Third Time Lucky: A Look At The New Sub-Surface Signalling Plan


We have reported previously on the ongoing problems with the resignalling contract for the Subsurface Railway (the Circle, District, Metropolitan and Hammersmith & City lines). The contract is an important one, for it underpins much of the work currently underway on the Underground. In advance of a TfL Board Finance and Policy meeting later this week a paper on this contract has been published. We look at its contents with a critical eye to see if a sound plan for the future has begun to emerge.

It is one of John Bull’s working principles that the more boring the name of a committee, the more interesting its remit is likely to be. As if to prove the point, the Finance & Policy committee, to whom proposals to spend large sums of money have to be justified, often presents rich pickings for those who want to get beyond the hype when looking at future plans. It is another of John Bull’s principles that one should read, or a least re-read, a press release from the bottom up, as organisations tend to put the most awkward truths at the point where they hope you have, at the very least, lost concentration or, better still, stopped reading. It therefore comes as no surprise that the final item on the agenda for the Finance & Policy committee is very interesting indeed.

It is hard to imagine that even occasional readers to London Reconnections are not aware to some extent of the numerable problems that have emerged in trying to resignal the Subsurface Railway (SSR). We have covered the subject many times but, in essence, an initial contract made under PPP was cancelled for various reasons and the replacement contract for this was abandoned when it became clear that Bombardier’s theoretical claims for the product they were offering were simply not achievable in reality. Something that became obvious to London Underground when the company failed to reach even the most basic of milestones expected in the project’s lifetime. A third, and hopefully, final contract has since been agreed with Thales but the planned completion date has again slipped by a number of years. We have recently reported that completion was not expected until 2022, but now the details available here show that even this is not quite true. The project may not be fully complete until 2023 – so five years after the original deadline of 2018 for the second contract.

One could be forgiven for wondering why this particular delay should matter. Delayed signalling projects have been a firm feature of Underground times past. A nuisance, certainly, but signalling projects always get delayed. Indeed if they didn’t, Roger Ford of Modern Railways might struggle to fill up his monthly column inches as he has long provided consistent reporting on issues with resignalling schemes. Yet at the end of the day the railway carries on running and problems with signalling replacement schemes rarely hit the mainstream headlines. Indeed, from the public’s perspective it is the fact that the railway keeps getting resignalled that is apparently the problem. For they probably don’t think of a “signal failure” on a line where no replacement signalling work has been done for years as a “past investment failure” whilst they, and politicians, are quick to link signal failures to any current or recent signal replacement scheme.

The importance of resignalling

The most obvious reason to be concerned with the delay is that the SSR resignalling is important in that, combined with other works, it will dramatically increase the capacity of around 35-40% of the Underground network (the exact percentage of the network affected depends on whether you measure this by passengers, track mileage or another plausible metric). The SSR links a lot of vital areas of London – most notably many mainline termini – and is conveniently accessible in central London as it is generally located only just below the surface so tends not to involve using long escalators to access it. Therefore, a convenient SSR with a frequent service contributes to the economy of London in providing an efficient and functional means of transport in central London that can handle the required demand.

Critical path

Perhaps more importantly the SSR resignalling is on the critical path for investment – in non-project speak, that it potentially blocks the start of other work. A lot of the reason for this is budgetary. The signs are that, until the SSR resignalling is out the way, TfL do not have the billions of pounds they need to spend on the New Tube for London project. This is an absolutely vital project to replace the Piccadilly Line trains and signalling before it falls apart before moving on to the Bakerloo, Central and Waterloo & City Lines. As it stands this will likely be followed straightaway by the Northern Line (or possibly one half of a newly segregated Northern Line then the other) with the replacement of the Jubilee Line following that. If a potential delay to the NTfL were not bad enough, it is hard to envisage any extension of the Bakerloo, as planned, until it is clear that there will be sufficient new rolling stock to cover the entire extended line. In simple terms then, the SSR resignalling is not only holding up other resignalling schemes, it is also holding up progress on efforts to update rolling stock which is either already in dire need of replacement or will be so in the next decade. No wonder that the latest report, referring to the complete SSR upgrade project, states that:

The Programme is London Underground’s top priority

As is often the case, the “do nothing” option here really is not feasible. Apart from not getting the most out of expensive, newly acquired rolling stock, signalling components will inevitably deteriorate and forty years is generally regarded as the maximum desirable working life of electrically based signalling – a figure that continues to be borne out by evidence as signalling failures tend to rise in equipment of this age or older. Much of the signalling for the SSR is in fact already much older than forty years.

