Something we have been wanting to do for some years is an overall look at the update plans for London Underground rolling stock (aka trains) and, independently, line resignalling. With all the uncertainty around this subject, the time never seemed right. It is probably time to face reality and accept that, in the foreseeable future, the time never will be right.
It was also difficult to decide the right time to start this story. Arbitrarily, we look at this from the start of the 21st century which, as it happens, nicely fits in with the formation of Transport for London in its current form. It is slightly easier to decide when to stop. Any rolling stock upgrade programme really needs to take into account all of the Underground lines which roughly takes us to the 2050s.
With some of rolling stock on the deep-level tube lines (Jubilee, Northern, and Victoria) we really have no idea of what their actual life expectancy will eventually turn out to be – and we rather strongly suspect no-one else does either. For others (Sub-surface Railway) upgrading or refurbishing the existing stock to keep it fit for purpose is taking place now, and for some other deep-level lines (Central and Piccadilly) there are known plans to replace the signalling in future even if the timescale is uncertain.
The Importance of Looking to the Future
It might seem that looking twenty or thirty years ahead is unnecessary when the future is hard to predict and many of those alive today will be dead and so past caring about what may happen beyond our lifetime. However, it is a bit like buying a leasehold property. As time gets closer to the leasehold expiry date the value of property goes down. Even if you are not immediately worried about the property diminishing in value, the people behind you will be and this will need to be taken into account when considering personal future plans and how much the property is worth should you wish to sell it.
In the rail industry, financial planners need to identify future liabilities and that will, to some extent, determine what money is available to be spent today on infrastructure schemes. From TfL’s perspective, it is pointless starting to build the Bakerloo Line Extension with your own money if that money may be needed elsewhere just to prop up the existing system. The spectre of a half-completed project being abandoned is something one generally wishes to avoid.
A classic example affecting the Mayor of London and the Government would be Crossrail 2. If you know you are going to have a massive budgetary hole in the 2040s and Crossrail 2 is predicted to take over 10 years just to construct it (as it is), you are going to seriously hesitate about progressing with the project if the money is not going to be available later on in the project due to the known need to spend huge amounts of money elsewhere.
A potential further problem lies with future investment in London. If potential investors can see there is no credible strategy for keeping the main-line and Underground trains running in the future, they are not going to invest in a city they perceive as falling apart.
We will look at recent past and planned future investment on roughly a line-by-line basis.
Four Lines Modernisation – Trains
When it comes to the London Underground, possibly one of the best decisions made this century was when the first TfL Commissioner, Bob Kiley, fought hard for all the sub-surface lines rolling stock to be replaced as a single package. This for the first time enabled near-homogeneity on the sub-surface railway (Circle, District, Hammersmith & City, and Metropolitan lines). It was a golden opportunity to replace the following rolling stock:
- A60/62 – long overdue replacement.
- C69/77 – functional but disliked and dated.
- D78 – In many ways this had a potential for a longer life, but its single-leaf doors caused significant delays on an Underground line that was getting busier with each passing year.
Building of the replacement stock, known as S-stock, commenced in 2008 and so big was the order of 192 trains that construction didn’t finish until 2017.
These trains are longer and much more reliable than the trains they replaced. They still feel quite new despite having been around for more than 15 years. From the perspective of the passenger, the non-diffused fluorescent tubes are the only thing that really makes the rolling stock look dated. That said, the passenger information screens could probably benefit from a wholesale replacement more in the style of the Elizabeth line or Thameslink trains, which seems to be the modern way of doing things now.
Crucially S-stock is air-cooled. Air-cooling is effectively as good as air-conditioning in the dry air environment of the sub-surface lines and means that the trains should not be unbearably hot towards the end of their working life.
One would expect, with a decent mid-life refurbishment, that these trains would be suitable to remain in service for around forty to fifty years, like its A-stock predecessor, meaning we are looking around 2060 for a replacement and sufficiently far into the future not to worry about it. Fifty years may seem unduly optimistic, or pessimistic depending on how you look at it, but this is effectively modern well-built main-line stock not given the harsh hammering that deep-level tube trains in London are subject to due to the sharp curves prevalent on lines built to avoid tunnelling under private property wherever possible.
Four Lines Modernisation – Signalling
In many ways Four Lines Modernisation signalling is the London Underground 21st Century horror show and one loses track of just how delayed this resignalling project is. Originally destined to be complete around 2012-2014, two contracts got cancelled due to a lack of progress being made, and the current one looks like being finished almost a decade later than planned. It would have seemed that the final stage was due to be in December 2027 when the delta junction at Aldgate is was due to be relaid so as to be optimised for trains of S stock length. However, the latest report on renewals from TfL suggests this would be complete ‘by 2027/28’ which may mean there could be a further delay. When complete, this will finally permit the long-planned 32tph timetable for central London sub-surface lines that was promised years ago.
At one stage pre-Covid, progress on this was seen as urgent and management attempted to apply pressure to avoid delays. The delays on the Elizabeth line signalling took the focus away from this project (at least in this case there was an existing, working signalling system in place) and the reduction of passenger numbers post-Covid took away the urgency. Nowadays it seems to be a case of ‘go slowly if you must – but get it right’.
It should be the case that, once fully installed on the remaining sections of track on which it the signalling project has not been cancelled, it should be good for another 40 or 50 years and maybe longer. The use of radio frequency and the lack of lineside signals should help its longevity. If it does ever get replaced one would like to think that ETCS/ERTMS (European Train Control System/European Rail Traffic Management System) would be sufficiently mature and advanced by the middle of the 21st Century which would mean a standardised replacement signalling system rather than being forced to specify a bespoke one.
Victoria line 2009 stock and signalling
At the time of its introduction, the 2009 stock Victoria line trains was seen as the ultimate in deep-level tube stock technology in London. The 36tph service made possible with the complementary Siemens advanced automatic train system replaced the increasingly unreliable 1967 stock operating a 27tph timetable. The acceleration ability of the new trains was in contrast to that of the old and the seats were much more comfortable.
Today, the 36tph on the Victoria line is still impressive but the trains are already showing their age due to intensive use. The same could be said for the entire line that seemed so modern when opened by Queen Elizabeth II in 1969 and it now appears very dated and unambitious in design, with relatively little having been spent below ground level on stations (Victoria is a notable exception) since it was fully opened to Brixton in 1971.
