Two birds with one stone: the importance of segregating services
How to increase both capacity and reliability
If you ask most ordinary rail users, they probably don’t care much about tram-metros, passing loops or through running. Often they have only two (entirely reasonable) requests: ‘I want to get a seat!’ and ‘I just want the trains to run on time!’ The two issues are in fact closely connected.
Sometimes there is an undersupply of seats because the trains are inexplicably too short, but it is safe to say that on many lines we have exhausted the option of lengthening trains. What these lines need instead is more capacity, so they can run a more frequent service.
Regarding delays, an individual train can get delayed for all manner of reasons: technical faults, passengers getting trapped in the doors, geese, train crew being in the wrong place at the wrong time, and so forth. These delays can be reduced in all the boringly-obvious ways you can imagine, although it is of course impossible to run a railway with 100% reliability.
But some delays are caused by a cascading effect: train A is late, which causes train B, which has so far been running smoothly, to be late as well. Train B might be a fast train which gets caught behind a late-running slow train. Or, train B might be timetabled to use the same platform as train A. Or, B might need to cross a junction just after A, which means it misses its slot. Train B in turn might delay train C, and so forth.
The link between capacity and reliability is as follows. Railways face a trilemma: they can choose up to two of frequency, reliability and complexity of services. (‘Complexity’ refers to service patterns like local and express trains sharing tracks, or branch lines that feed into the same trunk line, or passenger trains sharing tracks with freight.) This trade-off exists because, if the railway attempts to operate services that are both frequent and complex, then there are far more opportunities for delays to propagate to other services.
Britain’s railway services have to be frequent, given the demand for train travel, the size of our cities, and the fact that the railways attract more passengers if they run a more frequent service. We also want them to be reliable, given the issue’s salience among passengers, the extremely disruptive impact of cancellations, and the fact that a more reliable railway needs less redundancy, so can sweat its assets more efficiently.
That leaves complexity. Complexity is, in fact, often a bad thing: a mind-boggling array of different service patterns is hard to understand and deters casual passengers. Even when complexity is useful, it is generally preferable to prioritise frequency and reliability: getting rid of a half-hourly nonstop service, so that a half-hourly stopping service that takes five minutes longer can run every 15 minutes, is actually good for the passenger, because the average wait time goes down from 15 minutes to 7.5 minutes.
In other words, the solution to both delays and a lack of seats is often the same: make services less complex. There are several ways of doing this, but this post focuses on one: building infrastructure to segregate different services from one another.
Segregating fast from slow trains
As everybody knows, Japan’s trains basically run on time. Often this is ascribed to the punctiliousness of Japanese culture. The Japanese certainly are punctilious, but this is not the reason their trains are so reliable. Their trains are reliable because lines running different kinds of service are segregated from one another.
The diagram below (known as a stringline) shows a two-track line on which every train stops at every station. Time is on the x axis, and the stations, i.e. distance, are on the y axis. The different colours show trains going in opposite directions.
This type of service is easy to operate reliably. The trains run every ten minutes, so even if one train is delayed by a couple of minutes, then it won’t affect the others heading in the same direction. There is also plenty of spare capacity: you can add more trains in between the existing ones, up to the limits imposed by the signalling system.
Now, add in a nonstop train using the same tracks.
The blue path of the fast train crosses one of the orange paths of the stopping trains, which means that train cannot run. It will also be harder to run the service reliably, because the fast train is getting perilously close to the stopping train: if the stopping train is delayed by a few minutes, then it will cascade over to the fast train. Both capacity and reliability have been reduced by the fast train.
This is not a problem, however, if the fast train can run on its own tracks. Segregating fast from slow trains is part of the reason Japan manages to operate so many trains, so reliably. The most obvious way of doing this is the Shinkansen. Beginning with the most congested part of the Japanese railway network, the line from Tokyo to Osaka via Nagoya and Kyoto, almost all of the fast long-distance services use tracks that are separate from the slower commuter trains.