The other alternative of “replace like with like” signalling is also not very practical. For all the aggravation of getting a new system to work, at the end of the day any system is expected to last for around forty years so it is important to get it right. Furthermore, despite the eye-watering expense, a new technologically advanced signalling system would not cost that much more than basic traditional signalling and, combined with extra trains, is probably the easiest and cheapest way of adding capacity to the London Underground network in the places where increased capacity is needed.

A strategy for success this time?

Although the draft paper to be presented to the board may seem a little incoherent at times, it does seem to be based around a few key strategies intended to make absolutely sure that this resignalling project goes well or, in the worst case scenario, that TfL avoid the dreaded “Thermocline of Truth” and become aware of any problems in sufficient time to take action. These strategies appear to be:

  • Bring forward any benefits that can be made with the existing signalling
  • Minimise the project risk by critically examining what really needs to be included
  • Implement continual and rigorous assessment of progress
  • Follow a realistic phased implementation

We now look at these in detail.

Bringing forward benefits

The West Coast Main Line resignalling, carried out around 20 years ago, demonstrated the danger of putting all one’s eggs in one basket. The Jubilee Line Extension signalling in a similar manner really emphasised the need for a back-up plan. In the latter case London Underground had to rapidly install conventional signalling but, because this had not been envisaged, no allowance had been made in the tunnel design for suitable places to locate conventional signals with suitable sighting distances. As the SSR is not a new railway the back-up plan, if it exists, appears to be to keep maintaining the old system until such time as the new system is ready to take over.

It is clear that we can’t wait until the 2020s for further improvements to the SSR. Worse, we still don’t know with absolute certainty the timescale can be achieved – you never can be when planning to use technology that does not currently exist. What we are now starting to see are conventional upgrades to the SSR. We have already seen signalling changes to accommodate the S7 trains which are longer than the trains they replaced. Recently approval was given for a conventionally-signalled new crossover at King’s Cross to enable 20tph to be terminated there in the event of disruption.

According to the report:

Given the above commissioning dates, sufficient train availability and the necessary track works, the options being explored, subject to final confirmation, could include:

(a) Extension of “shoulder” peak period services on busy sections by up to an hour by 2017;
(b) Additional trains between Earl’s Court and Wimbledon from 2018; and
(c) Extending further east District line off-peak services which currently terminate at Tower Hill by 2018.

So the idea is to eke out the extra capacity that is still available on the current system. In simple terms this seems to mean that if something can be beneficially brought forward – even if it needs to be signalled the “old” way – then that should be done as a means of achieving a quick win.

Item (a) – Extending the shoulder peak – is an obvious thing to do and this is just part of a general trend on virtually all underground lines. Once ‘S’ stock becomes universal and ‘D’ stock is eliminated – which should happen by 2016 – then common driving characteristics of all trains should make this easier to do.

The only surprise about item (c) – extending District off-peak Tower Hill terminating trains eastward – is that it seems it will not happen sooner. Some of the previous off-peak Tower Hill terminators have already been extended, leaving just 3tph that currently terminate at Tower Hill. With no shortage of off-peak trains it is surprising that these will not be extended sooner to Barking to eliminate the gaps in the even intervals of trains.

More trains through Earl’s Court before the signalling upgrade

Item (b) is more interesting. As the report explains.

5.4.6. Work remains ongoing to ensure benefits are realised at the earliest point during roll-out, and opportunities to deliver benefits ahead of these dates are being explored. A number of infrastructure improvements will improve operational flexibility including the commissioning of additional track point work in Summer 2017 at Earl’s Court. This will reduce congestion, particularly for Wimbledon branch services.