The Victoria is now officially the hottest line on the Underground which is quite worrying when you consider that, when it first opened, they had a problem with signal faults due to damp, cold air. It shows how much it has heated up in less than less than 60 years and it would appear to be the case that it will only get hotter still.
The heat is bad enough in the platforms but at times it is almost unbearable in the trains which will almost certainly be the last ever built for London Underground without air-cooling. The worry is that, without sufficient remedial measures elsewhere, the trains will just become too hot to be useable forcing premature replacement. Otherwise, one would hope that they would last for the usual forty years which is generally the life-expectancy of deep-level Tube stock.
Another factor which may prompt premature replacement is the fact that, of all the Tube lines longer than the Waterloo & City line, driverless (unattended operation) is probably the most feasible on the Victoria line. With its straight platforms aiding installation of platform edge doors, its frequency and being entirely underground (which eliminates problems with wet rails and trespassers), it is the ideal candidate.
In the intervening decades, technology would have moved on and even today new metro systems being built for trains with drivers are getting very rare indeed, with both Sydney and Paris (amongst others) committed to making all new lines driverless. As regards the issue of unions and the need for drivers, it may well be that demographic changes and a demand for a universal shorter working week puts a greater pressure in the future to eliminate jobs that are not strictly necessary.
Heat and driverless trains lead to consideration of early replacement of trains, and possibly signalling, on the Victoria line. Thirty years would be a reasonable minimum working life of a deep-level tube train, with forty years being the expected length of time in use before replacement, so it looks like somewhere between 2040 and 2050 serious money will need to be spent on the Victoria line including its rolling stock.
The Stagnant Years – or were they?
After the introduction of the Victoria line 2009 stock there would appear to have been very little happening, infrastructure-wise, on London Underground for more than a decade. It is true no new trains appeared, but the Victoria Station Upgrade was completed and Bank Station Capacity Upgrade was well on its way to transforming a critical station. Following on from the Victoria line upgrade was the conversion of the Jubilee line to Automatic Train Operation (ATO) followed in turn by the Northern line.
More importantly to our story, a lot of development work was carried out on a programme referred to by Boris Johnson as ‘New Tube for London’. It was extremely ambitious including objectives beyond the state of the technology of the time.
Whilst it may be true that London Underground was not keeping up with the need to renew during this decade, this certainly was not true of TfL in general. The Elizabeth line (formerly Crossrail) was the biggest single transport construction project in London for years – if not ever. And this all happened at a time when the successful 2012 Olympic Games were a major distraction.
Finally New Investment: New Piccadilly line trains…
For quite some years it was clear what the next big investment needed to be. It was clearly an upgrade of the Piccadilly line with new trains and new signalling. The signalling was old but could more-or-less go on forever, although it would become more unreliable and limit frequencies to 27tph. With the implementation of Four Lines Resignalling there should be a huge stockpile of spare parts becoming available enabling an old but simple and fairly reliable signalling system to continue indefinitely. The same could not be said for the trains.
The existing trains, 1973 stock, were the first of a new generation of rolling stock in a digital age. These inevitably became obsolescent very quickly as technology marched on. They also suffered a problem with quality of the wiring used within the train which degraded over time. The stock is coming up to its fifty years anniversary in service and this is well past its design life and ‘best before’ date.
Out of the New Tube for London project came a more realistic plan for commissioning new trains. The Piccadilly line was prioritised due to condition of the existing trains – most notably with the aforementioned wiring. The implementation of new stock was intended to be followed shortly afterwards by resignalling the Piccadilly line thus enabling 36tph to be run, subject to some other critical upgrades. Given that the Piccadilly line cannot quite currently run 24tph in the Monday-Friday peak periods and that the new trains would be slightly longer, this would be a huge capacity increased on a heavily-used line.
It seemed inevitable that the new trains would be approved because it was becoming increasingly expensive and difficult to keep the old trains running, meaning that buying new trains was beginning to look line the cheapest option with the ‘do nothing’ scenario both risky and expensive.
Siemens won the contract to build 94 new trains at a cost of £1.5 billion. This works out at £16 million per train. So far, they have not disappointed in innovation. The trains will be the first deep-level tube trains running on existing lines in London with air-cooling. Space was made available for the cooling units by an innovative design for trains (but common on trams abroad) of having an odd number of carriages with the odd numbers having wheels and the even numbers being wheelless and suspended from the adjacent carriages. The absence of wheels from the even number carriages means there is space to install the air-cooling below the floor of the train. A reduction in the number of wheels overall also makes the train lighter than it would otherwise be. It weighs roughly the same as an existing train despite being longer and containing air-cooling equipment. This also reduces track wear compared with what it would otherwise be. Along with more efficient state-of-the-art motors, fewer wheels helps contribute to the reduction in energy needed by the train, which in turn should help reduce the rate at which the tube tunnel heats up.
In addition to all the above innovations, Siemens have managed to implement open gangways between carriages – a first for deep-level London-sized tube trains. Almost at a stroke, Siemens are going to introduce rolling stock which will make all other deep-tube trains look very dated. One could argue that the last rolling stock on a deep-level tube in London which had the same level of innovation was the 1938 stock, which pioneered the locating of all equipment below the floor (more strictly the solebar, which is the outer longitudinal beam of the train chassis, to which several components and supporting cross members are attached), making almost all of the space above floor level available for passengers for the first time.
The new Piccadilly line trains were expected to start entering service later this year (2025) but this has been delayed – apparently because of some incompatibility between the trains and the tunnels which we are told was not foreseeable. Until recently the intention was to fully replace the Piccadilly line 1973 Tube stock by the end of 2027 but 2028 is now looking more realistic.
… But not new signalling or a critical station enhancement
By not introducing new signalling on the Piccadilly line, it has been calculated that a maximum of 27tph could be run with the new trains. This would therefore be a slight improvement of around 12% over the existing frequency.
However, even with new signalling, another limitation appears. Holborn station will be at passenger capacity with 27tph on the Piccadilly line. Worse still, to go beyond 27tph it is calculated that a new substation in central London will be needed and the current plans would see this incorporated into an enhanced Holborn station. The station upgrade was consulted on in 2017 when the estimated cost was put at £200 million. We suspect, with an increase in building costs and general inflation, it would be more like £500 million at current day prices.
