Easier Splitting/Joining?

[Topological]

Thinking about the Castlefield Corridor problem in Manchester, but more generally about paths in and out of busy stations, would there be a benefit in having a fleet of local trains that could split and join very quickly? I am thinking technology that allows the split within a standard station stop time.

Take an example of a line between A and B where there are branches then to C, D, E and F. (For example Manchester to Bolton then on to Southport, Kirby, Preston, Blackburn). If the trains take time to split and join then it will be inefficient trying to plan their journey on the busy A to B section. However, if it does not matter which destination attaches/detaches then things are much more flexible. You can have paths on the A to B section and then just send whatever has managed to join. Splitting will be less of an issue in this case.

Yes, there are issues with people getting in the wrong part of the train that would need to be mitigated. I am sure there is a work around though.

In terms of trains, I am thinking of something with 2 carriages which can run in up to 12-car formations (though the Manchester example may need an 8-car limit for Oxford Road)

IF there was a path for the "joined" train every 15 minutes from A to B, then the effect on the various lines could be transformational.

 

[Farnborough]

SWR regularly splits off-peak (8 or 12 car) Class 450 and (10 car) Class 444 rakes... eg Waterloo-Woking-Basingstoke/Alton splits at Woking, while Waterloo-Bournemouth-Poole/Weymouth splits at Bournemouth.

 

[Magdalia]

Splitting will be less of an issue in this case.

IF there was a path for the "joined" train every 15 minutes from A to B, then the effect on the various lines could be transformational.

Splitting is the easy bit. In most cases, splitting will also involve joining, which is much harder.

Without joining too there will be little to gain in terms of utilisation of capacity, rolling stock or traincrew.

Joining requires permissive signalling to allow the second portion into a platform already occupied by the first portion. Even if the infrastructure is there, that takes time, because the second portion can't approach at the same speed that it would use if the platform was empty.

Joining is also where most of the service resilience issues arise. What do you do with the first portion if the second portion is running late? What do you do if two portions that work fine separately won't work together?

In what was the Network SouthEast area splitting and joining is much less common than it used to be, because it is a significant performance risk on a congested railway. As someone who has long experience of travelling on the GN, I can say that, when most fast Fen Line trains split and joined at Cambridge, the splitting and joining was a significant source of delay.

 

[Snow1964]

It used to be common on the Southern, obvious reason one long train is better than two short ones where paths are at a premium.

There used to be semaphore signals with C (calling on) indicators, modern colour lights can have a permissive equivalent. Although has fallen out of favour.

From memory used to regularly happen at Ascot, Faversham, Haywards Heath, Southampton, Woking, Worthing, etc

 

[zwk500]

Thinking about the Castlefield Corridor problem in Manchester, but more generally about paths in and out of busy stations, would there be a benefit in having a fleet of local trains that could split and join very quickly? I am thinking technology that allows the split within a standard station stop time.

This would need a completely redesigned coupling system to allow the detachment and attachment to take place without needing to lock the doors on both units. At the moment, the units need to stop, open the doors, close the doors, split or join, then open the doors again before starting the despatch sequence for departure. If running late, they sometimes may do the split or join before opening the doors to save that time.

However the actual coupling/uncoupling option is, as @Magdalia says, a relatively small part of the time required. The bigger issue is that before the join or after the split the two portions become completely separate trains to the signalling system. Splitting is more straightforward - the 2nd train waits for the 1st train to get far enough down the line to receive a separate Green (or yellow/double yellow if appropriate) signal. Joining is harder - the 2nd train must be coming from far enough back to get it's own green signal initially, but as the first train is sitting in the platform it will be checked down to a red signal before the station and given a separate 'permissive' aspect, which requires the driver to approach cautiously.

This requires the 2nd portion of a splitting train and the 1st portion of a joining train to stand in the station waiting for the other portion.

 

[TheGrew]

I don't think this would be a viable solution for Castlefield, unfortunately, because the rolling stock is far from uniform. Where you typically see this type of working today (thinking SWR/Southern) the rolling stock is identical or a compatible sub-class. That would be rarely the case with Castlefield.

 

[Fawkes Cat]

I don't think this would be a viable solution for Castlefield, unfortunately, because the rolling stock is far from uniform. Where you typically see this type of working today (thinking SWR/Southern) the rolling stock is identical or a compatible sub-class. That would be rarely the case with Castlefield.