The reference to Earl’s Court almost certainly refers to reinstating a scissors crossover to the west of the station and probably another crossover the east which were removed during a period of passenger downturn in the 20th Century. A quick look at Harsig’s website reveals the more flexible former layouts at Earl’s Court in 1936 and from 1957-1966. The diagram below has been taken from the ever invaluable Carto Metro and the likely reinstatement has been added in red.

Earl's Court Modified and amended

Earl’s Court District Line layout with possible reinstated track in red

The benefit of reinstating the sets of points marked in red is that it enables eastbound trains to Upminster and trains to Edgware Road from Wimbledon to cross before reaching Earl’s Court station. This gives the signalman (or the ATC computer-controlled supervisor) an additional option that may prevent one train being delayed by another. Basically the signalman (and the timetabler) gets two bites of the cherry. They can route the Wimbleware (Wimbledon – Edgware Road) service to platform 1 and also route a train from Richmond or Ealing Broadway to platform 2. Despite not being able to do both simultaneously, this is good because further conflict at the eastern side of the station is avoided. We can assume that often both trains will be stationary in their respective platforms at Earl’s Court at the same time due to the long dwell time there. The advantage is that trains can now depart from their respective platform as soon as they are ready to go as there will not be any conflicting movement ahead. Alternatively, if the signalman realises that allocating the trains to the platforms as described will mean one train will definitely be delayed on approach to Earl’s Court, he can route them into the platforms the other way round and give one priority over the other when departing – as happens now.

This extra flexibility on the eastbound lines mentioned above does not need to be matched on the westbound lines as the westbound lines already have a diveunder present to the east of the station. The only move westbound which is not fully flexible is from Edgware Road to Richmond or Ealing Broadway but as trains are very rarely routed to do this it hardly matters.

It is fairly probable that in reality these additional trains will terminate in one of the two bay platforms at High St Kensington (a revival of something that used to happen until recently). It is also the case that until recently there used to be an all-day Olympia – Earl’s Court – High St Kensington service which was usually little used in peak hours – especially the morning peak. This was originally intended to continue after the SSR resignalling which would suggest that there is actually a little potential spare capacity here.

The major downside of the plan to run additional services from Wimbledon to Earl’s Court is that people from the Wimbledon branch do not typically want to go to either Earl’s Court or High St Kensington. These additional trains will effectively take a lot of people and dump them somewhere they don’t want to be and leave them to continue their journey on already packed trains.

Minimising the project risk

It is clear that a lot of effort has gone into minimising project risk. An in-depth review to identify the minimum scope to deliver Programme benefits was carried out. One has the fear that some things removed will be regretted later and added back in at a later date.

Things mentioned by the report that are new are:

  • the Wimbledon branch of the District line, which is Network Rail managed, will now have an ATC signalling overlay rather than full resignalling (matching the established solution for the Richmond branch and consequently reducing design variants)
  • Piccadilly line trains will not be fitted with new signalling equipment
  • the section of the Piccadilly line between Rayners Lane Junction and Hanger Lane Junction will no longer be resignalled by the Programme
  • the signalling solution for the interoperable sections of the SubSurface Railway (SSR) and Piccadilly lines will include conventional signals to allow for continued operation of Piccadilly services on these sections
  • the number of planned track layout changes to be delivered by the Programme has been reduced following an extensive review
  • In a smart move the paper states:

    In parallel to the re-tendering of the ATC contract, LU engaged Thales to undertake a series of Interoperability Trials to demonstrate the feasibility of implementing Thales’ radio-based signalling system (CBTC) across the Northern and Jubilee lines as well as understanding the implications for implementing this across the four lines.

