This was originally written to be posted in the #Railnatter discord, but got too long, so I started a blog to post it on instead.

There are a lot of destinations in Europe reachable in a day if you can catch a pre-10am Eurostar, but not otherwise—or at least, otherwise it gets much more awkward. But from a lot of origin stations, that then requires an overnight stay in London, one of the most expensive cities in Britain, not to mention significantly slowing the journey down, increasing the amount of annual leave you’d need to take to use it for holidays, or time out of the office if travelling for work. If you use Interrail’s route planner, then to catch the 0931 to Paris, from Swansea you’d need to take the 0346 service, and I can’t imagine a worse fate. So I wonder if there are other options?

Reinstating the direct Eurostar services to the south of France would help on some routes: you don’t lose an hour going from Gare du Nord to Gare de Lyon, and you don’t need to worry about leaving extra time to avoid a missed connection, plus you save some time on the line in and out of Paris. But that just shifts the problem further south: there will still be a point where it would be practical if you could reach a pre-10am Eurostar.

Obviously, Regional Eurostar and Nightstar would help too. But the border control issues, tunnel rolling stock safety approvals, rolling stock network compatibility, etc. make this basically a non-starter in 2025.

But is there a way to get to St Pancras earlier, without needing to sit on Swansea station platform at half past three AM in the morning? (Substitute your favourite city north or west of London here.) Most lines in Great Britain are too short to justify a sleeper—either you’d have to depart Swansea before the last express service (at which point you could just get a hotel) or you’d be boarding at Bristol at 1am.

What about if you could have a “slip coach sleeper” service? You rock up to your local station at 9–10pm and board a single sleeper car, and go to sleep. It shuttles off into a siding (or is sitting in a very small bay platform already), sited to the west of the station. Then as the main sleeper service (which could start in Carmarthen, or even further west to meet the ferry) pulls into the station, the carriage pulls itself out of the siding and couples to the train, which then continues onward.

Animation illustrating the above paragraph

(Please excuse the terrible PowerPoint animation…)

Goes the long way around from Reading (assuming coming from the west, at least) and pulls into either King’s Cross or St Pancras at 0630. You wake up at a reasonable time, ready to check in for your Eurostar. Not early enough to meet the first train of the day, but fine for the 0801 or 0931. Similarly to the Caledonian Sleeper, occupies the platform for a couple of hours to let people rest if they want the 0931 so can afford to sleep in. Then for the reverse trip, decouples each coach at the relevant station, and they shuttle into their respective sidings/platforms.

Requirements:

  • Multiple unit sleeper stock, each capable of propelling itself up to maybe 15mph with gangway doors exposed. (If you need a shunter to move the coaches when they need to couple, this entire idea falls apart*)
  • All kinds of accommodation present in each vehicle (possibly with the exception of seats, where one could expect people to wait at the station and join the train as it passes)
  • Automated coupling, or at least more seamless than currently.
  • Approval for automated operations on restricted (a few hundred metres) sections of lines into stations. (If each station needs a separate driver available, or the main driver has to get out at every station to drive the joining carriage, this doesn’t really work.)

Looking at a service from South Wales specifically:

  • Carmarthen already has a platform-adjacent siding to the east, and there used to be one to the west that could seemingly easily be reinstated. As a terminus station though, there’s an issue if the service starts further west. Either you allow the train to sit on the main line while the carriage approaches and couples, or you allow extended autonomous operation to Ferryside and couples there, or you allow coupling at both ends and run gangway-ended all the way (with the resulting restrictions on speed).
  • Swansea is similar to Carmarthen: you have space for a small platform 0 that would seemingly just require some track reinstatement, but the terminus layout means you either couple on the mainline near Landore, or run autonomous to Llansamlet, or couple at both ends.
  • Neath has a little room to the north for a siding to the east, but it would need to be elevated as the existing line is. Alternatively, there is room for a bay platform 0 to the west, but then you’d need to do a little jig through platform 1 to platform 2 to couple. Either way you’d need a new crossover.
  • Port Talbot already has sidings, and enough space to build a platform for one of them, although I suspect the existing sidings couldn’t reasonably spend most of the time blocked by a sleeper coach. You’d also need a new crossover.
  • Bridgend: platform 3 already exists, as do the crossovers needed to turn the carriage around here
  • The throat at Cardiff Central is a notorious mess, but in principle there’s room for a platform -1 and enough crossovers to get to and from the through lines. Otherwise, there might be enough room for a little siding on the viaduct, or further west near the Network Rail building.
  • Newport potentially has room for a bay platform 5 if you borrow some car park, but is lacking the crossover needed to turn the carriage around.
  • Bristol Parkway already has a bunch of sidings. Not really room for a new bay platform, but plenty of space to the northwest for a new small siding. Might need a new crossover.
  • Swindon looks awkward. Could you maybe split the big island platform and add a second bay platform 2a next to platform 2?
  • Reading has had so many reconfigurations to its approach recently, can we spare it another one? It’s close enough to London that getting to St Pancras in time for an early departure becomes less of an issue—you can get the Elizabeth line to Liverpool Street for the 0931 quite easily. Meeting the 0801 is still annoying.

One thing that would make some of this easier would be virtual coupling: where two vehicles can move under the same signal block without being physically connected, by coordinating their movements directly via a communications link. Then sidings would not need to be westward of each station, and physical coupling wouldn’t need to happen at every station. For example, the main train goes through Landore junction and is immediately joined by the sleeper carriage from Swansea, they continue along the line a few dozen metres apart until they reach Neath, where they stop and couple. Virtual coupling is proposed as part of some version of ERTMS, although whether any railway will ever fully implement ERTMS to the point that virtual coupling could be done practically and reliably is an open question.

* Edit 2025-03-08: Thanks to user painting4 on the #Railnatter Discord for pointing out the existence of these adorable little battery shunters. While current models cannot operate autonomously or provide hotel power, producing a variant that does this and allows standard locomotive-hauled sleeper stock to be used is likely to be far more achievable than producing EMU sleeper stock from scratch.