I suspect that the other issues raised mean that this idea is a non-runner - but 'the fleet isn't uniform' is a solvable problem in that all stock is eventually replaced. Were that to be the only thing in the way of joining and splitting being a viable solution, then a new fleet could be acquired, and the existing stock cascaded to elsewhere.

 

[Nottingham59]

I don't think this would be a viable solution for Castlefield, unfortunately, because the rolling stock is far from uniform. Where you typically see this type of working today (thinking SWR/Southern) the rolling stock is identical or a compatible sub-class. That would be rarely the case with Castlefield.

Splttitng and joining is going to be really important when HS2 starts. The restriction on paths through Handsacre and the lack of 400m platforms at any of the main destination cities (except perhaps Edinburgh) means that there will be huge capacity benefits of a timetable that involves lots of splitting and joining.

The HS2 Train Technical Specification has ambitious targets for this process, though I note it doesn't specify a reliability measure:

When coupling, the maximum jerk and acceleration experienced on either Unit shall be low enough for Passengers to be able to board and alight the stationary Unit during the proces.

9.8.2.1.3 TTS-251 - Maximum Couple Time (PQTS-151) The time to couple two Units shall not exceed 120 seconds. This is measured from the time when the second Unit contacts and mechanically couples with the stationary Unit until the combined Train is ready to depart. It includes:  the time to make an electrical connection and re-configure systems into a single operational Train;  the time for normal dwell activities (e.g. close Exterior Doors), which may overlap coupling and reconfiguration; and  the time for the combined Train to request an ETCS MA.

 

[TheGrew]

I suspect that the other issues raised mean that this idea is a non-runner - but 'the fleet isn't uniform' is a solvable problem in that all stock is eventually replaced. Were that to be the only thing in the way of joining and splitting being a viable solution, then a new fleet could be acquired, and the existing stock cascaded to elsewhere.

Indeed, I can't think of any obvious places to combine trains on either side of the corridor either. On the western side, you have five routes (C2C, Chat Moss, Orsall Curve, Manchester-Preston & Manchester-Wigan) that converge without an intermediate station except Salford Crescent for the latter two. The eastern side is pretty similar as well.

 

[Topological]

Thanks all for the replies. Certainly, the signalling is the big issue as it makes joining more than a couple of services very hard, unless there is a junction next to the platform.

Whether you could design a signalling system that would accommodate tighter headways for trains that were going to join I am not sure.

I can see why you would not want to be sat in a platform waiting for multiple portions to arrive (remember this is only if the trains arriving were on time, and would only be for the booked waiting time) if signalling was only using standard headways.

For reference in the Manchester case, the trains would all be going to the Airport so all splits/joins would be at Salford Crescent (platform 2) or more likely Bolton.

 

[Bletchleyite]

This would need a completely redesigned coupling system to allow the detachment and attachment to take place without needing to lock the doors on both units.

It's physically possible to couple trains without closing the doors (the software might not like it, but that's easily fixed). The reason it isn't done is that there would be a risk of people being knocked over just as they board/alight or worse onto the track if the trains were banged together a bit hard. You're not going to get around that.

The main cause of delays with joining in particular (less so with splitting) is what you do if one portion is late.

 

[zwk500]

Whether you could design a signalling system that would accommodate tighter headways for trains that were going to join I am not sure.

Yes, you can it's called ETCS. However there are practical limits of what ETCS can do (conventional signalling can have planning headways of 2 minutes in some areas, ETCS is unlikely to get a planning headway below 1.5 minutes except in extremely limited conditions), and a cost associated with the greater amount of equipment required.

Castlefield corridor is, IIRC, a 2.5 minute planning headway currently under existing signalling.

It's physically possible to couple trains without closing the doors (the software might not like it, but that's easily fixed). The reason it isn't done is that there would be a risk of people being knocked over just as they board/alight or worse onto the track if the trains were banged together a bit hard.

AIUI most trains have a power interlock with the doors open. So to solve it you would need to have couplers that could be extended towards each other while the train remained stationary. And then either retracted as the train moved off/did a brake check or locked into the extended position to enable a sharper split.

Perfectly feasible to do so, just would cost a fair whack of money to get it through the approvals. The other option would be to just not open the doors first and advertise the arrival 1 or 2 minutes later than planned to allow the attachment to take place straight away.

