The Bakerloo Line extension to Hayes

TfL has recently been consulting on extending the Bakerloo Line to Lewisham, and possibly Hayes. The idea is an old one, and in general terms has much to commend it. The Bakerloo Line has spare capacity at its southern end, and the area south and east of Elephant & Castle is crying out for a rail service to replace some of the extremely intensive bus services currently operated.

However, the wisdom of replacing the existing National Rail service is much less clear. Many are jumping up and down shouting 'this will be marvellous', but appear not to have given much thought to what it would actually entail. Some of the information contained in TfL's consultation documentation was incorrect and misleading, particularly regarding journey times.

Journey times

In the TfL consultation document, the current journey times are all over-stated and the new journey times are all under-stated. I have compiled a corrected table below. The result is that significant time savings are presented, when the reality is the opposite.

Working from the data included in the consultation, I have calculated the planned journey times on the BLE from Hayes to various key stations in central London. This is for the longer route via Camberwell. Times could be 2 minutes shorter by the more direct route following the Old Kent Road. On the other hand, these figures are presumably also based on 3 intermediate stations either way, as shown in the consultation, but this seems rather inadequate considering needs in Southwark, and 4 or 5 stops would be more realistic, so the times could equally be around 2 minutes longer.

These times do not appear to be realistic for reasons explained later, so I have added a second column with my estimates.

The next few columns show current journey times on the Bakerloo and Northern Lines from Elephant & Castle, and then current off-peak journey times on existing Hayes line services. Note that this is based on the normal timetable, not the current temporary timetable during London Bridge works from August 2016 all trains (including peak services) will permanently stop at London Bridge.

Connection times are in brackets. The figures in the first column are official TfL figures, presumably based on changing to the Northern Line at Elephant & Castle, for which an allowance of 3 minutes has been made, which seems accurate assuming the first train can always be boarded. The times for Cannon Street are presumably actually for Bank, which is only a short walk away so an acceptable alternative. Interestingly, changing at Lewisham would generally be quicker for London Bridge and Cannon Street, depending on connections, but I have ignored this.

In the second column, I have corrected these and also added a time for Westminster as representative of the Jubilee Line (changing at Waterloo, assuming a 4 min connection time).

For NR I have calculated connection times using a connection time of 5 mins to Northern or Jubilee Line at London Bridge, and 6 mins to Bakerloo at Charing Cross, all incorporating 1 min waiting time but using fastest available NR trains. Changing to the Jubilee Line at Waterloo East/Southwark is more convenient than London Bridge, but slower overall.

Note, either Waterloo LUL or Waterloo East times can be compared for access to Waterloo mainline station, as the walk time is about 4 minutes from either.

PlaceBakerloo Line ExtensionCurrent LULCurrent NRJourney time difference from ...
TfLCorrectedBakerlooNorthernHYS-CHXHYS-CSTLEW-CHXHayesCatford BridgeLewisham
Hayes 00 00 - - 0000- -- -- --
Elmers End 07? 08 - - 0909- -1 -- --
Catford Bridge 14? 16 - - 1818- -2 -- --
Lewisham 18 20 - - - 2300-3 -1 --
Elephant & Castle 30? 32 0000- - - -- -- --
London Bridge (37)(39) -03293408+10+12N/A
Cannon Street/Bank(39)(40)-05 - 38- +2+4N/A
Old Street (44)(44)-08(39)-(18)+5+7+6
Waterloo LUL/East 34 36 04-34- 13 +2 +4+3
Westminster -(41)--(38)(43)(17)+3+5+4
Charing Cross 363806-38 - 17 + +2+1
Oxford Circus 404210-(48)-(27)-5-3-4
Paddington 495018-(56)-(35)-5-3-4

Thus journey times would be longer for all passengers other than those travelling wholly between Hayes and Lewisham (including those changing to DLR), and those who already change to the Bakerloo Line at Charing Cross. I would estimate that around 10% of current Hayes line passengers fall into each of those categories, so the majority (80%) would be worse off.

All times are of course point to point train timings and take no account of time taken to get in and out of stations, walk along platforms or wait for trains. Trains are currently low frequency, so it is reasonable to make an extra allowance for waiting for a train. In the morning most passengers only arrive a few minutes before the train is due, so would wait no longer than for a more frequent Underground train, but in the afternoon people tend to arrive more randomly so average waiting times may be 5-10 minutes, which largely offsets the journey time disbenefit. However, the increased frequency is also stated as a benefit in its own right, so you must not count the effect twice.