    One of the major differences between the Thales system planned for the SSR and actually in use on the Northern and Jubilee Lines is that the former uses radio to communicate signalling functions with the train and the latter two rely on loops of wire between the rails. Clearly near-continuous loops of wire risk getting damaged more than the occasional radio beacon but the loops are technically easier to install. By changing just one element of a known working system one can quickly establish, relatively painlessly, whether or not the new element is working satisfactorily. This means one can establish early on if there are any problems and, for example, if it were to cause the service on the Mill Hill East branch to be suspended it would not be a major disaster – and the public won’t relate it to the SSR resignalling project.

    Continual and rigorous assessment of progress

    Clearly learning from the previous curtailed Bombardier signalling project, a lot of planning has gone into integrating the Thales signal engineers with LU staff. There has also been a determined effort to get experienced LU staff onboard – basically those that worked with the Thales Jubilee and Northern Line upgrades. They will even post an LU site resident engineer at Thales’s base in Canada. The intention is clear – with Bombardier, LU engineers simply did not know what was happening. This was made worse by a lot of the development work (coincidently) also being carried out in Canada. This new approach is clearly intended to remove that scenario

    There is much in the document about governance. One of the simpler items is

    A “Crossrail-style” Programme Board report has also been introduced that summarises progress, performance and issues, which is issued each period.

    Another item, clearly learnt from the abortive Bombardier contract, is to monitor the Thales system where a similar installation has been commissioned recently. Expect to see a lot of LU engineers making trips to Hyderabad and showing a lot of interest in both the section of the metro already completed and the work in progress.

    A realistic phased (peak period) implementation

    Current service

    It is clear that Thales were not going to be bullied into committing to LU’s desire to try to maintain, or nearly maintain, the original timetable. The simple fact is that as the only realistic bidder they were in a strong position to resist this, regardless of any political pressure LU was under to deliver the final product in 2018, and there can be no doubt that Thales were fully aware that all the cards were in their hands. The result is inevitably an extended timetable as the table below shows.

    Proposed Timetable Change

    This does show a fairly well established pattern of increasing frequency as has happened recently on both the Victoria and Northern Lines. The first stage is to take the slack out the system caused by signalling and permanent way (track) improvements. After that you run a more intensive peak timetable, but only for around 90 minutes. You do this incrementally and with the SSR already running 27/28tph it does seem logical to go to 30tph and then to 32tph. Once confident that this will work you extend the hours of operation.

    Planned Peak Service

    It is notable that the service via Finchley Road is also being increased in increments. This will start at 24tph (currently around 22.5tph) before increasing to 26 and finally 28tph. One can understand this being the last phase of the peak hour upgrades and it is probable that increasing the number of trains terminating at Baker Street from 10tph to 12tph whilst having 16tph continuing onward to Aldgate is going to be quite a challenge.

    Off-peak enhancements

    What is probably the most interesting and most curious feature of the table is the final item, which simply refers to “off-peak enhancements”. Normally this would not get a mention, and in any case why delay with off-peak enhancements? This would be easy to pass by if it were not for earlier official comments about “Once completed most Circle line customers will see a train up to every 4 minutes instead of 10.”

    It is now we are going to get a bit speculative. For a start we will presume that this announcement would not be worth mentioning if it were not for it being something quite dramatic. Secondly, we are going to presume an off-peak service of 30tph. Such a frequency would not be at all unreasonable, though admittedly challenging to run all day with little slack in the system to recover from delays.

    We now get a bit more speculative and presume that there is no appetite to dramatically increase the off-peak service north of Baker Street. This seems to be borne out with only 4tph proposed for Watford Junction and no increases proposed for Amersham and no increase easily possible on the single track line to Chesham. Baker Street – Uxbridge off-peak services are expected to be provided at approximately 8tph.

    Now if you want to combine that service north of Baker St with a better more frequent service on the north side of the Circle Line you hit problems of integrating the train service. Meanwhile the terminating platforms at Baker Street are heavily underused. To get around this you could extend the Circle line from Edgware Road to Aldgate in the off peak. An example of the service that could be run if that were to happen is shown below.

    SSR off peak 2023

    Whilst we have had had no confirmation, official or unofficial, that this is what is planned, the consensus at LR Towers is that it represents the most plausible way forward that is consistent with recent announcements about the final off-peak service on the SSR.