 

[Bletchleyite]

The other option would be to just not open the doors first and advertise the arrival 1 or 2 minutes later than planned to allow the attachment to take place straight away.

This is what you'd do - done slickly it only adds seconds.

 

[Mgameing123]

Thinking about the Castlefield Corridor problem in Manchester, but more generally about paths in and out of busy stations, would there be a benefit in having a fleet of local trains that could split and join very quickly? I am thinking technology that allows the split within a standard station stop time.

Take an example of a line between A and B where there are branches then to C, D, E and F. (For example Manchester to Bolton then on to Southport, Kirby, Preston, Blackburn). If the trains take time to split and join then it will be inefficient trying to plan their journey on the busy A to B section. However, if it does not matter which destination attaches/detaches then things are much more flexible. You can have paths on the A to B section and then just send whatever has managed to join. Splitting will be less of an issue in this case.

Yes, there are issues with people getting in the wrong part of the train that would need to be mitigated. I am sure there is a work around though.

In terms of trains, I am thinking of something with 2 carriages which can run in up to 12-car formations (though the Manchester example may need an 8-car limit for Oxford Road)

IF there was a path for the "joined" train every 15 minutes from A to B, then the effect on the various lines could be transformational.

I think you going to need to talk with the people behind DSB’s IC3 and the IR4 (the electric version of the IC3):

 

[zwk500]

This is what you'd do - done slickly it only adds seconds.

Bit more than that.

For joining: Train 1 arrives, releases doors. Train 2 approaches, train 1 closes doors. Train 2 stops short, receives confirmation doors are closed, couples, then performs pull-away test, then is able to release all doors again.
For splitting: Train arrives and stops (at this point passengers are hammering the door open button), driver 2 boards, presses uncouple, sets train back slightly or gives indication to driver 1 to pull forward.

I'm not sure if the driver of train 1 in either instance would need to change cabs to be in the cab nearest the coupling/uncoupling operation itself or if the driver of train 2 would be sufficient.

 

[Magdalia]

I'm not sure if the driver of train 1 in either instance would need to change cabs to be in the cab nearest the coupling/uncoupling operation itself or if the driver of train 2 would be sufficient.

When class 387s first took over the Kings Lynn trains, it required 3 drivers to do a split at Cambridge. I don't know if that is still the case.

 

[bahnause]

AIUI most trains have a power interlock with the doors open. So to solve it you would need to have couplers that could be extended towards each other while the train remained stationary. And then either retracted as the train moved off/did a brake check or locked into the extended position to enable a sharper split.

I can't see why it would be necessary to move a train after uncoupling it before the actual departure. Our procedures don't involve closing the doors before attaching another train, nor stopping before attaching to another trainset.

 

[PeterC]

I think you going to need to talk with the people behind DSB’s IC3 and the IR4 (the electric version of the IC3):

Not invented here so its not going to happen on this side of the North Sea.

 

[zwk500]

I can't see why it would be necessary to move a train after uncoupling it before the actual departure.

The train is moved a short distance (about 30cm IIRC) to allow the physical couplings to be clearly seen to have disengage properly.

Our procedures don't involve closing the doors before attaching another train, nor stopping before attaching to another trainset.

Different railway, different units, and more importantly a different legal framework. If the doors were open, passengers would try and board. If there was a jolt as they did so and a passenger injured due to the attachment of the units, the railway would potentially be liable to a legal action.

 

[Mgameing123]

Not invented here so its not going to happen on this side of the North Sea.

So invent it? That is probably the worst argument I’ve ever heard against something.

It’s the same as saying: “We haven’t invented high speed rail so therefore HS1 is never going to happen”

 

[zwk500]

So invent it? That is probably the worst argument I’ve ever heard against something.

It’s the same as saying: “We haven’t invented high speed rail so therefore HS1 is never going to happen”

The bigger reason why it is extremely unlikely to happen in the UK is the regulatory approach to risk. A system designed to uncouple on the move has an inherently greater chance of doing so when not supposed to than a system designed to only allow uncoupling at a stand. It may be an extremely low risk, but the ORR and RSSB look extremely unfavorably on any additional risk that is not exactly 0.

Not to mention the system designed for uncoupling will likely cost more as it has to uncouple with greater forces acting on it and presumably has speed and location controls to prevent unauthorised decoupling?

