The only way to have express and stopping trains on the same track is to have frequent passing places. Scheduling and signalling get fiendishly complex. On somewhere like the underground, where building those extra passing places is really hard it just isn't worth it.

This is one of the big missunderstandings about the HS2 line. Politicians focus on the shorter journey times, but the big win is actually increased capacity. HS2 will take the express trains off the normal line. With just HS trains on that line you can run more of them than you can if you have stopping trains sharing the track. In addition you should actually be able to run more stopping trains faster since they don't need to be fitted in around the expresses with no extended stops waiting for a delayed express to pass at a station or passing place.

A high speed line with passing places is a non-starter.

The signalling barely works at the best of times. It's literally a museum of technologies, from the pneumatic to the electronic, and it's one of the most common points of failure on the network.

But when it's working it's been improved to the point where there's almost no spare line capacity at peak times.

And if you're making new tunnels it's so difficult and expensive to get a Tunnel Boring Machine into place that it makes no sense to bore short sections.

Crossrail is supposed to be London's east/west express line. Obviously it's not ideal for District Line users, but it should free up some peak hour capacity for Circle and Hammersmith & City journeys which may translate to fewer District/Circle passengers.

There's also talk of a north/south Crossrail 2.0, but that's unlikely to happen for decades.

There are overground express sections on the Piccadilly (District) and Metropolitan (Jubilee) lines but they all keep the lines separate.

There may well not be room for a tunnelled express in the central area. If it were up to me, I'd consider installing a good urban tram link from (say) Earls Court to Embankment.

> A high speed line with passing places is a non-starter.

Yep. Not sure if I miss-worded something but that definitely wasn't what I was implying, just the opposite. That the whole point of HS2 is to separate the fast and slow trains so you don't need passing places (either at stations or otherwise) and so can actually run more trains on both the new HS [line and the existing line.

Re HS2: I am familiar with the capacity-and-speed-difference argument, but I recently found out[0] that the WCML, MML and ECML are all already four-tracked, so you already have separation of fast and slow trains. Is further separation really still worth it?

[0] I don't live in the UK, so I didn't know that yet despite being a railway nerd.

The WCML has to accommodate very long distance express trains, regional trains, suburban/commuter trains, and freight.

I think you can read section 2 of [1], and the start of section 3. With enough trains, you can still fill up a four-track railway.

[1] https://assets.publishing.service.gov.uk/government/uploads/...

Freight 60-80mph constant speed. Stopping and semi-fast passenger up to 100mph but varies a lot. Fast passenger between 125-140mph. Fast-passenger is moving to HS2 albeit at 10 times the cost it would be to do the same thing in France.

(For easier reading and understanding: freight 100-130 km/h, stopping/semi-fast up to 160 km/h, fast 200-225 km/h.)

Could you have odd / even trains that miss stops?

Yes, you can do things like that, but that ends up with you having two different sets of stations that are not easy to get between. The normal pattern is express trains that stop a few times with stoppers that "fill-in" between, but there will be different patterns of stopping to try and optimise for different groups of travelers.

Scheduling/timetabling is a wickedly hard problem, especially in a system like a railway, it's the sort of thing people get Maths and CS PhDs in. The contstraints that you have to solve are complex and interconnected and are part of a bigger network, and you are also trying to please a lot of people with very different (often contradictory) needs.

Yeah I've previously worked on some constraint programming tasks and it was fun but mind-boggling.

It was one of my favourite course at uni, I'd love to get back into doing something like that at some point.

I didn't use them in the end, but the Google OR-Tools is a pretty amazing resource for this. I don't have a CS degree, so just seeing the names of the algorithms was helpful ha.

https://developers.google.com/optimization/routing

This is already the case for some of the very low-volume stops on that line. For non-express trains a combination of the following is currently (or at least was pre-covid) in use, with the mix varying throughout the day:

  * stop more or less everywhere
  * stop only at larger stations
  * even/odd stopping at small stations

The scheduling already seems pretty clever, and that's just from observing as a passenger. I suspect behind the scenes there's a whole lot more to it that's not obvious to someone who is just trying to get to work.

Yep, there are whole departments of people who work hard to make the scheduling on railways work as well as possible. People always like to complain, but they always ignore the fact that the timetable has to try and satisfy a huge number of people with often very different needs, on networks with some very fixed constraints that are expensive and difficult to change.

Not without passing places, because you can't miss a stop without the train in front of you also missing it, at least without an eye watering body count.

Depends on how closely you are trying to run those trains! If they are both skipping every other stop, they will both have the same mean speed, and thus shouldn't have to pass one another if there is sufficient initial spacing.