    The idea would appear to be quite a clever one. What it does is enable the trains to run at almost peak period frequency in the central area and on the District Line branches whilst at the same time have a much reduced and more appropriate level of service north of Finchley Road – roughly identical to the service provided today.

    Passenger advantages and disadvantages

    The most obvious benefit to passengers is the frequent off-peak service in the central area. Additionally, as far as passengers are concerned, the old Circle Line is effectively reinstated as it will in future be possible to make any plausibly sensible journey between two Circle Line stations without having to change. Off-peak at least, no longer will passengers south of Edgware Road need to change there. The only main downside for passengers is possible confusion when on some platforms. A westbound passenger at King’s Cross could catch a Circle Line train to Hammersmith or one advertised as “via Victoria”. Eastbound, a Circle Line train may or may not terminate at Aldgate. Metropolitan passengers on the Uxbridge branch would marginally lose out. The service frequency would go down from 8tph to 7.5tph and with it goes the convenience of an hourly clockface timetable.

    Operational advantages and disadvantages

    Operationally the idea is a bit of a mixed bag of benefits and disadvantages. Decoupling the need for Amersham, Chesham and Watford Junction services from the workings of the Circle Line would certainly help off-peak working. These trains would terminate at Baker Street. Reducing the number of through workings south of Baker Street would reduce the conflicting movements at Baker Street Junction to 7.5tph with only the Aldgate-Uxbridge services crossing the path of another service. However these 7.5tph, which need to be timed to fit in with what is happening in real time on the Circle Line, need to share tracks with the Amersham, Chesham and Watford Junction trains which need to run to the timetable as much as is possible.

    In a similar way to the lines north of Baker Street, there may well be a conflict on the line from Gunnersbury to Richmond. The District Line trains will need to fit in with real-time running on the District and Circle Lines whilst London Overground, which shares the tracks, needs to do the utmost to stick to the timetable. Hanger Lane Junction, north of Ealing Common (where both junction and station current share tracks between the Piccadilly and District Lines) will also have the same issue.

    Extending the Circle service from Edgware Road does re-introduce a potential disadvantage. To get this to work the time taken to travel from Aldgate to Edgware Road and vice versa via Victoria must be 36 minutes to nicely slot in four minutes after the previous Circle Line train. The current running time is 35 minutes and with Automatic Train Control it should be relatively easy to arrange for this to happen without delaying other trains.

    Not enough off-peak S7 trains?

    A potential problem with any plan to run a very frequent off-peak service is whether there are enough trains when necessary maintenance is taken into account. In this case it is worse because extension of the off-peak Circle Line service would require more trains whilst the off-peak service is already roughly 94% of the peak service. The problem is that it is the S8 stock on the Metropolitan Line that is being cut back off-peak and the S7 stock that will be given yet more work. Whilst S7 stock can (and occasionally does) run on the Metropolitan Line, it would not be possible for the longer S8 stock to run on the other lines.

    Night Tube

    One final item in the SSR resignalling is omitted, and that is any proposed introduction date for the Night Tube on the SSR. As it is said that this cannot be run until the resignalling work is done then this would suggest a date of May 2021 – nearly six years after it will be implemented on five other deep tube lines.

    Under Scrutiny

    A few years ago the New Tube for London was seen as the most important project for London Underground, with the SSR resignalling regarded as just one of many run-of-the-mill projects that were in progress. It is a sign of how attitudes have changed that it is the SSR signalling that is hitting the headlines and is under scrutiny with the New Tube for London project scarcely getting a mention. If there is a saving grace in all this then it is that Crossrail should do a lot to relieve the problems of the Sub-Surface Railway and buy a few more years to get its signalling up and working but, if the population of London increases as predicted, come the early 2020s we are going to desperately need the extra capacity on the Subsurface Railway.

    London Underground, and indeed London itself, will be hoping that when it comes to SSR signalling there is no strike three.

Written by Pedantic of Purley