Although the Shinkansen is the most famous Japanese railway project, it is by no means the only one. Around Tokyo, Japanese National Railways invested heavily in quad- or sextuple-tracking the five main lines radiating out of the city between 1965 and 1982, separating fast commuter services from all-stations trains. The primary motivation for this was capacity in the postwar boom years: depending on the line, this ‘Commuting Five Directions’ operation increased capacity by anything between 50% and 150%. But it also enabled greater reliability. Slow trains cannot delay fast trains, and vice versa.
Passing loops are conceptually similar: you four-track around a handful of stations, rather than along the whole line. Osaka only has one four-track line, which means that the railways around Osaka are much more reliant on passing loops than those around Tokyo. But there comes a point where a line is so busy that something more than just passing loops is required to create enough capacity and reliability.
Likewise, Austria has invested in segregation. The Westbahn is the country’s most important railway artery, going from Vienna to Salzburg via the city of Linz. There are now four tracks on the busiest bit, from Vienna to Linz, and another pair of tracks are under construction between Linz and Wels, the next city west towards Salzburg.
The Westbahn, however, is a very curvy line. Austria has therefore four-tracked it by investing in bypasses for the curviest bits, or in places where adding four tracks in situ would be difficult, and adding four tracks next to the existing right of way in the rest of it.
This has killed three birds with one stone. Austria has increased the capacity and reliability along its main artery, and has built a higher-speed line into the bargain, enabling 250 km/h (155 mph) maximum speeds.
What is noteworthy is that these improvements could be justified in spite of the small size of the places that it connects. The Westbahn connects Vienna (1.9 mn people within 10 km) with Linz (320,000) and Salzburg (250,000), and from Salzburg it goes onwards to Munich (1.6 mn) and Innsbruck (220,000). Although Munich and Vienna are quite big, Linz is the size of Derby. Salzburg is the size of Barnsley. This is because segregating non-stop from stopping services is extremely beneficial, even for cities that are not the size of Tokyo.
Separating traffic from different directions
This is a diagram of the track layout of part of Shinjuku station in Tokyo, the world’s busiest station.
The yellow lines are the Chūō-Sobu line, which runs east-west across Tokyo. The green lines are the Yamanote Line, which runs in a big loop around Tokyo. The orange lines are the Chūō Main Line, which goes from Tokyo out to its western suburbs. The blue lines are for the Saikyō and Shōnan-Shinjuku lines, which run north-south.
We therefore have four lines, going in different directions, which are entirely segregated from one another. A delay on the Yamanote Line cannot cascade to the Chūō Main Line. (In some places there are tracks shown in black which link them together, but these are only used for rolling stock movements.)
In a similar vein, The Netherlands entirely rebuilt Utrecht Centraal station in the 2010s. Because it is roughly in the middle of the country, and has services to nearly every other part of The Netherlands, a delay in Utrecht can easily affect all of the rest of the country.
English Wikipedia has an unusually good summary of the project, which involved redesigning the track layout so that the main traffic flows were all fully separated from one another, inspired directly by the way that services are segregated from one another in Japan. This has, again, enabled greater reliability, and greater capacity.
Historic station layouts were often designed to ensure that every platform track could be accessed from every approach track. Superficially, this seems to improve reliability: if platform 7 is blocked, then the train can go into platform 9 instead. In practice, though, it is much better to adopt the Japanese approach. The new layout of Utrecht reduces redundancy in respect of individual lines, but it means that lines are not interlinked: a train from Amsterdam cannot delay one to Rotterdam, because they are fully separated from one another.
The other superficial advantage of the old Utrecht layout is choice over origins and destinations. To slightly simplify, the layout made it possible to run a direct train from anywhere to anywhere via Utrecht – but many origin <> destination pairs make no sense. There would never be Amsterdam <> Rotterdam trains via Utrecht, because there is already a direct line between those cities. It follows that there is no need to have the infrastructure to support those services.
