I have a maglev train starting in Paris, going through Siberia, bridging over the Bering Strait, and running down the West Coast to San Francisco. There are other major lines splitting from this one, but this one is simply "vein of the world" in my settings. It gets you from Paris to San Francisco in 40 hours for a personal ticket price in economy class of 200 Eur.

For the sake of my question, do not scrutinize operation of a maglev and the maintenance or construction of railroad in Siberian climate. Let's assume the rails are always maintained and under heavy observation.

Problem and Solution

The train can travel up to 500 km/h. As we know trains are heavy and a big portion of the track length is spent accelerating or decelerating. I came up with way how to minimize this by creating auxiliary track near desired cities (for example in Europe it is only Paris, Berlin, Warsaw, Minsk and Moscow), where the main train will slow down to 200 km/h and an auxiliary train will match the speed and create bridge between them, allow people to transit. I imagine this creating a transit window for around 20 minutes, then they will disconnect and auxiliary train will deliver people and goods to train station. That will need two parallel tracks running for about 100 km (5 min reserve before transit and after) and then the aux train deliver the people into the city (that might take another 30 minutes, depends how far are transit points from the city and its train station). Like this I can make the Paris-Moscow trip from current 40 hrs to 12 hrs.

Regardless of price, is this transit system sound? I can imagine I might have little problem with safety committee. I mean a person stuck between doors is a dead person. Mishap in speed of trains (Flexible docking clamp can compensate only small abbreviations) can be fatal to hundreds of people when the trains are docked.

Are there any other really major red flags in this system? Is there safer way to make the transit without stopping the train?

  • 4
    $\begingroup$ Standard advice: wait 24 hours before accepting an answer. Unsolved questions attract more attention, and we have users in all time zones who can contribute. $\endgroup$
    – L.Dutch
    May 7, 2020 at 7:31
  • 2
    $\begingroup$ I recall lots of supertrain studies in the 1970s and 1980s using classic tracks, maglevs, train-like vehicles, single lifting-body vehicles, U-shaped guideways, etc. The big problem they couldn't overcome wasn't speed or energy usage -- it was track-switching. A 500km/h vehicle requires a physical switch over 150m long for a comfortable ride at speed. Without the convenient ability to switch tracks, you are limited to point-to-point line (bridge carrier) markets; cannot form a network that meets different demands, and cannot reach trip endpoints. The mode change reduces the speed advantage. $\endgroup$
    – user535733
    May 7, 2020 at 12:53
  • 2
    $\begingroup$ It took me a few readings to understand that you want to COUPLE trains moving at 200 kph. Put the gimmick in bold in the first sentence when you write future posts, please. What we have here is called "burying the lede" or "burying the good part in fine print". Perhaps "transfer" in the title would be clearer than "transit" $\endgroup$ May 7, 2020 at 15:43
  • 3
    $\begingroup$ I would question your speed limit of 500 km/h. Even non-maglev trains can reach speeds above 500km/h. -> en.wikipedia.org/wiki/Project_V150 $\endgroup$ May 7, 2020 at 15:57
  • 1
    $\begingroup$ Maglevs are fun for fiction, not very realistic for real life. The best high speed trains run on normal rails that are well groomed, whereas maglevs float on a cushion of money. $\endgroup$ May 7, 2020 at 16:56

9 Answers 9


Stops aren't that big a deal.

In fact, in the vast arctic, your stops won't be for passengers - but for crew.

Take a commuter train; a stop adds about 1 minute. You can compare the schedules of express and local commuters and that's about what it amounts to. In your case, you have a great deal more acceleration to do. But you climb through the first 130 kph even faster than a plain old commuter train, and get to 300 kph acceptably fast. Most of your accelerating is over 300 kph, but you're moving at a fine clip during that time, so it doesn't cost you that much time. The upshot is you can expect stopping proper to add 7 or 8 minutes IMO.

However, because of what I'm about to say, stops will be more of a "big production"; even with staff cueing people when to move, you won't have a 1-minute stop - it'll be more like 7-8 minutes of fussing and fiddling.

But have fewer of them.

Take a high speed overnight train from NYC to Chicago. That train is not going to stop at 125th St, Yankee Stadium, Yonkers, Hastings-on-Hudson, etc. It'll fly all those stations at speed, and make its first stop at Albany, the end of the short-haul regional district. If you're at Yonkers and want to ride it, you either backtrack to Grand Central, or you take the commuter to Croton-Harmon and change to the regional to Albany. There you change to the fast train, and your stops are:

  • NYC
  • Albany (transit hub; end of short-haul service from NYC)
  • Buffalo (big, end of intra-state service, transit hub for Ontario)
  • Detroit (big; beginning of short-haul service to Chicago)
  • Chicago

Note how we let the intra-state service in NY state and Michigan make it completely unnecessary to have intermediate stops. Someone going from Syracuse to Ann Arbor will simply board an Empire service to Buffalo, take the HSR to Detroit, and change to a Michigan service train to Ann Arbor. Connections will be timed, of course, and often "across-the-platform".