Journey time assumption

It struck me that the big reduction in journey times between Hayes and Lewisham did not seem right. Current trains are generally scheduled to take 23 minutes, although this includes 1 minute of "padding"; TfL are claiming 18 minutes for exactly the same calling pattern. This is a distance of 8 miles 30 chains = 8.375 miles, so the current scheduled time works out as 21.8mph, and 18mins would be 27.9mph. The latter speed is close to the maximum anywhere on the Underground.

Tube trains might have marginally shorter dwell times, possibly saving a minute or so. They would also have faster acceleration (current trains on the line have comparable capabilities, but are restricted because of power supply limitations). Automatic train operation would allow faster braking. On the other hand, tube trains also generally have a top speed of 45mph, as against 60mph applicable on the Hayes line (although this maximum speed is typically only achieved between Lower Sydenham and Catford Bridge). Overall, improved train performance might possibly save another minute, getting us down to 20 minutes. By adopting the same tricks, NR could achieve 21 minutes there is no reason to think this will not happen, although as the National Rail network is a mixed traffic railway the technology is being implemented much more slowly.

Can we find a benchmark? One of the fastest sections on the Underground is the Central Line between North Acton and West Ruislip, which is almost straight with relatively widely spaced stops it is built on part of the old Great Western main line from London to Birmingham. This is a distance of 13.28km = 8.25 miles, which trains complete in 17 minutes, giving an average speed of 29.1mph. This includes 6 intermediate stops; at a steady non-stop speed of 45mph this distance would take 11 minutes, which implies about 1 minute needs to be added for each station stop. Hayes Lewisham is almost exactly the same distance, but includes 8 intermediate stops, and also a long sharp curve south of Elmers End which takes approximately 1.5 minutes to traverse. Just 1 minute longer is clearly not enough to cater for these three extra sources of delay. So 20 minutes does seem to be an accurate estimate, which I have used above.

It is also odd that TFL have assumed quite heroic time savings on the former NR route, but assumed a slightly longer time between Elephant & Castle and Paddington even though this will gain similar benefits from the line upgrade works! I have used the current running times as a fairer comparison; any future time savings over this section would occur regardless of whether the line gets extended.

As a matter of interest, Lewisham Elephant via the Old Kent Road alignment is approximately 4.52 mile giving a speed of 27.1mph, whilst via Camberwell it is around 5.35 mile giving 26.75mph. This seems realistic for the widely spaced stops proposed.

Train capacity

There would be a much more frequent service, but the trains would be much smaller shorter, narrower, and to tube height which is more cramped inside. Here are details:

8-car 465 160m6961200
10-car 465200m8641500
10-car 376200m4561500
12-car 465240m10441800

Exact seating capacity of new style Bakerloo Line stock is not known, but is based on other recent LUL stock. Total capacities are realistic figures rather than official. In fairness, it is quite possible that 376-style standee trains will become the norm anyway in future, especially if the Hayes line ever gets transferred to London Overground. These are however an aberration and offer no operational or passenger benefits whatsoever, and many disbenefits.

There is no inherent reason why trains should not all be 12-car but there is simply not enough demand at present to justify it. If they were, we would have up to 6264 seats and 10,800 capacity in total, in the busiest hour. The Bakerloo line promises 27 trains per hour giving 6588 seats and 18,900 total capacity, or 5% and 75% more respectively. However, there would only be 21tph south of Catford Bridge and 15tph south of New Beckenham, meaning a sharp reduction in seats; even now many passengers have to stand even south of Catford Bridge, meaning some passengers will be in for a lengthy stand.

Furthermore, Hayes line passengers would be sharing the space with existing Bakerloo Line passengers, and "new" passengers from the various stops between Lewisham and Elephant & Castle (particularly those changing from other trains at New Cross Gate), never mind any from Bromley South if the line runs there as well. Current bus services deliver around 11,000 passengers per hour from the Walworth Road and Old Kent Road corridors to Elephant & Castle, many of whom would switch to the Bakerloo Line (though some already do at E&C). It therefore seems that the line would not actually deliver sufficient capacity for the likely demand, let alone allowing for future growth.


The plan relies on many passengers changing to the Northern Line at Elephant & Castle to complete journeys into the City. However, this is already one of the most overcrowded sections of the Underground. Changing at Lewisham onto National Rail services is a better alternative, but if passengers do this then rather obviously the claimed benefits will not materialise!