For example, in the simple case where the stations are evenly spaced, if train B arrives at station 1 at the same time that train A arrives at station 2, then they will leave at the same time and arrive at stations 3 and 4 respectively at the same time, and never catch up with each other.

Yes, this was the sort of thing I was thinking for London. Not "express" per-se, but "missing stops" (i.e. just like what happens when stations are closed for platform maintenance during normal ops).

You can definitely do that.

However on something like the underground that would complicate the "turn up and go" approach that most people take to the tube if you start skipping different sets of stations on different trains.

It's something that's done, but I'm not sure how well it would scale on the tube if you did it a lot. Each train would need to miss roughly the same number of stops to prevent blockages or you need more passing places. With the small gaps between many tube trains the margin of error for scheduling can be very small.

I think it's one of those things that works, but the advantage for the vast majority of people would be pretty small compared to the complexity and added fragility it would introduce if you tried to do it a lot.

This creates a very significant passenger challenge. How do the customers know which train to get on and what if they get on the even stop and need to get off an odd stop? They have to change train. If they can't pass each other then they don't save much time.

In NY, the expresses are easier to understand because the rules are very simple.

They used to do it in Chicago for a long time. Trains and stops are labelled A or B and some stops would be labelled AB if evens and odds both stopped.

For large parts of London's underground (and overground[†]) it is simply the case that it wasn't really designed — lines were slapped in willy-nilly by disparate commercial interests with relatively little forward planning, and only later became something like a coherent whole with some consideration for coordinated thinking. Once the lines are in, upgrading them is more difficult than building them in the first place, particularly if you don't want to stop service for large parts of the improvement work.

> just like in any other number of countries around the world

Other cities had extra benefit of hindsight, being able to design around the problems identified in older systems (particularly London's).

> They could have used the same rails, no need for a separate line

You at least need passing places around stations in practise. In theory* you could have many extra points and pass trains between the existing two lines to work around each other instead of keeping one line dedicated for each direction (as is the case for most of the track length) even at stopping points like stations but that gets complex to manage, has more moving parts (which are difficult to maintain in the confined space), would considerably slow down flow at busy periods as the trains can't move as fast over the points (particularly if they may need to switch line at them) and will spend time waiting for an opposing train ahead to switch out of the way, the tunnel around each change point needs to be wider (for the train partly, unless you redesign them too, for maintenance even more so), … It might work for a small number of non-stop trains worming their way through the system around the majority stop-start services, but that number of services would be so small to the point where the investment would not be nearly worth the small overall gain in reduced journey times.

[†] only about 45% of the line distance of the current tube is actually underground[‡] [‡] though that includes large overground sections in the outer zones if you are only considering central London I suspected that %age is considerably higher

Thanks for the insight @dspillett, makes sense.

The majority of the tube network is pretty much already at peak capacity, with trains on the Victoria, Jubilee and Northern being full autonomous with moving block systems as separation distance can now only be maintained through autonomous systems. I think an express service would be wonderful but it would require careful orchestration.

I think it might be worth increasing the line speed through signaling upgrades and more automation.

They are effectively building an express system as seperate lines. E.g. the new Elizabeth line is effectively an express Central line.

Isn't the problem with automation on the tube less the technology but more the unions ?

No, it's the technology or perhaps the companies supplying it (unions are only ever a problem in the US - or maybe they're never a problem and there's only a propaganda system against them in the US). London Underground are currently in the process of literally the third attempt to upgrade the signalling on the subsurface lines (i.e. including the District) - the previous attempt by Bombardier failed outright (as London Underground were aware it would from the early days, but they were politically obliged to wait until the company admitted as much), and the one before that (a similar story with New Labour ideology-driven PPP) was also abandoned.

Full automation—without drivers—would typically require platform edge doors, which are expensive and troublesome to introduce without disrupting service, and very difficult on some of the curved platforms.

The benefit for the cost involved really just isn't there; while drivers aren't cheap they aren't impossibly expensive in comparison with the average number of passengers per train.

The unions don't help but they reflect largely the interest of their members. Who doesn't want constantly improving pay and conditions and no possibility of redundancies? In effect you have to bribe the train union / its members to land automation that they perceive as potentially weakening their future bargaining position. They don't care about notional safety -- or at least no more than most individuals do in practice -- they care about their personal outcomes. One common bribe for drivers is time set aside for "education" that can be done at home and isn't assessed. Say you give drivers 30 minutes per week for this "education", well, now you've got 0.5 hours per driver of driving needed, you can't hire the necessary drivers quickly enough to cover that, so, you pay overtime to cover it. Lo and behold every driver is keen to do the extra overtime. Now your workers are doing the same job for more money, but when asked about it their union can say it's not that you paid more money, it's because you agreed they needed time for appropriate "education" which is quite different and will contribute to improved safety.