Your train will use that trick also. If you are in Paris, you must ride a TGV to Brussels to pick it up.

  • London
  • Brussels
  • Berlin
  • Warsaw
  • (skip Minsk, they can't even get the Bug canal open.)
  • Moscow

Now at this point, you're largely doubling the Trans-Siberian Railroad, and you'll expect that to handle all the local traffic. A Siberian customer might have to ride the TSR for a day to get to the station for this train. (though realistically your route would also accommodate more local/short-haul trains, reducing that to 6 hours for those people.)

Crews, though

Because an automated train in the middle of Siberia ain't gonna happen.

In fact, there's so little out there in Siberia, Alaska and British Columbia, that your main reason to stop will be crew change.

The simple fact is that it's not humanly possible to know 3000 miles of route. Amtrak has 800 km crew districts, and assuming your grade-separated route eases some of the knowledge load, let's say 1600 km. So 7-8 stops (depending on terrain) between Moscow and Hope BC. I chose Hope because terrain favors it, to use the word "favor" lightly. The last 7000 km have been rough terrain, hugging the Ring of Fire.

Terrain is not loving this plan

I know you want to go to San Francisco but the terrain will make you fight your way through tough mountains all the way down from Anchorage to the Golden Gate. You're better off swinging hundreds of miles inland, e.g. through middle Alaska, NWT, Edmonton, Fargo and on to Chicago, and avoiding the Rockies altogether.

In Asia, there's just no good answer. This has been looked at for awhile. I don't know Russia well enough to guess.

Crossovers and special work is not that big a deal, really.

It is possible to have higher speed special work. It's just more expensive. As far as taking up space, it doesn't need to be in a particular location; you can push it up the line a few kilometers and just have 2 tracks. Crossovers don't need to be a single package in the normal transit form-factor of an "X". They can be a simple switch onto a "branch line" which then rejoins the other side. Below it, grade separated, can be the other side of the X, allowing trains to swap tracks at speed without interference.

Special work right outside of places you're stopping anyway does not need to be fast. So the crossovers and yard throat coming into Moscow station, don't worry about it.

Only the regularly used special work out in the country needs to be fast. When trains are being crossed over for occasional maintenance, that can slow. Slowing isn't that expensive timewise; you get right back up to speed and it only sets you back a few minutes at your speeds.

Freight pays the bills

Strictly as a purely rational (non sci-fi fantasy consideration)... one of the gigantic mistakes that high-speed rail projects have made, and I'm thinking particularly of ones with huge engineering problems to solve... is to overlook freight. They think they're gonna make it on passengers? Get real, nobody makes it on passengers. The freight business doesn't work in Europe because it's too short-haul and suppressed by stupid regulations coughopenaccesscough. But America proves freight is ridiculously profitable in long-haul. They unload container ships in Long Beach, rail them to New York and reload for a sail to Europe, for profit, without government subsidy.

You'll probably want a plain-ole-rails freight railroad to shadow your line just for construction logistics. So plan to develop that into a serious thing that's ready to carry an endless parade of double-stack container trains at 120kph at the closest spacing possible. When you have frontier crossings like the Bering Strait, accommodate freight there too. The profits will offset much of your total cost.

  • $\begingroup$ I do not see a problem when Siberians needs to travel 1 day to nearest maglev station (It could be probably even more). Thats just how the world works. Tracks in history ruined cities and elevated or even created new ones. And people either moved closer to tracks or had to commute.. The crew is really good mark I need to think more on. And I also appreciate the lesson on US geography - The Rockies are good challange. For Asia.. I dont have solution, I will just divert attention from that and make it solid fact. - not making fuzz about it. $\endgroup$
    – Prahara
    May 9, 2020 at 1:52
  • $\begingroup$ About freighting I have something similar in my had about what you said. I am just not detailing it as it is not in Plot any important. $\endgroup$
    – Prahara
    May 9, 2020 at 1:54
  • $\begingroup$ I've driven the middle Alaska to West US coast route numerous times. As long as you are willing to swing in from the coast by a couple hundred miles, there is relatively flat terrain. It isn't mountains that are the problem. It's muskeg. You'll have to sink your supports fairly deeply into the permafrost over much of the route. $\endgroup$ May 9, 2020 at 3:41
  • 1
    $\begingroup$ "nobody makes it on passengers" - Plenty of systems in East Asia do. $\endgroup$
    – Vikki
    May 9, 2020 at 23:12
  • $\begingroup$ I'm sorry, but there are plenty of big cities in Siberia, and in European part of Russia as well, certainly way bigger than paltry 300k of Anchorage. $\endgroup$
    – Alice
    May 10, 2020 at 4:31