Given the other projects that are pressing, we suspect that no money can be found at present beyond that required for the new trains, and that the new signalling and Holborn station upgrade will have to wait for some time in the future when money is more freely available.
Replacement trains for the Bakerloo line
If one thing has been currently giving overriding concern when it comes to capital investment has been replacement of the 1972 stock on the Bakerloo line. The Chancellor’s spending budget for TfL including the all-important commitment to provide the money for the next four years means that TfL should be at last in a more predictable position. Unfortunately, the amount allocated does not seem to be sufficient to enable new trains for the Bakerloo line to be ordered, when other essential projects are taken into account.
The 1972 stock on the Bakerloo line was actually introduced in 1972 (a rare thing for the year of manufacture to coincide with the year of introduction) meaning they are already 53 years old. Worse still, they are based on the 1967 Tube rolling stock built for the Victoria line as the trains were needed urgently. These were the last of the pre-digital-era trains and as such they could theoretically last indefinitely by being patched up and repaired, but the cost of doing so would continue to escalate. Already, the service on the Bakerloo line has had to be reduced to 20tph from 22tph because of the need to allow for more trains in the workshop for maintenance.
As reported by IanVisits, spare parts from existing Piccadilly line stock common to Bakerloo stock will be salvaged to help maintain Bakerloo trains. It comes to something when in 2026 or 2027 spare parts are taken from 1973 stock to help keep 1972 stock running.
Currently there are 36 1972 stock trains which in normal circumstances could provide 22tph. Ordering 36 trains of 2024 stock would cost roughly £600 million but TfL may order fewer or more. Original plans were to provide a 27tph service but that assumes that the existing signalling on the Bakerloo line could handle that many trains. £600 million for the new trains would now seem to be at least possible, if challenging, using some of the money the government has allocated TfL over the next four years. With government finances tight, probably only an extreme optimist would expect the government to suddenly produce yet more money for these trains.
Unfortunately nowadays, even more than before, new underground trains need to be accompanied by considerable extra expense to other infrastucture to maximise the benefit of the new trains and ensure they can operate safely. We have absolutely no idea of how much this would cost, other than to suspect it is not cheap.
It may be that the government is keen for the Mayor to find extra funds to cover the shortfall by fare rises, a higher council tax precept, or another way of raising the money. In a way, the Mayor and TfL may be forced to take unpalatable measures, because they have long stated that continuing indefinitely with the current stock is not sensible from either a reliability or financial perspective. The tipping point has supposedly been reached where the sensible long-term cost-effective solution is to buy new rolling stock, almost matching that which has been designed and tested and due to see service on the Piccadilly line in the next year or so.
A further complication with Bakerloo line stock is that there is an option to buy further trains from Siemens to follow on from the Piccadilly line order which expires in late 2026. One wonders if TfL will order the trains anyway and somehow find the money, whatever it takes, to pay for them when payment becomes due.
The Central line Horror Story
Bad as the Bakerloo line saga is, at least one feels there is hope that the trains will be ordered within the next couple of years and trickle into service in the next decade. Despite the Central line trains (1992 stock) being around 20 years newer, in many ways the stock was worse than the Bakerloo line. The DC motors featured electric technology (Gate Turn-Off thyristors) that were about to be superseded by AC motors with electronic rectifiers. The latter would have been much better as they require less maintenance and, more critically, would reduce the number of trains out of service. The DC motors were also unreliable and, at one stage in 2003, the Central line was closed or partially closed for around six weeks whilst urgent remedial action was taken as a consequence of a traction motor falling onto the track.
Coupled with the bespoke and obsolescent automatic train operation signalling system, and the seats that never were comfortable and are now positively uncomfortable, the 1992 stock has not aged well.
With new trains seemingly forever the wrong side of some transport planning event horizon, there emerged a somewhat bizarre and organic policy. It had its roots in a 2017 project for Bombardier to supply replacement motors for the Central line trains. It was clearly felt that TfL could not wait for the new trains which, realistically, were not going to arrive until the 2030s. The project was given the title of Central Line Improvement Programme (CLIP) though, at that stage, Central Line Motor Replacement Programme would be a more accurate description even if it didn’t result in a neat acronym.
As the expected date for replacement trains advanced it was recognised that other components were going to need replacing as well. It then followed that it made sense to do it at the same time as the motor replacement. Once the need to strip the carriage down became inevitable it then made more sense to add further updates to the programme rather than risk having to replace these at a later date.
In this way, a ‘simple’ replacement of old and unreliable DC motors that represented a past era of motor development would be replaced by modern AC motors that could work on 4-rail DC traction supplies, developed into a programme where just about everything that could be replaced on the rolling stock would be replaced.
To add to the ambitiousness of the project, it was decided to carry out the work in-house and become the largest in-house project ever undertaken on rolling stock. To this end, new workshops would have to be constructed at Acton and staff recruited. The feeling was that the previous methodology of sending the trains to various far-flung destinations including places in Scotland was neither appropriate nor offering value for money.
The eventual CLIP project, valued at £500 million seemed remarkably cheap compared to the cost of new trains. The biggest downside was the trains would not have air-cooling, air-conditioning, or anything to reduce the heat. With tunnels getting hotter and climate change, the worry is that, before the trains have otherwise come to the end of their useful life, the trains will have to be prematurely retired or the heat will become unbearable and passengers would be reluctant to travel by these trains.
Progress on CLIP has been described as ‘glacial’. Not surprisingly, what theoretically works when planned shows shortcomings when the plan hits reality. As of February 2025, only two trains (out of 85) were complete, with a further three trains due to be complete by January 2026. After that, a seemingly heroic push will see 25% of trains complete by November 2027 – so 16 further trains completed in 23 months. Then, miraculously, in the space of just over two years, a further 64 trains will be completed, enabling the ‘end of 2029’ deadline to be met. The argument goes that production will speed up rapidly once the initial teething troubles are sorted out.
A slight concern here is that TfL refers to ‘completed’ trains rather than ‘in service’ – there may be quite a lag between a train being ‘complete’ and ready to leave the Acton workshops, and being delivered to the Central line depot and having accumulated sufficient out-of-service fault-free running to enable passengers to travel on it.