Oh and it really doesn't save very much time against coming to a stand and then performing a swift decouple before the doors open, and by definition cannot have an equivalent for the other direction. It also doesn't address any of the other inefficiencies of splitting and joining operations such as poor utilisation or resources and punctuality issues.

 

[The Planner]

Splttitng and joining is going to be really important when HS2 starts. The restriction on paths through Handsacre and the lack of 400m platforms at any of the main destination cities (except perhaps Edinburgh) means that there will be huge capacity benefits of a timetable that involves lots of splitting and joining.

The HS2 Train Technical Specification has ambitious targets for this process, though I note it doesn't specify a reliability measure:

Which is a bit moot, as that spec assumes ETCS. None of which is likely to be on the WCML even by the time HS2 gets to Handsacre.

 

[Mgameing123]

The bigger reason why it is extremely unlikely to happen in the UK is the regulatory approach to risk. A system designed to uncouple on the move has an inherently greater chance of doing so when not supposed to than a system designed to only allow uncoupling at a stand. It may be an extremely low risk, but the ORR and RSSB look extremely unfavorably on any additional risk that is not exactly 0.

Not to mention the system designed for uncoupling will likely cost more as it has to uncouple with greater forces acting on it and presumably has speed and location controls to prevent unauthorised decoupling?

Oh and it really doesn't save very much time against coming to a stand and then performing a swift decouple before the doors open, and by definition cannot have an equivalent for the other direction. It also doesn't address any of the other inefficiencies of splitting and joining operations such as poor utilisation or resources and punctuality issues.

The system only allows you to uncouple when the train is going under 8 kmph so therefore even if a train uncouples it would be at a relatively safe speed and would ground to a halt instantly.

I’m not suggesting that trains must be able to couple to use the Castlefield corridor but if they must I do suggest talking to the designers of the IC3 to create the most efficient solution because it is a very efficient solution they use as trains can just depart just as fast as if the trains didn’t need to uncouple.

 

[zwk500]

The system only allows you to uncouple when the train is going under 8 kmph so therefore even if a train uncouples it would be at a relatively safe speed and would ground to a halt instantly.

But it therefore has a system to release the control before the train is at a complete halt. That risk is unlikely (from prior examples) to be accepted by the UK safety authorities.

I’m not suggesting that trains must be able to couple to use the Castlefield corridor but if they must I do suggest talking to the designers of the IC3 to create the most efficient solution because it is a very efficient solution they use as trains can just depart just as fast as if the trains didn’t need to uncouple.

1 train benefits by in the order of 2-3 minutes. The 2nd train doesn't benefit at all. In the other direction, also no benefit. How much does this system cost for that benefit?

Worth mentioning the UK did invent slip couplings, as it happens.

 

[Nottingham59]

Which is a bit moot, as that spec assumes ETCS. None of which is likely to be on the WCML even by the time HS2 gets to Handsacre.

Certainly, but one would hope that HS2 trains will also be able to split and join on the classic network too, in roughly the same timescale.

== == == ==

One option for HS2 that I think would work well with Phase 1 is for 400m trains to Manchester to split at Stoke, with only the rear half alongside the station platform. The front half proceeds immediately on to Stockport and Manchester, whilst the rear half calls at Stoke and terminates on platform 2 at Macclesfield.

Returning to London, Colwich will not be a constraint for these trains, where they will only have to cross the path of southbound freights. So these units could rejoin at Interchange or Old Oak Common, where there will be spare 400m platforms to wait if one unit is delayed.

 

[RiverDon]

It's physically possible to couple trains without closing the doors (the software might not like it, but that's easily fixed).

If you try to couple in a 158 or 170 with the doors energised on the stationary unit then the couple relay might not set correctly and you could very easily get a wrong side failure on the door control circuits ie. the doors could energise or open on the non platform side on one or both units. There's no software involved, it's just electro mechanical relays.

Also you won't get a brake release if the doors are energised so you can't do the brake continuity test, which is vital to know that the units are coupled correctly.

 

[Nottingham59]

Thinking about the Castlefield Corridor problem in Manchester, but more generally about paths in and out of busy stations, would there be a benefit in having a fleet of local trains that could split and join very quickly? I am thinking technology that allows the split within a standard station stop time.

Yes, quick and robust splitting and joining should be possible to design into any new multiple units.

But to me the answer is to run longer units.