Instead of your current solution, I'd like to propose another solution:

Detachable Wagons

Instead of slowing the whole train and trying to match another to it, just disconnect the last wagon of the main train for everyone who wants to get off.
Parallel to that, everyone who wants to get on gets on a specific wagon in the station, which then gets brought up to speed and connects to the the train when it passes by.

It'll require some timing, but if speed and distance are constant it should be fairly predictable. Everyone on the main train hardly notices anything.

  • 2
    $\begingroup$ Excellent idea. Of course you will need to deal with the problem that if the main train is going 500 km/h then the feeder train will need to accelerate to circa 550 km/h in order to catch it, but this seems to require much less technological investment than trying to make a parallel track transfer anything like safe. $\endgroup$ May 7, 2020 at 15:31
  • 3
    $\begingroup$ ...or the main train can slow to 450 km/h to let the feeder catch up. Still way better than slowing to 200 km/h or even coming to a complete stop. $\endgroup$
    – Matthew
    May 7, 2020 at 15:35
  • 7
    $\begingroup$ Upvoting not only because this is a practical solution, but also because the entire concept practically begs its use in a heist film of some sort. $\endgroup$
    – bta
    May 8, 2020 at 1:40
  • $\begingroup$ No significant speed difference needed. For the feeder to connect: parallel tracks, both trains going 500 km/h. Tracks merge, feeder train is scheduled so it is just behind. It accelerates slightly (or was already going slightly faster), then connects. Same thing in reverse at the other end of the train. $\endgroup$
    – sfink
    May 8, 2020 at 4:07
  • $\begingroup$ Can you make more clear how this is different from my answer? $\endgroup$
    – L.Dutch
    May 8, 2020 at 5:22

You are creating a solution in search of a problem. Unlike conventional trains which are limited by the friction of their wheels, Maglevs can accelerate quickly. Your limit is usually passenger comfort, not the capability of the system. They decelerate equally quickly, recuperating energy very efficiently. If you have many stops, so the transfers add up, just run some express lines that bypass the intermediary stops.

Just stopping at the stations bypasses all the safety issues with stuck doors and fatal crashes. I don't see a way how you can make a transfer at high speed failsafe. Customer trust in safety is extremely important, otherwise your business dwindles to nothing.

  • 6
    $\begingroup$ I don't see a way how you can make a transfer at high speed failsafe. You mean... like resupply spacecrafts docking on ISS at 7.66 km/s? $\endgroup$ May 7, 2020 at 8:44
  • 4
    $\begingroup$ Orbital mechanics is a wholly different beast. Transfer at high speed with a ground vehicle, if that makes you happy. $\endgroup$
    – Whitecold
    May 7, 2020 at 10:44
  • 3
    $\begingroup$ @Whitecold Pretty much the same system: match speed, link up / lock together, walk across. The rails mean it's fairly smooth - like walking between train carriages, only perpendicular, rather than trying to do the same thing in cars which can go anywhere. Engineers are already working on it, and it's not a new idea $\endgroup$ May 7, 2020 at 14:13
  • 1
    $\begingroup$ The difference between doing this in orbit and doing this on Earth is that in orbit there is no friction to suddenly slow one of the objects down. On Earth, there is friction everywhere. $\endgroup$
    – NomadMaker
    May 7, 2020 at 17:57
  • 1
    $\begingroup$ @Chronocical I am not saying you can't make the docking work ever. What I am doubting is that you can a) make it work safely enough to be approved, and b) that it is worth the trouble compared to simply stopping at the station $\endgroup$
    – Whitecold
    May 8, 2020 at 6:39

The train can travel up to 500 km/h

the Main train will slow down to 200 km/h

This is the point defeating your whole system. The energy expenditure for accelerating from 200 km/h to 500 km/h is 5.25 times the energy needed for accelerating from 0 to 200 km/h. You are just complicating the whole system for a marginal saving.

Just have the exchange wagon be the only one to accelerate/decelerate, and have it go from the convoy to the station and from the station to the convoy, without changing the train velocity. In this way you are investing energy in only a wagon, not in a whole train. You can even make a loop at both ends of the track, so that the whole train has never to come to a full stop within its operating life.