What is more worrying is that one presumes that TfL intend to get at least 10 years useful life out of the CLIP-refurbished trains, which takes us at least to the end of 2039. This does appear to be confirmed in an answer to a question posed at Mayor’s Question Time.
By that time, the trains will be approaching being 50 years old. Even with a full mid-life refurbishment this seems to extending their desirable service life to the limit. But now replacement in the 2040s will probably clash with the desire to replace the Victoria line stock if something needs to be done about the heat on that line – and this clash of replacement timescales is going to get worse.
Central line Signalling Incremental upgrade Project (CSIP)
The chances are you have never heard of the Central line Signalling Incremental upgrade Project – and we hadn’t either. Below is a quote of the entirety of what we know about this project.
Central line Signalling Incremental upgrade Project (CSIP) within Signals – the Programme is undertaking a 15–20-year incremental upgrade of the current Central line signalling and control system. This will be replaced with a modern, digital Communications Based Train Control system.
We have to say it makes some sense. The Central line signalling is already obsolescent. It was implemented when Automatic Train Operation (ATO) was not nearly advanced as it is now. To give but one example, it actually has a separate train describer system that keeps track of the trains, their train number, and where they are heading. This interfaces with the ATO system. Today, no-one would dream of designing an ATO system that couldn’t drive the customer information screens or tell the controllers what train was where and where it was planned to terminate.
Clearly when it becomes time to replace Central line stock, one doesn’t want to have stock customised to run on the existing Central line signalling. It could be extremely beneficial to have a modern type of signalling already installed. Furthermore, it is uncertain how long the existing Central line signalling system could keep going. In particular one must worry about availability of spare parts and having the engineers with specialist knowledge to service it.
A phased approach for a new signalling system would seem to be the best approach. As parts are predicted to wear out or otherwise fail, portions could be switched over to the new system so avoiding catastrophic failure. And given the tendency of state-of-the-art signalling projects to overrun by many years, starting planning for this now would seem to be a sensible approach. One’s only concern is that signalling is not cheap and this is going to involve a huge sum of money to be spent from TfL’s internal budget – albeit spread over many years.
Waterloo & City line
Little is being said about when one can expect Waterloo & City line stock to be replaced. In the ‘New Tube for London’ plan, which changed considerably whilst the project had that name, Waterloo & City line stock at various stages of the plan occupied all four positions in the sequence of the four lines affected to receive the replacement rolling stock.
With only Monday-Friday working, peak flows in just one direction and extremely poor economics for the line which discourages investment, one can take a guess and suspect that some elements of CLIP will eventually be put together to partially upgrade the Waterloo & City line stock. This would enable the age-old tradition of running ancient rolling stock on the Waterloo & City line to continue. Given the total current allocation of trains is just five 4-car units, replacement of Waterloo & City line trains with an equivalent number (or maybe one extra) cannot be regarded as a major issue.
Northern line
So far, we have been mainly discussing replacing tube trains, with some mentions of upgrading signalling. The only feelers we have for replacement dates for Northern line 1995 Stock is that, in the past, the date of 2040 has been suggested in TfL papers. This would mean that the stock would be in service for 42 years. This figure is slightly on the high side (40 years is generally regarded as a desirable maximum for deep-level tube stock) but not unreasonable.
If it were a simple case of replacing the rolling stock, this would not be a big deal, although the sheer number of trains (currently 106 6-car trains) would make it a big project. What makes this replacement a particularly big deal is the long-standing desire to increase capacity in central London by effectively running the Northern line as two separate lines, which would be High Barnet – Morden and Edgware – Battersea Power Station.
To make best use of the opportunity presented, you would need to run many more trains in order to get a service of at least 32tph in central London and all the way down to Morden. More trains mean more stabling sidings and more workshop space. Finding stabling sidings for the extra trains, should this upgrade take place, is going to be extremely challenging and would probably call for some inventive thinking.
The level of the challenge was put into perspective by a comment on the District Dave website. To fully upgrade the Northern line to run in two sections with significantly more frequent trains would require sidings to match the number currently at Morden depot. And Morden depot is currently the largest depot on the Underground serving just one line. For that number of trains it wouldn’t be sufficient just to build more sidings. The trains would have to be serviced. So, effectively, you would need the equivalent of a new depot the size of the existing Morden depot to provide facilities for the new trains to be maintained. To a find a space for such a depot or depots would be a considerable challenge and the cost would also be significant – for the land if nothing else.
Camden Town station would need to be enhanced to aid interchange at this critical station. There was a scheme for this but it had to be postponed indefinitely due to budgetary crises. The good news is that TfL have already acquired the land for this – the former Hawley Infants school. Like Holborn station enhancement we suspect that the original £200m estimated cost would probably end up being more like £500 million at today’s prices.
To largely separate the Northern line would be a golden opportunity and the ideal time to do it only comes around roughly once every 40 years. It would be a shame to miss the opportunity. A further factor here is that, eventually, Euston will see HS2 trains and, in all probability, work on Crossrail 2 will not have even commenced. Increasing Northern line capacity is the one thing that can relatively easily be done to help disperse HS2 passengers. As both central London sections of the Northern line serve Euston, a modest increase in Northern line frequency has a doubling effect.
Jubilee line
In many ways, the Jubilee line’s 1996 stock is practically the twin of the Northern line’s 1995 stock. Both were originally six car trains. Not surprisingly, the Jubilee stock will need replacing at around the same time as the Northern line stock.
In the case of the Jubilee line there are only 63 trains. Four of these trains were subsequently ordered as seven car trains and came into service in 2007 but the previously-existing ones had a carriage added. Like the Northern line, the Jubilee line will probably need more trains as there was a plan to increase the service from 32tph to 36tph which was abandoned because of the lack of rolling stock when a proposed purchase of extra trains was cancelled. Therefore, the Jubilee line will probably need an increase in the number of trains though not nearly as big an increase as on the Northern line.
It would be surprising if, in the late 2030s, there wasn’t a plan to replace Jubilee and Northern line stock with a common new stock – albeit of different lengths for the two lines.
The Crunch
What is becoming apparent is that, by our calculations, most of the deep-level tube stock will need replacing in the 2040s. By then, only the Piccadilly line and the Bakerloo line, hopefully, will have stock that could in any way be described as modern. Assuming, unrealistically, that money was no object, it seems that the only way this could happen is if your different train builders building new stock for different lines.
