As crazy as it sounds, one way to make HS2 even remotely affordable might be to put the whole thing in a tunnel. The current Y-shaped HS2 plan to both Manchester and Leeds has 531km of track and a probable cost of £85 billion or £160 million per km. The Groene Hart Tunnel in the Netherlands, built for the Thalys high speed line, cost £77.5 million per km. Less than half. The £160 million per km does include stations, but as we shall see, the buried line could still be built for far less.
A note on the numbers. The 7.16 km Groene Hart Tunnel was built in 2000 for €426.5 million or £338 million using purchasing power parity exchange rates or £47.2 million per km. Using a UK construction cost inflator, this is equivalent to £77.5 million per km today.
And the Groene Hart Tunnel is not a cherry-picked example. It is single bore, large diameter, TBM-built tunnel for high speed rail to run at speed. (Because of the air pressure built up in front of a high speed train, a high speed tunnel must be much larger in diameter than just the train itself.) In fact there is a UK example that was built for even less. The North Downs Tunnel dug for HS1 cost £28.2 million per km in 1999-2000 or £46.25m per km today. But it was driven in such favourable geology that it didn’t even require a TBM and is thus not a reasonable cost comparator.
In addition to tunnelling, the rails, the electric supply and signalling system need to be installed, but these aren’t actually that expensive once all the other engineering work is done. The rails, switches and catenary run around £2.5 million per km and a signalling system sufficient for automatic train operation costs about £2.5 million per km. Also, there needs to be some land acquisition for vents and emergency escapes, and that is hard to estimate, but I will put in £1 million per km with the expectation that there will be a surface structure every few kilometres or so.
It might seem baffling that a tunnel could possible cost less than a surface route. But tunnelling itself is not actually that exorbitant. The tunnel boring machines operate with very few workers, and after the starting pit is dug and the machine set up, the main ongoing operation is feeding in new tunnel liner segments and taking away the spoils. Soil conditions need to be monitored continuously and cutter wheels replaced, but that is accomplished with a small crew. The sticker shock comes from the stations. The Crossrail 2 estimates also show this. For the metro-style line, the cost of the stations is £5.32 billion is while the tunnels are only £1.83 billion. And reviewing the HS2 plans show what an undertaking building a totally grade-separated railway on the surface actually is. Every time a road, railway, stream or chicken needs to cross, there must be an overpass or an underpass. And high speed rail requires relentless cutting and earthmoving to give it the wide-radius curves that it needs. All this on top of the cost of the land.
The disadvantages of a tunnel:
- Running a train in a tunnel at speed is less efficient than en plein air. The sleekest train in the widest tunnel is what we are looking for. As well as a system of baffles that equalizes the pressure between the two halves of the tunnel at the most opportune time.
- No view.
But once we get past the essential barminess of building an England Underground, there are a number of advantages of a tunnel:
- There is minimal interference in above-ground life. No noise, no visual obstructions, only air vents and escape routes that can be disguised as houses, barns or discount carpet warehouses as surface conditions require.
- The route can be direct. As the crow flies, or rather as the mole digs. A straight shot from London to Manchester is only 270km. A Y-shaped tunnel from London to Manchester via Derby plus the segment from Derby to Leeds is only around 390km compared to the current 531km HS2 plan. A 26% cut.
- No interference from weather or the wrong kind of leaf falling on the tracks.
- Automatic operation. A driver really isn’t necessary for high speed trains above ground and certainly not below ground. Automatic operation improves the economics of running shorter trains more frequently which is better for passengers and for station design.
- Putting the train in a tunnel allows it to work more like metro. A costly item in the HS2 plans is expansion of large terminus stations in London and Birmingham. Railway planners elsewhere are busy trying to get rid of terminus stations and go to through-running stations because they take up so much less expensive urban land.
- In London, where there will be several stations, the high speed train will have to slow down to conventional train speeds leaving surplus space in the large diameter tunnel not needed for the pressure wave. Thus there will be space to also feed a suburban rail service through the tunnel as well creating a Crossrail Lite type service.
The current HS2 plan:

A tunnelled version of HS2: HSTunnel. This would be Y-shaped like the current plan but entirely in a large diameter bored tunnel designed for 300kph operations with automatic train control and short headways:

Thus far most countries have found that high speed rail has provided the most benefit over the 300km to 600km range. So I have pencilled this out to focus on the London to Manchester and London to Leeds over London to Birmingham; although, the West Midlands will also benefit from a rail connection at Northampton for London trains and the Derby to Leeds segment for routes to the north:

(The HSTunnel route is distorted on this stylized network map, but as shown in the map above, the route from Manchester to London is nearly straight. The lightest blue line on this map is the route of the current HS2 plan.) Obviously this HSTunnel route isn’t the only possibility. In might in fact make more sense for the new line to intersect the old network at Milton Keynes and not Northampton. And the bifurcation point could be Nottingham instead of Derby, but I chose Derby because it seemed to integrate better with the existing network, allowing Birmingham to Leeds trains to take a more direct route. Obviously this plan anticipates using high speed trains on the conventional network that would enter the high speed system, switch to automatic control, and then proceed at speed. The current HS2 plan anticipates building and expanding traditional terminus train stations. (A terminus station is where the line ends at the station so trains must leave by reversing down the same tracks that they arrived on.) This shows the expanded footprint of Euston Station:

A new Curzon Street Station in Birmingham:

And an expanded Piccadilly Station in Manchester:

Terminus stations require expensive urban land and block cross-town travel, and railway planners now are trying to get rid of them or retrofit them into through-running stations. The trouble with terminus stations was an impetus for the creation of the RER system in Paris and the original Crossrail scheme in London:

Through-running makes it easier to have multiple connections to the rest of the transportation systems. Air transport can only connect to the rest of the transport system at particular points, airports, but trains ought not to have that limitation. They ought to be able to provide multiple connections to the rest of the transport system. The following map shows how an HSTunnel line, in light blue, might intersect with the rest of the transit system in London:

The main London stops would be at Euston Station and London Bridge Station in what would be effectively large metro stations. Turnstiles would prevent ticketholders from proceeding down to the platform too far ahead of their trains to avoid overcrowding the platform, but otherwise, the operation would be very much like any other metro. At peak time, trains would arrive every two or three minutes, wait for 20 seconds for alighting and boarding, and then depart. At least one report noted that Japanese high speed rail patrons are much better and getting in and out of the train quickly compared to Europeans who have a tendency to lollygag, but passenger education and possibly a third door per car should help with that.
Also shown on this London area map are points at West Hampstead and South Bermondsey. These are interconnections with the surface rail network, essentially ramps up and down, to allow the surface system to feed into the tunnels. The high speed trains would have to slow for the London stops freeing up space in the large diameter tunnels no longer needed for the pressure wave developed by high speed running. The tunnels would be large enough to fit four trains in tight configuration allowing for a suburban line to run from the southeast sector to the northwest sector in continuous operation on its own tracks separate from the high speed system. The result is something like Thameslink. Intermediate stations would make this service more useful, and I have penciled in stations at Holborn and Blackfriars. Those would only be for the suburban trains and would be paid for out of local transport funding like Crossrail.
There is no particular necessity for an HSTunnel route to stop at Euston Station or London Bridge, other station pairs or triplets would work. But Euston or St. Pancras have the advantage of being next to numerous tube lines as well as Crossrail 2:

Having the second station at London Bridge leaves open a connection to HS1 south of the river and continuous operations from Paris to Manchester that would pass directly under London instead of skirting it. The current HS2 plan has mooted a connection to HS1 between Euston Station and St. Pancras, but it is kludgy and does not serve most of the London market well:

Through running and the complete integration with the surface network would require big changes to the franchises currently in place, but a tunnel plan would take a long time to build which would give plenty of time to plan for the change.
This plan accomplishes all of what the current HS2 plan does. With trains operating at a maximum speed of 300kph, London to Manchester is about one hour and London to Leeds in about one hour fifteen minutes. Because the current network lacks a direct connection between Derby and Manchester, the HSTunnel route would also greatly speed traffic between Manchester and Derby, Leicester, Nottingham and Peterborough.
Currently there are around 33 million train journeys per year between London and the West Midlands, East Midlands and the North West. With such fast service, there is good reason to believe that most of these trips will transfer to the HSTunnel Line plus a significant number of car trips. Thus 33 million journeys per year ought to be a good estimate of the initial usage of the line.
Cost: A Lot
At Groene Hart Tunnel costs plus the installation of the rails, signalling system and power supply, the line would cost £83.5 million per km for a total cost of £32.57 billion for the 390km long route. The stations and rail connections are estimated at £250 million each in the smaller cities and £0.5 billion in the larger cities:
North Leeds Rail Connection | £250 m |
Leeds Station | £250 m |
North Sheffield Rail Connection | £250 m |
Sheffield Station | £250 m |
North Manchester Rail Connection | £250 m |
Manchester Station | £0.5 b |
Derby Station | £250 m |
North Leicester Rail Connection | £250 m |
Leicester Station | £250 m |
North Northampton Rail Connection | £250 m |
Northampton Station | £250 m |
Hampstead Rail Connection | £0.5 b |
Euston Station | £0.5 b |
London Bridge Station | £0.5 b |
Bermondsey Rail Connection | £0.5 b |
Operations Yard | £1 b |
Total | £6 b |
Train cost are estimated at £50,000 per seat, so a 200m emu with 400 seats would be £20m per trainset or £400 million to buy 20.
Tunnels | £32.6 b |
Stations | £6.0 b |
Rolling Stock | £0.4 b |
Sub-Total | £39.0 b |
Design and Management 10% | £3.9 b |
Contingency 10% | £3.9 b |
Total | £46.8 b |
Thirty year gilts are now yielding around 1% meaning the crude carrying cost of the HSTunnel would be around £468 million per year, interest only. If the line does attract 33 million journeys per year, the capital carrying costs would be £14.18 per trip on average (though it ought to be skewed by distance traveled). Actual repayment of the debt must come from increased ridership or debt attenuation due to inflation. I do not know whether an average £14.18 charge would be realistic after covering operating costs. Automatic operation ought to decrease operating costs, but the integration with the rest of the network will require train drivers and plenty of employees so the decrease might not be dramatic. Likewise maintenance. Having the track in a tunnel will protect the track and equipment from the elements which also ought to decrease maintenance costs. However, high speed operation requires very high reliability and thus high preventative maintenance.
The ultimate capacity of the tunnel is limited by train capacity and train frequency. At 300 kph, trains every two minutes would leave 10 km between each train. This is fine because the stopping distance of a train at 300 kph is less than 10 km. (The Shinkansen emergency braking system can stop a 275 kph train in 4 km and experimentally a 360 kph train in 4.3 km using the most advanced braking controls and physical wind resistance brakes that pop up at the end of each car.) And for the main stops – at which all trains would stop, eg Manchester, Euston and London Bridge – all the trains must slow in series, decreasing the distance between them, but also decreasing the braking distance. For intermediate stations like Derby, Leicester and Northampton where only some trains would stop, turnouts would need to be provided for the trains to operate at maximum frequency. Thus all trains would have to slow to the maximum speed of the point, 200 to 240 kph for the high speed points, at which the stopping trains be switched to the station tracks. Since all trains would have to slow down in series to the maximum speed of the point, the required trains separation beyond stopping distance would be maintained.
Two minutes between trains implies 30 trains per hour. At 400 seats per train, this is 12,000 passengers per hour per direction. At 18 hours per day of train operations and 365 days per year, the ultimate capacity of the line is 78.8 million journeys per year per direction. This is assuming that each train covers the entire route, which for a Y-shaped line is impossible, so there will be even capacity in trains that do not run the full length of the network. This is for relatively short 200m trains. While it would be truly crazy to tunnel all the way to Scotland and also a waste of a scenic train trip, the natural second phase of this plan would be to extend the arms of the Y north:

Because of the directness of the HSTunnel route, a two hour Flying Scotsman from London to Glasgow would be possible with an extension north.