My favourite example of how unionised workers whose union invariably presents their preferences as "safety" care little about actual safety was signallers. Historically some mainline rail signallers (same unions) in the UK worked 12 hour shifts at outlying boxes. So you maybe do 3 x 12 hour shifts = 36 hours that's a week's work or take an extra as overtime for 48 hours. Roster maybe a dozen people to work a box and it remains open 24/7. Annoying scientists said, wait a minute, humans don't remain useful and attentive workers for 12 straight hours, especially at night which explains these incidents where a signaller makes a grave error at like 0400 after 10 hours at work. So the safety regulator wants to limit shifts to 8 hours like for air traffic. The unions are apoplectic because if you're a member working a mixture of 3x12 and 4x12 and now they want you to work 5x8 that's up to two days per week extra. Suddenly railway safety vanishes as a concern...

It is possible to make gradual progress. When I was born there were still guards on tube trains. Why? Well there had always been guards on tube trains. Today of course London Underground does not have guards. It just took a long time to make it happen, with I believe the Northern Line being last to stop having guards.

It is mainly due to an enormous expense and wanting to keep everything running during upgrades.

If you are already costing the taxpayer X million per year and then you want another 400M for an upgrade, are you likely to get it?

Express services on shared track isn't common on metro systems. NYC's express subways all use dedicated track, for instance. The tendency to not keep to a strict timetable and the close-running of metro systems makes it hard to slot express trains into gaps between local trains properly.

Trivia. The timetable, or Working Table, for all London Underground train movements is here https://tfl.gov.uk/corporate/publications-and-reports/workin...

Crazy that the time between trains is measured in seconds

Even with a strict timetable you need extra spacing between trains and potentially slowdowns to allow passing to be done properly. These systems on busy lines tend to run trains as close to each other as possible without causing safety issues so there's no buffer room. And if something goes wrong for any reason the whole system starts getting cascading delays.

> The tendency to not keep to a strict timetable and the close-running of metro systems makes it hard to slot express trains into gaps between local trains properly.

To be fair, "close-running" is not a word that tends to be associated with the District line ... "signal failure" is, however ! ;-)

The central section (including the Circle Line) has a service interval of 2-2½ minutes, depending on the time of day.

The rest has roughly 2-10 minute intervals.

One of the things you need to remember when thinking about the London Underground, is that it was a scheme that was ultimately unfinished. There were great plans between the world wars to extend it in size and functionality but ultimately these were all stopped by WW2 and an impoverished UK couldn’t afford to do them afterwards. It was planned to extend the network well into Surrey, Kent and other places.

They also chose not to build many of the suburbs that the extended lines would have served, and instead to keep those areas rural as the Metropolitan Green Belt and to build new towns further away from London.

I grew up in the resulting green belt with restricted residential development of Metroland. Of course in practice what happens is that many people who "must" work in the City but have the salaries that come with that, choose to move, especially when they have young families, out to "the countryside" where the Metropolitan Line ends at e.g. Amersham.

There you can walk from your house (with a view of woodland and fields across the valley) to a London Underground station (you probably don't actually catch an Underground train, even the express takes too long, you catch a "normal" commuter train serving the same station but these days it's the same price because it's the same system) in the morning and the reverse even evening. And your children grow up away from the noisy polluted city, but not so far away that you can't take them to see a stage show or one of the museums on a whim.

Which is nice for them, but hardly screams "sustainable" as a society.

There are not many/any high-frequency metro systems that can do this on the same rails. At peak times, when you could do with the expresses the most, there can be stopping services 1/2 minutes apart on average, which means non-stoppers are simply not possible.

On regional railways, where the service might be every 30 minutes or longer, it is a different prospect since you can send the stopper out immediately after the express and it gets 30 miuntes to get out of the way.

The tube has shorter times between trains than the New York subway. For example, Picadilly/Central/Victoria line trains run at <120 second headways on the central section. You couldn't control the gaps between trains precisely enough to demerge and remerge trains once you account for random factor of passenger loading and unloading and changing gaps between trains as they accelerate and decelerate between stations. Better to build a whole separate line, in fact this has been done already, the Victoria line is pretty much the express version of the Piccadilly line and likewise Crossrail will be the express version of the Central line.

> random factor of passenger loading and unloading and changing gaps between trains as they accelerate and decelerate between stations.

Solution: Never arrive early, never depart late, and keep the stopping durations fixed, at the longest time possible (which is one quarter the time between the two closest neighbouring stations).

That's not actually a solution. What do you do when someone holds the door open? Unless you have someone next to each door ready to boot such people out of the train, you simply can't stop yourself departing late.

Have station doors, and slam shut them before the train doors close.