Together with signalling upgrades and some related station reconstruction, this would appear to be a huge budgetary spike that no-one is even thinking about yet. It is understandable. When the alligators are snapping at your feet it is hard to focus of the main objective of draining the swamp. Yet one day this must be faced and it is a problem far too large for any London Mayor to deal with on their own.
An obvious solution is to lease the rolling stock, but that probably won’t work because normally no lessor is going to take over leasing custom rolling stock designed for one Underground line and of virtually no value elsewhere. An exception to this is the Northern line stock which is leased. However this is exceptional and is probably related to the fact that Alstom (who built the trains) also had a long-term contract to maintain them rather than the normal situation of the trains being maintained by London Underground themselves.
Politically, any huge investment in London will have to have something like proportional investment in other parts of the country. One just hopes that the government of the day starts planning for this in the 2030s.
All photos have been taken from the relevant article on Wikipedia and sometimes cropped. Details of copyright restrictions can be found there.
Thanks to various members of the London Reconnections team who have provided a considerable amount of background information to make this article possible.
Free Content Costs Money
We continue to be a free content site. However, we do have a Patreon for donations, which goes solely to paying for the not inconsiderable server costs – given how many reads and views the site receives annually.
Interesting point regarding depot space for the Northern Line split. Whilst there might be some potential for some additional stabling space at Morden, Highgate, East Finchley and Mill Hill East, it doesn’t look as if this would be sufficient to provide the space needed, would not provide additional maintenance shed space and in many cases are a bit disconnected from the existing depot and sidings and would result in the loss of car parking unless decks were provided over the new stabling.
Along a similar line I presume a new depot will also be required were the Bakerloo line to be extended to Hayes? Have there been any suggestions as to where this might be?
Overall it seems a shame that a similar approach couldn’t have been taken to the deep level tube rolling stock replacement as to that for the sub-surface lines. I appreciate it would be incredibly expensive but a huge order for the replacement of the Picadilly, Bakerloo plus extension, Northern, Central, Waterloo and City and Jubilee over a number of years would be useful in bringing the overall unit cost down, providing standardisation across the network and facilitating frequency increases and extensions.
The land to the west of High Barnet station would have been a possibility for extra depot space but has been designated for housing. There are no obvious easy answers for more Northern Line depot space.
Great article and good to demonstrate the “crunch” in 2040 – which presumably for reasons of capacity and resourcing TfL would want to avoid.
Given Central Line trains will have had ( if things go to plan…) a fairly substantial refurb in late 2020s, might it make sense for another set of rolling stock to be replaced ahead of them, after Bakerloo, in late 2030s (E.g. Northern). Of course wider politics of tfl financing may make this impossible.
@ Biggerisbetter, I suspect that the platform screens on the Jubilee line might prevent them using the New Tube for London trains, as I doubt the doors line up.
On the subject of lack of stabling space on the Northern line (Edgware branch), obviously the only logical answer is to extend the line at Southern end beyond Clapham Junction and take over one of the SWR suburban lines that has stabling space – maybe the Feltham Loop? You could even dig a tunnel to Heathrow to provide even more options to get to and from the airport… Sorry, I’ll put my crayons away.
@peewee Re Jubilee platform screens–certainly a consideration and I suspect you are right they won’t line up for New Tube.
Likely needing to be a different train model (shared with Northern probably), maybe taking some concepts from New Tube and sharing some components, but ensuring the door placement matches current rolling stock.
If New Tube was to be used in current form for Jubilee, it would practically mean the launch of new trains would need to be either “big bang” (i.e. complete fleet changeover all at once with a short outage of a few weeks to allow the screens to be replaced with new spacings as quickly as possible–likely they would need to be renewed anyway for mechanical reasons, perhaps more than once from now). Could run mixed fleet in testing; maybe in partial service avoiding the central section if the pathing would permit (I’ll presume that having a central section shuttle with old trains would not fit well with pathing for 20+ TPH)
or
there would need to be a temporary removal of screens while the fleet was changing. This would be significant change in operations and safety for a period of some months; unclear if a safety could be mitigated through the use of additional platform dispatchers, capacity controls at some stations to avoid platform overcrowding, etc.
That likely all nets out to Jubilee being either last to change if New Tube was proposed (maybe further right on PoP’s chart even); I think PoP’ chart would need to presume that Northern/Jubilee would be a different train model to New Tube (lessons learned, common design principles, some parts and software commonality maybe, but different).
@ Biggerisbetter, in an early TfL consultation on extending the Bakerloo line beyond Lewisham, an alternative to Hayes was to extend towards Hither Green and the sidings there. Is that still a possibility? (And if so, and this is a crayons moment, a further southern extension to Bromley North, might also be possible.)
It is an emerging truism that heavily computerised equipment can have a shorter service life than older electro-mechanical equipment, unless extraordinary efforts are undertaken early on to ensure longevity of supply of:
* circuit boards and key components like processors, ASICs, FPGA etc.
* software development supply chain (operating systems used to develop software, tooling to implement the build etc.). Use of Open source can help, failing that, it’s vital that source code remains available via an licensing or escrow process.
* And for FPGA (if used): the tools to custom program the FPGA from their manufactured state.
* ASICs can be a real concern: “application specific IC” i.e. bespoke short order productions, potentially proprietary to train/component manufacturer.
Maintaining all of this is necessary to ensure that certifiable versions of software can continue to be available. Change some of the main components? Need to retest in a test rig at least, maybe mileage tests out of service as well. In some cases may need to rebuild the software from source code to account for processor etc. change
Ideally the train software should be open to third party scrutiny (to ensure manufacturer has not put in “crippling code” to disable trains in case of contractual disputes, as we have already seen with some locomotives in central Europe), but also will be willing to supply source code later in life of product and novate licensing to allow TfL or nominated 3rd party to take over support. (And code in escrow in unlikely event manufacturer ceases trading).
More than once I have seen in my working life in IT examples of computerised systems becoming reliant on second hand spare parts purchased via specialists or (horror of horrors) eBay, sometimes after just 20 years, due to OEM ceasing supply. Computer companies move on and cease supply and support for key items!
