What if sleeper trains, but high speed?
Originally published on Observable in Febuary 2020 — I’m just consolidating my blog posts in one place.
Sleeper trains are a very time-efficient way of getting from place to place. They’re gaining in popularity in the context of the climate emergency, especially amongst those who want to reduce their carbon footprint. But they tend to be substantially slower than high-speed daytime services on similar routes.
I’ll aim to find out whether there are opportunities for new high-speed sleeper services to play a part in rail mobility across Europe.
Introduction
The expansion of high speed rail in many regions of the world has really transformed medium-distance travel. In particular, Europe has seen a huge development of the network since the first line of the French LGV network opened in 1981, to the present day state shown on the map below, which includes significant coverage across Germany, Spain, Belgium, the Netherlands and Italy. Similar developments have taken place across the world, notably in China and Japan. Even the USA is starting to catch up.
Across many city pairs, such as London-Paris and Madrid-Barcelona, high speed rail has taken significant market share from airlines — even reaching a 23% market share between London and Amsterdam, despite requiring a change of trains in one direction. On short journeys (in the hundreds of kilometers), high-speed rail is competitive on time once you take into account the time taken to travel from city centres to airports on either end and requirements for check in times.
Despite concerns about a decline in popularity which led Deutsche Bahn to withdraw all of their remaining sleeper train services in 2016, sleeper train services are now regaining popularity, especially in light of the climate crisis and the resultant flygskam phenomenon. ÖBB’s railjet service, the backbone of the sleeper train services across Europe, has recently announced expansion. However, the vast majority of rail sleeper services operate at significantly lower speeds than their daytime counterparts, using older rolling stock.
However, there is an opportunity to combine the high-speed rail and sleeper train concepts. Reducing the journey times on a sleeper train by an hour or two isn’t normally considered to bring as large a benefit as a similar time reduction on a daytime service, but the real benefit of these services would be in opening up new routes which are not already served by sleeper trains, partly because the journey is considered to be too long. For instance, the longest duration of sleeper train service operated by ÖBB Nightjet is the Zagreb-Zürich sleeper, which departs Zagreb at 18:38 and arrives in Zürich at 09:20 the following morning, a duration of 14 hours 42 minutes. Most European sleeper services have much shorter durations, allowing for a late evening departure and an arrival in time for a morning meeting or a full day of sightseeing. High-speed sleeper trains therefore present the opportunity to cover much longer routes in a reasonable duration night-time journey.
China Railways is using high-speed sleeper rolling stock between Beijing and Shanghai (their CRH2E stock has a top speed of 250km/h, although they don’t exceed 160km/h in sleeper service), so there is a precedent for such services.
Working out the potential
So this all sounds like a great idea — but can we find some data to see how these services might stack up? Let’s try to answer the following question:
“Which routes in Europe (a) have sufficient travel demand to support a high-speed sleeper train service and (b) have a suitable journey length for a high-speed sleeper train service?”.
It turns out that Eurostat have some great data covering air passengers carried — which I’m going to use as a proxy for travel demand, since I’m principally interested in opportunities for mode shift from air to rail. And then Deutsche Bahn’s journey planner will help us find out the journey time lengths for daytime travel. I’m going to look only at the time spent on trains, rather than in connections, because we’re interested in planning new routes here.
An interesting thing to observe here is that there are many journeys where the high speed rail route takes a short time, yet still have a lot of air passengers. See the below table for some of the highest numbers (bear in mind, though, that this is 2017 data and so these may well have reduced). This definitely demonstrates that there are opportunities to mode shift further to rail on existing routes, but that isn’t our main focus here.
Instead, let’s move on to looking at the busiest air routes with a rail journey time of 8–15 hours, which is more typical of a sleeper service. The map below shows these, again restricted to those routes carrying more than 1 million passengers per year.
There are some definite clusters going on here — and the candidate routes are largely in Western Europe. This is down to a couple of things: a combination of high air traffic and good high-speed rail infrastructure — but might also hide some good routes in Eastern Europe, where Deutsche Bahn’s timetable data isn’t as good. However, we can at least start to plan out some possible routes based on this analysis.
There are, of course, complicated techniques I could use to optimise this routing, but I can’t be bothered, and I’d rather just pick some obvious wins here. Once I’ve picked the routes, I can then see what the total addressable market for those routes would be, basing this once again on the demand figures that we have, but this time looking at demand across all the city pairs that would be served. By estimating a required 500 passengers per night that would be required to make a service viable, we can then calculate what percentage of the market would need to be taken over in order to make the service work. Let’s see what that comes up with:
Well, there you have it! A series of routes where a sleeper service would only need a small percentage of the overall market in order to make sense as a service (of course, this doesn’t necessarily mean that those services would work in current market conditions — and note also that the total addressable market is double counted on some of the above routes). But there’s definitely potential there for a network of high-speed sleeper services across Spain, France, Italy, Germany, Denmark and the UK.
Potential barriers
There are plenty of barriers to launching such services. Some obvious ones include:
- Compatibility of technology. The rail network across Europe has wildly varying standards, including but not limited to: electricity supply voltages, signalling systems, loading gauge, and track gauge. One of the results of this is that finding rolling stock (particularly for motive power) that can work across national borders is more difficult. While dedicated high-speed lines are more standardised, most high-speed services require some distance to be covered on the conventional rail network, which is more diverse.
- Border control requirements. Largely to do with any services entering or leaving the UK (and not because of Brexit!). As the UK has never been part of the Schengen area, passport checks are required on entering and leaving the UK. In the operating concept of juxtaposed controls, these checks take place before boarding trains, requiring UK passport controls to take place at Paris, Brussels, Lille, etc. and French passport controls to take place in London, Ebbsfleet and Ashford.
- Security requirements. Some rail routes (in particular, the Channel Tunnel and routes in Spain) require security checks to be conducted on all passengers before boarding. This could cause challenges in requiring more infrastructure to be installed at stations running such services.
- Legislative, political, and business considerations. Despite a movement towards rail liberalisation in Europe, most cross-border services are run by national railways, joint ventures between them, or companies owned by national railways. There are many approvals required to launch a new service and this would likely make launching a new service difficult. Track access charges vary wildly and could completely change the economics of such a service. Potential for state subsidy could be an opportunity.
- Suitable rolling stock. Finding high-speed sleeper stock is challenging.
Limitations in this investigation
- Connecting flights. The air demand data used covers the number of passengers carried on each route. This includes connecting passengers, which is likely to have inflated the figures for origin-destination pairs which include hub airports like Amsterdam, Frankfurt and London.
- Missing rail data. Missing rail data, particularly in Scandinavia, means that there could be more potential routes which have not been covered.
- Islands. I have excluded from this analysis any consideration of journeys involving a ferry trip — which includes routes like London-Birmingham and to the Italian islands. Island trips would add another level of complexity like new physical infrastructure or trains being carried on boats, so I haven’t considered this.
- My own carelessness. This has been a fun little project, not a piece of serious academic research, so I’m bound to have missed things.
Conclusions
Despite the difficulties likely to be faced in launching new services, there is clearly a large market which could be targeted by high-speed sleeper train services. With the ever-looming climate crisis, we should be looking for all alternatives to heavily polluting air travel, and I think this bears investigation. I wouldn’t be surprised if we see higher speed sleeper train services on some of the corridors above in the years to come!
Want to see the code used for data analysis? Check it out here. Questions? Comments? Join the conversation on Twitter.
Originally published on Observable.