Even when it comes to more plausible links, there are still good reasons for reducing the complexity of the infrastructure. Of the possible direct A <> B connections which are quickest going via Utrecht, not all of them should have direct services. Good connections can suffice.
Hilversum, for instance, is a big town between Utrecht and Amsterdam, which does not have direct services to The Hague. The infrastructure was designed to make direct services to The Hague impossible, but this does not matter very much: the connection at Utrecht is quick and well-designed, and the reduced complexity enables more frequent services from Utrecht to both The Hague and to Hilversum.
In other words, the infrastructure at Utrecht, and in Tokyo, was designed to operate a particular service pattern, rather than to enable the operation of a wider, more speculative array of service patterns. They have planned the infrastructure and timetable together, which is how a railway can make best use of its assets.
The situation in Britain
These three short case studies all come from countries that manage to run their trains on time, while operating an intensive service. What is the situation in Britain?
In London, there is a fairly high degree of segregation. All of the major main lines coming into London have four tracks, and long-distance trains going from different directions are also pretty segregated because they go into different termini. Things are certainly not perfect: south of the river there is too little segregation between the different stopping services, so they cannot have tube-like frequencies. There is also too much track sharing between regional services to places like Bedford and intercity services to places like Sheffield.
Outside of London, however, the picture is very different. Take Leeds. There are seven railway lines funnelling into Leeds, none of which have four tracks. Three of them could be classified as main lines – the line to Sheffield and London, the line to Huddersfield and Manchester, and the line to York and Newcastle. The remaining four lines mainly host local services to Barnsley, Harrogate, and two lines to Bradford and points west, along with a few faster trains.
All three of the main lines should have four tracks, or two tracks built parallel to carry the long-distance services, at least on their busier sections. With the local lines, there is no need for full four-tracking, but some passing loops would help to enable them to run the express services.
There is also less segregation of origins and destinations than one would like. As the image below shows, the six lines that approach Leeds from the east funnel into three pairs of tracks which means, for instance, that trains from London share tracks with trains from Bradford.
Leeds is by no means the worst offender. Birmingham New Street has five lines entering it, of which four are used for both long-distance and local services. None of them have four tracks, and there is little of the kind of origin/destination segregation seen in Utrecht that would enable frequent, reliable services.
In Manchester, the same is true: only one line entering the city, the line from Stockport and London, has four tracks, and they go down to two on the approaches to Piccadilly Station. The notorious timetable meltdown in 2018 was partly caused by attempting to squeeze services from all over the North into a single two-track bottleneck in central Manchester.
None of the ideas in this blogpost are novel. The railways have four-tracked busy lines and separated traffic flows since the nineteenth century. In recent years, Britain has been doing this: in the 2010s four tracks were restored on the line between Bristol Parkway and Bristol Temple Meads, and Reading station was comprehensively rebuilt to segregate traffic in different directions. These solutions have convergently evolved because they are the correct responses to the problems thrown up by the technology of rail.
What is the solution to this? Through running is a large part: by putting suburban trains in tunnels, it will free up tracks and platforms at stations like Leeds, enabling greater segregation of services from different directions. But through running will need to be paired with investment into segregating long-distance trains from local services. This means adding four tracks next to existing lines, or investing in bypasses along the lines of Austria’s neue Westbahn, which has the additional benefit of enabling faster services.
I have no illusions that this will be cheap. But it needs to be done to make the best use of our railways. Most of our cities have extremely dense railway networks, relative to their size: Manchester has 20 railway lines within 20 km of its city centre, to serve 2.8 mn people, compared with just 16 in Berlin, which serve a population of 3.7 mn. That means there is enormous opportunity to give our cities the best public transport in the world, without having to spend extremely large amounts of money on building extensive metro networks. Rather, it requires spending merely large amounts of money. If we want good public transport, with all the benefits that flow from that, we have to be willing to pay for it.