At least we have (and can maintain) “heritage” industry skills to rebuild electro-mechanical items
On the subject of platform edge doors, I omitted this from the article as well as a lot of other things as it was getting rather long. It is certainly an issue and one that is recognised.
The fundamental issue seems to be: do you keep the existing door spacing (forever effectively) or do you keep up with the times and modern thinking (wider door doors, no single doors, Piccadilly style arrangement of train cars)? I strongly feel you should move on and take the consequences. It is certainly not going to be pain free.
It seems logical to have considerable commonality of Northern and Jubilee line stock. The problem is that if one train maker does it all then it will take years to fully introduce it. If you have a second train maker involved then the economics of scale and benefits of commonality would disappear. If you bring train construction forward then you don’t maximise use of the first trains to be decommissioned. If you leave it too late then you have the situation where you are trying to keep the final trains serviceable when it is becoming extremely challenging.
MilesT makes a very good point. One partial solution is to enter into a design, build and maintain contract like has been done with Alstom on the Northern line. This should guarantee keeping the trains in service but arguably all you are doing is passing the problem onto someone else.
An alternative to searching round museums, eBay etc for parts is to have their functionality clearly specified and documented so it can be replace with a more modern replica. But the critical thing is to recognise the issue from the outset and ensure that, one way or another, the spares can be sourced throughout the lifetime of the trains.
I really don’t know the solution to the depot problem. I do sometimes wonder if it would be possible to double-deck some of the depots. Golders Green would seem to lend itself to this with a lower deck below current ground level that had separate junctions located off the running tunnels to physically isolate it and make it independent from the existing depot. Mind you the disruption building it and earth moving required would be enormous. But at least it is a possible solution when everything else one thinks about seems to be ruled out.
Something not mentioned here was the Central Line – Signalling Life Extension project which was running in front of everything else with ambition of remaking the most obsolete aspects existing signalling in more modern form. Effectively trying to keep the whole lot limping along for another 20 years or so, presumably until it was overtaken by the CSIP programme.
I say ‘was’ deliberately. The last intelligence I had suggested that this has now been canned, presumably hoping that the existing kit can be eased ever onwards with the help of decommissioned equipment from elsewhere in the world.
The Programme Feasibility Study for Northern Line Upgrade 2 was reviewed in the August 2016 Underground News. That showed plans for a new maintenance depot at Highgate sidings for 4 trains, with 6 more trains from Morden depot, and 13 in new sidings on Finchley Central former goods yard. High Barnet would fit another 3, whilst 2 trains stabled in platforms there, together with 1 in Morden platforms. On the Edgware branch they planned 3 more from Golders Green and 10 more from Edgware, with 2 also stabled in Edgware platforms. New trains need much less servicing than current fleets, so less depot space. Maybe a 50% fleet increase could be maintained in the current space? There was discussion on new Picc trains whether Siemens should run one or two production lines, so delivery rate can be varied if it can be afforded.
This is really interesting stuff, and reading it it’s never been clearer to me that replacing LU trains is a rolling requirement, not a stop-start one.
That being said, I think it’s useful to add some nuance around the talk of “money” and “investment” – because the core business case behind replacing these trains is that it reduces the future operating (ie maintenance) cost. In theory, the cash required to make this investment should be much easier to get hold of than the cash needed to pay for operating costs, because basically the payback (ROI) is guaranteed. New trains will cost less to operate and easily pay back over their lifetime. Unfortunately, as many public sector organisations are finding, the whole national financial set-up is sufficiently broken that the cash to pay for what should be solid long-term benefit isn’t forthcoming. The PFI/PPP schemes of 25 years ago tried to address this problem but failed. LU is stuck in something of an investment trap – without the funds to make a step change reduction in operating costs, it doesn’t have the ongoing spare funds to continue to keep those costs down through more asset replacement, or to invest in expansion/enhancement.
So, onto CLIP – from the sound of it this is a classic case of in-sourcing optimism bias. It happens when organisations have outsourced things but been unsatisfied with the outcomes and costs, so internal people come forward with a persuasive business case for in-sourcing. And then it gets approved but they struggle to deliver, because whatever it is turns out to be harder than they thought. This isn’t in the end about whether in-sourcing or out-sourcing is better – my conclusion is that the most efficient delivery comes from a stable team conducting an ongoing pipeline of work, whoever they work for.
Which brings me back to the point I made at the start – the buy-maintain-overhaul-refurb-replace cycle is a rolling requirement for LU, and having a team conducting a rolling programme will offer the best value for money, whoever they technically work for.
Clearly no electronics are going to perform for a 50 yr life. Greater use of industry standard interfaces will allow modules to be swapped out and upgraded.
The trains will be stripped out and rebuilt 2 or more times so control systems can be included along with running gear and interiors.
Stabling can be managed by judicious parking in tunnels. According to maintenance schedules you could isolate different loops that are cleared.
Alternately rather than fanciful double decking you could add some battery locomotives and move coupled serviced units around maintenance trains as needed. Release into service in the morning from different starting stations until road cleared.
Aleks,
Storing trains in tunnels runs the very real risk that the train won’t start after an overnight shutdown and will mean no service. It is similar to the risk of a depot entrance/exit being blocked. To offset the latter risk, depots (and even sidings) are double ended where possible – even relatively small stabling sidings such as Gidea Park have two useable exits whenever possible.
Trains are already stored in tunnels to almost the maximum extent considered safely possible. For example, you already have three stored at Walthamstow overnight and these are rotated between platforms and sidings to keep on platform free on some nights for essential cleaning and maintenance. On the Northern line they are already resorting to the slightly desperate measure of storing trains at Nine Elms overnight due to the lack of an overrun siding at Battersea Power Station although I admit this is more to do with enabling an early start on the Battersea Power Station branch rather than a lack of capacity elsewhere.
The plan for the Bakerloo extension is to have separate tunnels away from the running tunnels to store trains south of Lewisham. That may be the way to go if locations above ground cannot be found but it would be expensive. Ideally, you would have the sidings tunnels slightly larger than the size of standard running tunnels.
Paul,
A very valid point in my opinion.
I remember the late Gordon Pettit, an excellent railway manager, complaining about cuts to British Railways budget by Margaret Thatcher. One of her arguments was that level crossings could be automated to save money (there were a lot of manually supervised crossings then). Gordon complained that she didn’t seem to understand that you needed to invest money to save money.
Victoria Line … the other inherited problem dates back to Treasury penny-pinching before opening.
Only two escalators at Walthamstow Central ( With appalling jams when one fails, as they do .. ) & the wedging of the foot-tunnels & escalators at Highbury – which also only has two, not three.
Piccadilly … You mean Holborn station is not already at or beyond full capacity, right now?
Lastly: wait for some time in the future when money is more freely available. – when the Boris deliberate (malicious?) cuts to TfL’s subsidy are reversed, perhaps?
Even-more lastly (!)
CR2 – CR1 showed an amazing (to outsiders) pick-up in traffic & users – over a million passengers in one day last month, I think. So that CR2 is an automatic “win” for London & it’s transport .. but getting the first cost past the Treasury is going to be a big ask. Incidentally, I seem to be rephrasing PoP’s point about spending-to-save, here.
@PoP – separate tunnels – maybe existing old ones like Aldwych, Trafalgar Square.
Odd sidings abandoned as too short, and some disused platforms – would surface stations be considered too open for the overnight ‘artists’.
The Jubilee Line 1996 trains have an older generation of traction motor electronics than the Northern Line 1995s, I believe they’re similar to those in the GEC powered Networkers.
This may require replacing, or the stock to be replaced earlier than the 1995s.
A Facebook post said that after CLIP completes at the end of the decade, the Jubilee fleet are to take their turn through Acton. Perhaps that is the reason why!
21st C Rolling Stock:
The fundamental difficulty about the whole mechanism of new or refurbished rolling stock comes back to paying for it.
Up until 2015-18, Tfl had the “General Operating Grant” ( A subsidy, in other words ) which was of the order of £700 million p.a. As a matter of record, the then-Mayor, one B. Johnson, “agreed with then tory government” – to stop this payment.
And TfL have been cash-strapped ever since, with very little prospect of ever building a significant new transport project of any size or use.
For comparison, & as far as I can find out { I would be glad to be corrected on this one } London is the only Capital or largest city in any supposedly-developed nation that does not get a “Transit Subsidy”
People will doubtless draw their own conclusions from this.
Rereading this excellent article again, I’ve realised that it is silent on a significant point. Overground rolling stock.
TfL presumably needs to take care of replacements and expansions there too (purchasing outright or continuing to lease as happens today). And that will potentially create an additional replacement pinch point/funding need.
Presumably the Class 378 Capitalstars would need replacing in 2040’s as they reach 30 years old, unless they are stretched to 40 or more (assuming they would not be replaced early with more 710’s for fleet commonality/better customer experience; unclear if there would be demand to cascade the 378’s to other operators in some form to enable this).
An additional fleet for West London Orbital would be also needed in the coming decade (assuming WLO happens before 2040), potentially as a specific subfleet although hopefully on an existing common platform
The 710’s Aventras wouldn’t reach 30 until 2050 i.e. around the same time as the 345 Aventras on the Elizabeth line, and several other Aventra based fleets across the UK. This is a different looming problem unless the procurement model changes to procure from Derby Aventras (and eventual evolutions of the Aventra), at a slow and steady pace, across all UK operators (a sensible use of public funds to provide long term skilled jobs). The other option would be to bundle the replacement with the next subsurface underground trains as a new competitive procurement (gaining a different fleet commonality and cascading the 710 Aventras elsewhere later in life).
On the other hand LO trains can be reasonably “off the shelf” without the unique characteristics of the deep tube lines that lead to extra expense/complication. Only slight wrinkle is the need for dual-current stock on the Mildmay Line (unless you could use straight ac stock with battery/supercapacitor power on the third-rail sections).
https://data.london.gov.uk/dataset/london-underground-average-monthly-temperatures-epr8d/ provides an eleven year history, which shows that the heat in the Victoria Line has seriously worsened in that time! Up 17% so, with no intervention, in 2035 it could reach 36°C.
The Central Line is also of concern, and the Northern line is deteriorating! The Jubilee line is younger and well ventilated. The Piccadilly and Bakerloo lines are planned to get air-conditioned trains, but how well will it work with heat being expelled from trains into the tunnels surrounding them? Piccadilly line upgrade 2 with future resignalling, could provide 36tph, a 50% increase on the current service, which might result in a similar heat rise to the Victoria Line.
Temperatures remain high in the months either side of August. Currently it is left to customers to decide whether to experience that heat, which may be bearable until an incident leads to stoppage of trains in the tunnels. In the future, will more responsibility fall on management to decide when it is too hot to allow passengers into stations? How long would line closures need to last? It helps to have trains pushing air around the system, and the train operator isolated in the cab has the benefit of air conditioning! Empty trains might still operate at such times.
Could the provision of air-conditioning urgently bring forward replacement of trains on these lines?
Greg,
I don’t really disagree with anything you have written though I might have toned it down slightly.
I am assuming there is only room for two escalators at Walthamstow which seems to be a massive oversight when, customarily, tube escalator barrels were built with space for three escalators and the centre space was initially used for a concrete staircase. However, I have some sympathy with the planners as Tottenham Hale, Blackhorse Road and Walthamstow were never expected to be that busy.
I think you need to look at ‘crowded’ from two perspectives when considering Underground stations. For the customer they are overcrowded when they feel overcrowded. I think TfL are more concerned with speed of evacuation in an emergency and dangerous crowd situations. So long as station can handle the exits there is not too much to worry about. If you can’t handle the number of people entering the station you can, if needed, restrict them. I suspect that when you get beyond 27tph on the Piccadilly line you can’t safely exit all the passengers so that prevents you running even more trains so the station is overcrowded in TfL’s eyes.
Most of us would love to see a start on Crossrail 2 but, as mentioned in the article, it is crazy to consider it if your can’t even fund replacing assets that need renewing. The amount needed would be more than the treasury could provide given borrowing and taxes cannot be safely increased and could be counter-productive and it is extremely hard to cut spending elsewhere. As John Bull would say, it is not the best scheme you need to consider but the best one that you can afford (or words to that effect).
I suspect there might be some interesting parallels here to how things are currently being done on the Paris Metro – particularly on the steel-wheeled lines. A massive order of 410 (according to Wikipedia) MF19 trains is replacing the existing fleet on six lines (3, 7, 8, 10, 12 and 13) plus the 3bis and 7bis shuttles, being delivered across a period of more or less ten years. There is somewhat more standardisation across lines on the Paris Metro than in London, although there are still considerations around platform edge doors (Line 13), variable train lengths (the 3bis and 7bis will have shorter trains) and the need to be automated in the future (planned for Line 13).
I suspect the advantage for Paris is a somewhat more stable funding environment.
If there are any Paris experts around, a comparison article would be a fascinating read!
PoP
Walthamstow’s escalator barrel is clearly wide enough for 3, but it was pared down to 2 with a concrete staircase in the middle, during the construction period.
I THINK ( not certain ) that the same was done for Highbury ….
Anyone out there with better information?
Getting off topic but Brixton was originally two escalators with a concrete staircase in the middle. The staircase was subsequently replaced by a third escalator. Can’t remember when.
Most of the Northern line stations on the southern branch also have 2 escalators with a staircase in the middle, so it’s not an uncommon configuration
Excellent article as always.
A couple of things from some involvement with the NTfL project, i wasnt part of the signaling or rolling stock teams but did interface with them. My memory may be faulty so would be interrested in others with more relevent expertise.
1) Like you there was lots of chat about how ideal the Vic line would have been for GoA4 operation for the reasons you mention. Another reason is that the Vic line has no, or very limited, interface with other lines? No shared PWay and no bother with different rolling stock. The water cooler chat was that it was a shame the Metronet upgrade wasnt a few years later as it would probably have been GoA4 assuming the issues of what to do in and emergency and detraining etc could be resolved?
2) If GoA4 were to happen then the depots would also have to be ‘driverless’? Given the number of people milling round and the number of points etc then that would be very expensive/difficult. If not then it would still require a number opf drivers to be retained and a bespoke system? if you have ‘drivers’ then the politics of ‘not being held to ransome’ by unionised drivers is lost – although most on here would recognise that the issue is a bit of a red herring?
thoughts?
@MetronetMole
On a GoA4 train, you probably would want human supervised movements within depots, but that can be achieved a small number of people, maybe not even classed as “drivers” and therefore with different unions/contracts
This manual movement could be controlled with a small control panel in the train in the passenger area if there isn’t a driver’s cab in the design (maybe with front passenger doors open for visibility, or video screens, and with a very limited possible top speed), and maybe also offer a wired or wireless external “walking” controller similar to what is used with some cranes and other construction equipment (similarly complex and dangerous environment). With an external controller, the depot personnel could “walk” the train to some points in the depot (where it is safe and sensible to walk, e.g. into and out of sheds), with good eyes-on situational awareness possible.
The DLR does something similar today with the in-train control desk.
And wireless controllers are used with some US freight locos for low speed shunting and similar operations (the driver stands on the end platforms of the loco with the controller, to better supervise the operation, and flip manual points quickly). So wireless rail traction controllers are proven tech with safety data available to consider for use in the UK.
An interesting question is whether a wireless or in train controller would benefit from instructing the signalling system to instruct the train to move (probably too complex), or, whether the train would signal to signalling system “I am moving slowly under manual control” with the signalling system able to send an “emergency stop” command to the train as a safety fallback.
ChrisMitch: There are also those stations on the Northern line (both north and south) with a pair of escalators but no fixed staircase. These stations previously had lifts hence the original spiral staircases are still in use.
@MilesT: see the Nuremberg metro, lines 2 and 3, as an example of GoA4 in Europe. The older trains have an unused driver cab, the newer trains only a control panel on the front row. This row only has lean seats, like the door areas of Jubilee line stock, and is often occupied by families with children.
Wikipedia tells me service staff is trained to drive trains in emergencies.
The difference with future London trains is that these have no platform edge doors, but sensors detecting people/objects falling onto the tracks instead.
I haven’t visited Nuremberg’s GoA4 lines but have visited Budapest which has adopted Nuremberg technology for GoA4 with no platform screen doors (PSD). Comparing Budapest with the London tube is like comparing chalk with cheese. It’s a new line and its island platforms make the huge island platform at Canary Wharf seem narrow, so crowding risks seem much lower. Although there are no PSDs, there is a light curtain over the track which, if penetrated anything much bigger than a pigeon, trains in the visinity are commanded to stop.
I note in previous comments the use of sensors in place of platform edge doors. Whilst platform edge doors are clearly the best solution, it is not always possible because of cost or engineering considerations such as whether the platforms can support their weight.
One thing AI is remarkably good at is identifying anomalies in pictures or specific objects. Consider AI in the use of medical interpretation of scans or the way your camera identifies people in a picture to try to ensure they are in focus. It seems to me that this would be the way forward when platform edge doors cannot be fitted. Better still, it is often the case that cameras are already present recording what is happening on the platform.
This technology could also be used to ensure it is safe to close the doors on the train. I think that inevitably in time this is going to be safer than relying on the driver or guard (where present) interpreting screens or looking down the length of a long train.
@100andthirty: you are right that Nuremberg is not comparable to London either, in terms of passenger numbers, train length and frequency. However, it is interesting because there initially used to be mixed automatic and manual operations, where line U2 was manually driven and U3 automatically. And my experience with the system is that it is very reliable.
The downside of the RF sensors is that they do not detect people/objects falling between a stopped train and a platform, depending on bystanders to press the emergency button in that case.
For London, I think platform edge doors are still the better solution as there is experience with them on the Jubilee and Elizabeth lines.
TfL Mayor’s Budget 2026/27 GLA Consultation extracts – November 2025 at https://content.tfl.gov.uk/tfl-mayors-budget-submission-2026-27.pdf shows funding for rolling stock and signalling post 2035 needs to triple current levels. (p.54)
It will come as no surprise this is going to get worse. Piccadilly depot reconstruction is going to blow the current budget and we are already looking towards the end of 2026 to get the first Piccadilly line train in service so sometime in 2027 won’t be unexpected.
https://www.standard.co.uk/news/transport/piccadilly-line-upgrade-new-trains-delay-tfl-b1262273.html
https://board.tfl.gov.uk/documents/s25543/pic20251211-item08-part1-piccadilly-%20line%20-upgrade-%20programme%20stage%201.pdf