If ATW want to run trains through Castlefield, they should be required to run at least 8-car trains, or terminate at Victoria. Four-car EMR units should terminate in the main shed at Piccadilly. Similary, if Lumo want to use scarce capacity on the southern ECML, they should run 10-car trains. EMR should have ordered 234m-long standard 9-car IETs, rather than bespoke 5-car 810s for the Midland Mainline. And Cross Country should be running longer trains too.

It's entirely wasteful for the country to allow short trains to be run on congested railway infrastructure.

 

[zwk500]

Certainly, but one would hope that HS2 trains will also be able to split and join on the classic network too, in roughly the same timescale.

However the timing calculation to receiving the ETCS MA is unachievable if not using ETCS. Whether the Signaller can clear the signal within 2 minutes of mechanical coupling on conventional signalling will depend on local conditions. In most cases they should. Many setups will not allow a permissive aspect to be shown towards a signal showing clear, but once the 2nd train has passed the permissive signal and the route released the platform signal should be able to be cleared basically immediately (assuming the route is clear, of course).

One option for HS2 that I think would work well with Phase 1 is for 400m trains to Manchester to split at Stoke, with only the rear half alongside the station platform. The front half proceeds immediately on to Stockport and Manchester, whilst the rear half calls at Stoke and terminates on platform 2 at Macclesfield.

Returning to London, Colwich will not be a constraint for these trains, where they will only have to cross the path of southbound freights. So these units could rejoin at Interchange

Surely it makes more sense to do the split/join at the same place? Unless you do pick up @Mgameing123's suggestion of the coupling system used in Denmark and the front portion doesn't even stop.

== Doublepost prevention - post automatically merged: 22 Dec 2025 ==

But to me the answer is to run longer units.

Yes

 

[styles]

You do both splitting and joining while the train is in motion if you are so inclined. DirectTrainS hope to roll it out with someone eventually.

 

[zwk500]

You do both splitting and joining while the train is in motion if you are so inclined. DirectTrainS hope to roll it out with someone eventually.

Assume these are the people: https://news.railbusinessdaily.com/...olutionary-dynamic-train-coupling-technology/

DirectTrainS, the rail technology startup developing aerospace-inspired solutions for the transportation industry, has successfully completed its first field test of the Dynamic train Coupling (DC) system

They've coupled 1 pair of wagons in 1 test run in June this year. Notably the speed this test was carried out at was omitted from the article trumping it's success:

The tests, conducted in collaboration with French engineering partner company CERTIA, demonstrated the DC system’s ability to couple and decouple high-mass trains in motion on a real railway line. Using full-scale, 22.5-ton carts for real-life coupling scenarios, the tests validated the precision and smoothness of the operation, even passing the “water cup test,” with no spillage from a cardboard cup placed on the fixture during coupling.

Forgive me if a lot of their well-prepared publicity feels a bit too good to be true. This system feels like it could only release meaningful capacity in areas that have lots of spare route capacity and limited yards and terminal capacity.

Let's take this line:

Both physical and software components are designed for seamless installation on existing rolling stock with no modifications required to infrastructure, signalling systems, or regulations, making it compatible with any existing railway operation.

First, what they don't say - Information Systems (planning, operation, traffic management etc) and Rolling Stock are conspicuously absent from the 'no modifications' line. Cost is also not mentioned. However, all are solvable with the right will.

Then onto what they do say: infrastructure and signalling systems will need modification to allow trains into blocks with moving trains (Currently only allowed in the UK by permissive (conventional) or on-sight (ETCS) MAs). Signalling sections will need amending to allow trains to actually catch up under Permissive aspects. Yards, terminals, loops etc may need modification to accept longer trains if the cargo load isn't reduced. Signalling systems will also need modification to handle trains to split on the move (especially ARS-type systems). Infrastructure may need modification to allow storage of the DC units in suitable places for shunting and marshalling. Regulations will need amending to allow the necessary changes to the signalling systems, and appropriate operational procedures for events such as decoupling failure, 1st train emergency stop, portions arriving in the wrong order, etc.

Assuming that they are planning for a loco to be marshalled mid-train with a driver sitting in their from the last stop before splitting, why not just split at that stop?

TLDR: In order to make this work, you'd need the DC units to be autonomous/ATO-controlled locos (at least for short distances) and CBTC/ETCS L3 or better signalling system on the mainline.

An interesting concept, but one that when you look past the 'cool factor' quickly collapses against it's own assumptions.