Release of train paths

This has been mooted as a possibility by Network Rail in their Route Utililisation Strategies, but no details are given, and no particular need demonstrated. Southeastern trains via London Bridge are mostly 8 or 10 carriages, but could all be lengthened to 12, providing roughly a 33% increase in capacity, enough for many years to come. The area has also benefited from several other rail schemes including DLR (x2), Crossrail and HS1, so it seems unlikely that the train paths that could be re-allocated from Hayes services will be needed for a long while. Mainline services are in any case constrained by track capacity south of Orpington, which is near saturation.

Also, note that regardless of whether the Bakerloo Line takes over the Hayes branch, it is bound to abstract some passengers from all the other trains which serve Lewisham and New Cross, releasing capacity on those services without any changes to the timetable being needed.

In the medium term, applying automatic train operation (as already adopted on much of London Underground) would enable more trains to run on the existing tracks anyway.

In the longer term, a NW-SE Crossrail-style route to National Rail standards would be a more appropriate way to incorporate the Hayes line although see below.


The Bakerloo Line extension is dependant on external funding, principally from developments along the line. Whilst there is plenty of development going on between Lewisham and Elephant & Castle, there is not likely to be much between Hayes and Lewisham. The Lower Sydenham industrial area is a possible candidate, but beyond that there are only a few small sites such as Catford Green, and nothing which would generate sufficient extra demand to exceed the existing capabilities of the line.

Line conversion

Downgrading the line from National Rail to tube standards would be quite disruptive. It is quite likely TfL would want to completely rebuild the platforms to the lower tube height, although raising the tracks would be a cheaper alternative. This would require complete closure of the line for several months, at least.


There is some potential for improvements to the existing National Rail service. Once London Bridge work is complete, all trains will stop there. Following doubling of the link from Lewisham to the fast lines at St. John's it would be possible for all Hayes line trains to run via Lewisham, as indeed they did during off-peak hours until the early 2000s the current 2tph service serving Lewisham is by far the biggest turnoff about the existing timetable. (This would however lengthen journey times on Charing Cross services by 2 minutes.) Longer trains are also likely as and when demand materialises.

One point to note is that whilst it would be possible for all Hayes trains to serve Lewisham, this is not true of the routes from Orpington and from Dartford via Sidcup. An alternative way of providing more frequent services at least between Hither Green and Lewisham should therefore be the top priority.

As to the Bakerloo Line itself, I would like to see both OKR and Camberwell branches built. There should also be more frequent stops in this area, which currently has no train services at all. Ideally everyone in that area should be within walking distance of a station, which implies stations at most a mile apart, not 1.5-2 miles as proposed.

The branches could be linked as a loop, but there is clearly a strong desire to serve Lewisham, so they could join back together again, or the Camberwell branch could terminate at Peckham. I would suggest building the line as far as Lewisham initially, but in such a way that it could be further extended if it proves still to have spare capacity. An extension to Catford in tunnel would be one option, or to Bromley North via Grove Park.

Bromley Council does not support the Bakerloo Line serving Hayes for the reasons outlined above. Its preference is to extend the DLR to Bromley North via Grove Park, taking over the existing branch, and possibly running on to Bromley South. This scheme has much to commend it, although the scheme described below would probably be better. An alternative would be to extend the DLR to Catford, where connections would be provided with services from Bromley and beyond, as well as Hayes Charing Cross services if they continue not to serve Lewisham.

East London Crossrail

There have been some suggestions of a new north-south rail link serving the Docklands, an idea first germinated as "Brighton Main Line 2". This would address the fact that most existing rail services are focused on central London, whereas Docklands is becoming an increasingly important destination for people over a wide area. Transport links from the north and south remain very poor, with most passengers obliged to travel via central London, which quite apart from anything else is a waste of precious rail capacity in central London.