By the way, the above concept is not coming out of my bag, it has been shown several times by some Chinese designers, I wasn't able to find the reference with the CGA showing how the whole thing would work.

  • $\begingroup$ Oh... well I guess it is not really that much difference to have transit track 250km instead of 100km, was just thinking about a way not to to have the transit track too long to accelerate to 500kmh.But I guess I overthink it. Thanks! $\endgroup$
    – Prahara
    May 7, 2020 at 7:34
  • $\begingroup$ Minor quibbling (5^2-2^2)/2^2 = 21/4 = 5.25 != 6.25 $\endgroup$ May 7, 2020 at 8:38
  • 5
    $\begingroup$ Major quibbling: regenerative breaking on maglev. $\endgroup$ May 7, 2020 at 8:40
  • $\begingroup$ You can even make a loop at both ends of the track => If the maglev main train can loop for an infinite duration in case of issue, then there's not time limit for the transfer. You're still left with a transfer at 500 km/h, which has its own dangers, of course. I do wonder how large the loop would have to be... $\endgroup$ May 7, 2020 at 15:08
  • 1
    $\begingroup$ An open-track train cutting through a major metropolis at 500 km/h... I see a little inconvenience with that. Of course we can switch to an enclosed track... and arrive at hyperloop concept. $\endgroup$
    – Alexander
    May 7, 2020 at 16:44

L. Dutch's answer and Martijn's are correct; uncouple only the carriage whose destination is the next station. It won't recouple to the same train afterwards, but to the next one, an hour later.

Having uncoupled, it will brake to a lower speed, so that switching tracks to a branch line is safe, (per comment to the question), take the branch line to the station, and halt. Meanwhile its predecessor has left, merged onto the main line, and accelerated, to wait ahead of the train, where it will couple at the front. Fun fact : the Russians developed fully automatic approach and docking for resupplying the MIR space station; these trains use a descendant of the same technology.

Every coach has traction, but they need not all have power collection. If you can't supply sufficient power via track induction, it always has at least 2 pantographs, normally 3, and usually only the central one is raised. (Vibrations from one pantograph prevent a second one making good connection). As coaches leave, the pantograph cars move aft, and detach (and are replaced) - generally at Moscow, and one of Yakutsk or Vladivostok, and Seattle.

Why uncouple at the rear, and couple at the front? Well this way round, you board the coach for Vienna or Moscow, or Anchorage, and you don't have to hike your baggage through half the train to the transfer car. Your carriage just peels off at the right place.

So, only the leading panto car and coaches ahead of it, leaving Paris, are First Class and travel continuously through to the continental US - these seats sell at a premium; others make a stop along the way, arriving a hour later, at a useful saving on the fare. (Upgrades are possible for a suitable consideration, or bribe...)

Back the line up with an HVDC interconnector, scaled up from current technology. Today that allows a link in excess of 3200km in a single stage, at 12GW - plenty to power the fleet and then some. For reliability you have at least two of these. Funnily enough, 3200km is the distance from Shannon to Gander, so you can transfer power across the Atlantic, and the Trans-Canadian Interconnector, and power the line independently from either end.

Conducting solar power from the Gobi Desert sunfarm (and others) across timezones using spare capacity is a nice little earner on the side for the railway company, nearly as profitable as the Silk Road goods line extension to Beijing and Wuhan...

And naturally, despite the HV interconnector crossing Britain to Shannon in Ireland, anything beyond the Paris-London TGV is still on track laid by Brunel and his Victorian contemporaries.

EDIT to move the safety case from comments into the answer...

A.I.Breveleri points out that coupling onto the front also obviates the need for the feeder carriage to ever go faster than the thru train, but - the feeder carriage is injected onto the express line exactly when a massive main train is approaching a 500 km/h. There must be at least three levels of emergency response to the case where the feeder carriage fails to accelerate properly. (No kidding; railways are among the most heavily engineered systems for safety).

While the coupling management tech does have space heritage, that's not enough for the safety case! However, there are at least 2 sidings to get the feeder out of the way if need be, and you don't let the train within several blocks (km) of the feeder until it's at speed, braking the main train at the first sign of trouble. And there could be some good plot points there...

As a Maglev can brake more heavily than a wheeled vehicle, passengers will be instructed to wear seatbelts through the re-coupling manoeuvre as a precaution, and where possible, mealtimes will be scheduled to avoid re-coupling during the soup course.

