘official’ target and the date quoted in the most recent press release.

Why change the plan?

The obvious question to ask is: why the change? Of course, this would appear from a passenger perspective to be a good thing, but maybe it is not. The obvious reason to bring ATO forward is to increase capacity on the SSR to make up for the fact that Crossrail is not yet open. Crossrail was, of course, intended to relieve the lines as they are today. It is hard to imagine any other explanation. So, perhaps, in a way this really isn’t good news after all as it could be construed as indicative of just how much Crossrail is expected to be delayed.

Signal migration areas are at the eastern end of the District line parallel to the electrified lines from Fenchurch Street

There is another potential benefit in opening early. All the stages that need to be covered have only limited exposure to the risk of 25kV interference from other nearby lines. For the next stage (Stepney Green – Barking) this is no longer true. So it makes a lot of sense to implement improvements prior to attempting to cut-over the first section of the SSR that runs parallel to a 25kV railway for a significant distance. That means that early benefits are no longer dependent on getting ATO working in an area with long stretches of 25kV overhead cables nearby.

The need to plan for more drivers now

Unfortunately, we were not privy to the proposed timetable. We presume it must be known, or at least the intended frequencies it contains are, as this will determine the number of drivers required. With the implementation due in less than a year, if it were of a substantial nature, London Underground need to be thinking now about advertising for the extra drivers. Things will probably get more complicated than one might initially think as some drivers like ATO and some don’t. So as well as newly trained drivers, London Underground probably needs to sort out all the requested driver moves to a different depot.

What could the March 2020 benefits have been?

As initially we did not know the exact details we speculated based on information provided in a TfL press release. We made two main assumptions. One is that the off-peak will be much as it is today. This seems to be sufficient to handle the current demand so why make things difficult?

The other main assumption, entirely wrong as it turns out, concerned peak frequency. We presumed a substantial uplift. The current service is roughly based on the 6tph Hammersmith & City and Circle line frequency. This could go up to 7tph but this is always messy. It would have been far more likely that it would go up to 7.5tph. This would then have meant that in sections where the original Circle line shares the line with other services, one could have expected 30tph in total. These sections are the Gloucester Road – Tower Hill section of the District line and the Baker Street – Liverpool St section of the Metropolitan line. This is only slightly short of the intended ultimate frequency of 32tph.

One unknown was how many trains could run between Aldgate East and Barking. We don’t know quite what the old (current) signalling can handle. Nor do we know if the power supply is sufficiently upgraded for 30tph. Under the scenario we speculated on, there may well have been some peak-hour District line trains that terminated at Tower Hill.

We then assumed 15tph on the District line beyond Barking to Upminster. In reality this is unlikely and it is probable that, as now, some would have terminated at Dagenham East to leave approximately 12tph going all the way to Upminster.

We also do not know how many trains would run north of Baker Street but we assumed that, in terms of frequencies, the Metropolitan line is basically unchanged although the press release did suggest there would be frequency improvements on all four lines.

It is was also assumed that there is not a problem implementing a slightly more frequent service between Gunnersbury and Richmond where the District line shares track with London Overground. As both services terminate at Richmond, it was presumed that any minor adjustments necessary could be easily made and achieved by varying the arrival and departure times at Richmond.  It is our understanding that the National Rail timetable was adjusted between Richmond and Gunnersbury as long ago as May 2018 to take into account the need for flexibility and to adjust to revised planned frequencies on both the SSR and London Overground.

These assumptions lead to this expected level of service from March 2020.

Expected frequencies on the SSR, March 2020


Had our assumptions been correct, in simple terms and in brief the benefits would have been

  • Approximately 10% increase in capacity between Gloucester Road and Tower Hill and between Baker Street and Liverpool St as a result of a more frequent service
  • Trains every four minutes instead of every five minutes on the Hammersmith branch, between High Street Kensington and Edgware Road and between Edgware Road and Baker Street
  • A slight peak improvement from Ealing Broadway with 7.5tph (up from 6tph)

A possible slight disadvantage was that marginly more Wimbledon trains would go to Edgware Road instead of via Victoria in order to provide an even service to both routes.

The Unfortunate Reality

It turns out that, not for the first time, there is more hype than substance and the improvements planned are somewhat underwhelming. This disappointment is particularly so as TfL were quick to provide significant improvements to the Jubilee and Northern lines once signalling upgrade work had been completed.

It seems that the only improvements planned in terms of capacity are ‘Additional peak City trains’. It is not even clear if these are extended trains continuing from Baker Street to Aldgate or entirely new services. Given that the rest of the Metropolitan line has not yet been resignalled we strongly suspect the former. With the delay to Crossrail until at least September 2020, the much-needed increased provision of services between Farringdon and Moorgate is welcome but little more than a drop in the ocean.

Critical Future Dates

Signal Migration Areas 1 -5

As far as can be ascertained, the critical dates are as follows

Stage 2: Latimer Road, Paddington (District and Circle) and Finchley Road to Euston Square.

This is a larger area than originally intended. There now seems little point in opening Latimer Road – Paddington separately, given the amount of testing that has taken place and the confidence now present in the new signalling. As they want to identify and fix all problems that can be identified and fixed on the first section, this is not expected to happen until July.

Note that this stage takes in the first of many critical junctions. In this case it is Baker Street junction where the Metropolitan line joins and leaves the Circle line.

The lookahead dates of planned closures on the Underground suggest that July 20th-21st is the most likely date but the sections of line closures don’t quite correspond with what would be expected

Stage 3: Euston Square – Stepney Green and Monument

This becomes a more challenging migration with the complex Aldgate triange being converted. This involves three flat junctions in close proximity to each other. It also involves platform 2 at Tower Hill becoming a through platform although normally, for operational purposes, it will be used in the same manner as today with trains arriving from the west and departing westwards.

This is rumoured to go live in early September which makes a lot of sense. It is not too soon after stage 2 and is before the Underground gets busier again as school term time returns and family no longer go on holiday.

Stage 4: Monument – Sloane Square

This is a relatively simple section of plain track with just a trailing crossover at Embankment to complicate things very slightly. It probably only exists as a separate stage because the two migration areas either side are already very large. It obviously has to take place after stage 3. Because TfL do not allow planned engineering work to close the Underground in December prior to Christmas we can be fairly certain this will be some time in late October or November.

Stage 5: Sloane Square, Fulham Broadway and Barons Court to Paddington (District and Circle).

This is another large area and challenging migration area with another triangle of flat junctions. It also includes the single line branch to Kensington Olympia. With the completion of this the Circle line will be entirely ATO signalled. This does not mean Circle line trains will only operate in ATO mode because they will still need to travel to depots located beyond the Circle line.

A statement in the latest Programmes and Investment Committee minutes states that:

We are targeting commissioning of the whole Circle line by the end of 2019 to support an early timetable uplift in March 2020.

Given the ban on engineering work prior to Christmas and the ‘end of 2019’ deadline it seems inevitable that this will be implemented between Christmas and the New Year. In fact one can go further as on New Year’s Eve there is always a shortage of train operators (drivers) during the day, as the day is relatively quiet and drivers are needed for the New Year’s Eve all-night service. So, all being well, this must take place between 26th and 30th December.

Hope for the future?

If all now goes to plan it looks like after a difficult gestation the Four Lines Modernisation is going to deliver its first actual improvement in frequency in little under a year from now. Let’s hope that this new timetable is delivered without the hiccups that seem to have plagued rail transport in 2018.

Thanks to ngh for background information on the signalling suppliers.

Written by Pedantic of Purley