The line would largely use existing infrastrucure which has spare capacity, along with a short new tunnel from South Croydon to Addiscombe (old station site) and a longer tunnel from Lewisham to All Saints (Poplar), where it would take over the DLR tracks via Bow Church to Stratford, and then replace the current Stratford Bishops Stortford service, taking advantage of the planned Lea Valley line upgrade. The core route could serve:

Branches and notes
Gatwick Airport004
East Croydon1784tph from Oxted line, potentially from Brighton/Eastbourne via Uckfield
Elmers End218
Clock House238
New Beckenham25124tph from Beckenham Junction, or possibly Bromley South or beyond
Lower Sydenham2712
Catford Bridge3012
Lewisham34208tph from Bromley North via Grove Park, possibly also Sidcup
Canary Wharf3820
All Saints4020
Langdon Park4120
Devons Road4320
Bow Church4420
Pudding Mill Lane4720
Stratford50208tph terminate here
Lea Bridge56124tph to Chingford via Walthamstow
4tph to Enfield Town via South Tottenham
Tottenham Hale60 42tph continue semi-fast to Hertford East
Bishops Stortford (fast)85 2
Stansted Airport95 2

Of necessity all existing stations would be served between Elmers End and Lewisham and between All Saints and Stratford, but otherwise trains would be limited-stop and could achieve speeds up to 90mph over much of the route, providing a fast direct link and relieving the DLR. There would only be three new under-ground stations to build, at East Croydon, Lewisham and Canary Wharf, so the scheme should be relatively cheap tunnelling itself is not expensive maybe around 2bn.

Key journey times would include:
Gatwick Lewisham: 34 mins
Croydon Lewisham: 17 mins
Gatwick Canary Wharf: 38 mins
Croydon Canary Wharf: 21 mins
Catford Canary Wharf: 8 mins
Gatwick Stratford: 50 mins
Lewisham Stratford: 16 mins
Lewisham Stansted: 61 mins

Existing infrastructure would be used to South Croydon, where the line would drop into tunnel with an under-ground stop at East Croydon. The tunnel would emerge on the trackbed on the former Addiscombe branch. It would be accommodated alongside Tramlink tracks from Blackhorse Lane to Elmers End and then join the existing line to Ladywell. The next tunnel would start under Parks Bridge between Ladywell and Lewisham, again including under-ground platforms at Lewisham station, and then at Canary Wharf. The tunnel would surface on, and take over, the trackbed of the DLR line between Poplar and Stratford, which was originally converted from a British Rail Line anyway. A new flying junction of some description would be required between Pudding Mill Lane and Stratford to cross over the Great Eastern main line. Alternatively, the tunnel could continue to Stratford with no intermediate stops, which would reduce the journey time by around 7 minutes. After that it is all on existing infrastructure, although a stop at the proposed new Lea Bridge station is included, as is use of a re-opened Hall Farm Curve to open a new link to Walthamstow and Chingford.

Some trains would start from Bromley North taking over the existing branch to Grove Park, then new segregated tracks to Parks Bridge. A surface route may be possible as far as Hither Green, followed by a short tunnel.

Specifically, by accident, the line would provide massive benefits to stations between Elmers End and Lewisham. (The existing services to London Bridge would be retained, although these might be re-routed away from Lewisham, and diverted to Charing Cross, apart from a few peak services).

How realistic this plan is remains to be seen, but retaining the Hayes line to Network Rail standards would be crucial.

High Speed 2 link

Another reason to retain the Hayes Line to National Rail standards is the potential need for a new train service to the HS2 hub at Old Oak Common. The current plan is for the line to continue to Euston, but large savings (believed to be around 5bn) would be possible by terminating at OOC instead.

There is understandable opposition to this as OOC is not in central London, but the reality is that Euston is not most passengers' destination either, and in fact, with Crossrail and Overground enhancements, OOC would be just as quick and easy to get to as Euston from most parts of central and Greater London, and indeed the wider area, with most rail lines having a direct connection with either Crossrail or Overground.

The one major exception to this is South East London, where access to Euston via the Northern or Victoria Lines is relatively easy, but getting Old Oak Common generally requires 2 changes. This could be avoided by introducing a new train service calling at the following stations:
Old Oak Common
Shepherd's Bush (not essential)
Peckham Rye (links to Southern route via Tulse Hill)
Lewisham (link to Dartford and Orpington lines)
Catford Bridge (not essential)
Beckenham Junction (link to Tramlink)
Bromley South (link to mainline services via Chatham and Maidstone)
Orpington (link to mainline services via Tonbridge)

This is the best, probably only, route for such a service, maximising interchange possibilities whilst taking into account existing capacity constraints on the network. End to end journey times would be around 45 minutes, which is similar to the current journey time to Euston. Once again, the Hayes line is used, between New Beckenham and Lewisham.

Published 12 March 2015
HS2 link added 30 March 2015