Side note : track switching aka "points" in Maglev technology may turn out to be safe and comfortable at any speed, using "software defined rails" at the points. In which case the feeder can accelerate on the branch line, reducing safety problems.

  • $\begingroup$ Coupling onto the front also obviates the need for the feeder carriage to ever go faster than the thru train. $\endgroup$ May 7, 2020 at 15:43
  • 4
    $\begingroup$ Coupling onto the front does, however, introduce a new safety problem -- the feeder carriage is injected onto the express line exactly when a massive main train is approaching a 500 km/h. I would like to see at least three levels of emergency response to the case where the feeder carriage fails to accelerate properly. $\endgroup$ May 7, 2020 at 15:48
  • $\begingroup$ @A.I.Breveleri Very good point. The coupling management tech does have space heritage ... but I'm not volunteering to write the safety case! However, it includes at least 2 sidings to get the feeder out of the way if need be, and you don't let the train within several blocks (km) of the feeder until it's at speed, braking at the first sign of trouble. And there could be some good plot points there too... $\endgroup$ May 7, 2020 at 16:00
  • $\begingroup$ braking at the first sign of trouble As Whitecold says in another answer, maglevs can brake more quickly than wheeled trains. But we'll have to avoid tossing the passengers around, so I imagine that they will be instructed to be seated and fasten their seat belts during the operation -- just in case. $\endgroup$ May 7, 2020 at 18:47
  • 1
    $\begingroup$ @A.I.Breveleri Indeed, and where possible, mealtimes will be scheduled to avoid re-coupling during the soup course. $\endgroup$ May 7, 2020 at 18:51

As other have said it does not work well. I would like to point out that the concept was tried at Disney World with the Tomorrowland Transit Authority people mover. It was a moving sidewalk that match the speed of the cars so you could enter the cars with them having to stop. It never worked well and was eventually discontinued.

  • $\begingroup$ I experienced that fiasco in person. Even a slight difference in speed appeared to significantly increase the chances a passenger would fall when transferring. I can't imagine scaling that up to 500 km/h where even a 0.01% speed difference is huge. $\endgroup$
    – bta
    May 8, 2020 at 1:30

In a maglev system why you would need a long convoy of wagons if those units can accelerate / desacelerate in independent ways without a locomotive? Each wagon can travel between two points per se without complicated merge connections or waiting exactly time schedules.

  • 1
    $\begingroup$ Even modern high-speed trains work like this. The German ICE3 trains don't have locomotives; they have per-carriage engines. $\endgroup$
    – MSalters
    May 8, 2020 at 10:57
  • $\begingroup$ The answer is traffic management. If you must maintain a specific separation between trains for safety, like 2 minutes between trains, then smaller trains will reduce carrying capacity. Doesn't mean it can,t be done though. $\endgroup$ May 8, 2020 at 14:57
  • 1
    $\begingroup$ @BrianDrummond the transport system of OP is easily managed by an efficient AI, even actual subway systems and promissed autonomy cars tecnology making "smart highways" are an actual path to it. $\endgroup$ May 8, 2020 at 17:08

If you really want the stop to be as "convenient" as possible (in terms of a short slowing-down time), why not make the passengers parachute (paraglide) out of the train at their stop? You could have an entire carriage that puts out glider wings and a parachute just before its stop, detaches from its wheels, lifts off and glides into a parking space at the station.


If you're writing some sci-fi and you're committed to the train-that-never-stops concept:-

Slip carriage

"In British and Irish rail transport, a slip coach or slip carriage is passenger rolling stock that is uncoupled from an express train while the train is in motion, then slowed by a guard in the coach using the brakes, bringing it to a stop at the next station. [...] This allowed passengers to alight at an intermediate station without the main train having to stop, thus improving the journey time of the main train." -- Wikipedia

LABIS train

The LABIS train is a proposed train design from the 1990s, where a shuttle would be detach from the back of a train to drop off passengers at each station, while another shuttle that could go faster than the train would allow passengers from the station to join the back of the train.

Chinese train with detachable shuttles on top

I've also seen a proposal for a non-stop train with a similar shuttle design, but with the shuttles on a separate set of rails above the regular train.

As you can see, other people also find the idea of a non-stop train with stopping carriages appealing! Although they will often speak in terms of commuter trains making 30 stops, and rarely progress beyond the concept stage.

Whatever arrangement you go for, you'll want to think about safety - in particular what happens if the train or shuttle has to emergency brake, and what happens if the shuttle and train can't disconnect as planned? If a new train design is prone to fatal accidents, or needs many more staff to prevent accidents, it's unlikely to replace current train designs.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .