France High-Speed Rail Paris

Expanded High-Speed Rail Access Planned for Greater Paris

Routing TGV in Greater Paris

» National rail company SNCF plots new TGV service for province-to-province trips, with coordinated development of stations in Paris’ suburbs.

With a high-speed rail system whose capacity and breadth expands every year, France boasts of an efficient national travel network. It is limited, however, because of its historic focus on its terminal stations in Paris, which account for a large percentage of overall trips. Today, customers traveling from region to region, uninterested in stopping in the capital, still are often required to switch trains in Paris, a time-consuming transfer that typically requires crossing the capital by metro. Meanwhile, the city’s suburban residents, who make up 80% of the population of the metropolitan area, are forced to enter the city to make most rail trips.

SNCF railways is plotting a solution through the construction of two new circumferential lines and five new stations in the suburbs. As the United States government prepares to commit to significant investments in new high-speed rail corridors, it should look to Paris to examine the French capital’s planning errors and its new efforts to remediate the situation.

The first three French high-speed lines operated on the same simple model: depart from a Parisian center-city station and extend to a major city in the provinces. The LGV Sud-Est extended to Lyon from Gare de Lyon in 1981; the LGV Atlantique departed from Gare Montparnasse towards Le Mans and Tours in 1990; the LGV Nord, heading to Lille, left Gare du Nord in 1993. Each provided excellent connections between Paris and regional cities, but commuters hoping to avoid the capital altogether were left without fast options. Though the required transfers through Paris may increase the stature of the capital city, it does a complete disservice for people who want to get from region to region. It encourages air and car alternatives.

That situation changed somewhat in 1994, when the LGV Interconnexion Est opened. The 35-mile link connected the Sud-Est and Nord lines and created two new stations at Charles de Gaulle Airport and at Euro Disney in Marne-la-Vallée. The project has made it possible to travel 600 miles from Lille, in the country’s north, to Marseille, on the Mediterreanean, in less than five hours, without a transfer in Paris.

But for people attempting to get to the west side of the country, there’s no simple solution. TGV trains traveling east-west through the Paris region have to share commuter rail tracks through the south of the city. Trains heading to the northwest of the country, which is currently disconnected from the high-speed network, have to use extensive sections of out-of-the-way, slow-running track.

SNCF’s solution to the problem will involve the construction of two new lines — one circling around the south of the city, and the other traversing the northwest. Each project would eliminate transfers for province-to-province travel. In the south, Orly Airport and Villeneuve-Saint-Georges would get TGV service. The northwest line, which would share tracks with the new circumferential automatic metro being planned for the Parisian suburbs, would connect Europe’s largest business district at La Défense and the growing hub at Pleyel, each of which would be provided brand new stations.

The French rail company envisions new terminals on the scale of Shanghai’s south station that will allow commuters direct travel to areas virtually everywhere in the country. The La Défense hub would have the honor of hosting the terminus of the LGV Normandie line, which would connect Paris to Le Havre on the Atlantic coast in just 1h15, compared to two hours today. No longer would all of the major hubs be located in Parisian city limits.

The plan has a number of substantial benefits. Eurostar trains would connect La Défense with the City of London in 2h15. The overload of trains and passengers at Paris’ existing train stations would be relieved by the implementation of new service at suburban hubs, which would reduce overall travel times for people who live outside of the city.

Many of these problems could have been solved years ago had the TGV network been built from the start with through-running trains with a tunnel under central Paris, such as one connecting the Gare Montparnasse and the Gare du Nord. That project, however, was never pursued because of its high cost. And the construction of the new south and northwest links would essentially solve the problem by allowing a detour around the central city. In its recent proposal for Midwestern high-speed rail, SNCF promoted the idea of constructing a bypass of Chicago in the project’s first phase, indicating that it had learned its lesson.

Indeed, the recent French interest in expanding access to the Parisian suburbs is indicative of the fact that singular central city stations are not always enough to satiate the rail travel demands of a region’s residents. Unlike air travel, which relies on just one or two airports per region, rail has the advantage of being able to access several stations; this allows people in different parts of a region quick access to service. If today, most residents of the Paris area have to enter the city to get on a TGV, SNCF’s proposals will open travel to people in the north and south.

In the U.S., where cities are more spread out and where public transport links are less well developed, stations outside of downtown such as those SNCF is developing could be an important element of preliminary high-speed plans. If the Washington-Charlotte rail line is ever improved to true high-speed service, for instance, a bypass around Washington to allow more direct service to areas north of the American capital could be necessary if capacity ever becomes a problem. A station in suburban Virginia would be obligatory — as long as it is placed in a reasonably built-up area to avoid inducing sprawl.

That said, as the French example shows, it may not be necessary to build those suburban connections first; the TGV has been enormously successful so far with mostly center-city stops.

30 replies on “Expanded High-Speed Rail Access Planned for Greater Paris”

In a country that has yet to get one line off the ground, it is encouraging to see an example of a country that continually upgrades its service like this.

I could understand a bypass around Boston so they don’t have to connect North Station and South Station, but isn’t through service possible in both Chicago and DC’s Union Stations? Is the point of the bypass to provide access to O’Hare or was Union Station not meant as the final destination of the midwestern high speed trains in the SNCF proposal?

Nice article but there is still many things to say on the subject.


The actual design was a nice feature until recent years. It permited SNCF to massified fluxes and used very efficiently the trains and HSLs to their optimum. Of course this was only possible with a very effective mass transit system in Paris itself.

Problem. The mass transit system is not so much effective and the over-crowding are geting worse each years. Today the incentive is great to serve the very dense suburbs around the capital. Because it doesn’t make sense to force people go through the core of the city on at capacity infrastructurs. It is of course very efficient economicaly speaking but less and less acceptable and desirable.

Second. Nowadays Paris is very well connected to the rest of the country by HSLs. The market is more or less mature there. The only rapidly growing HS market in France is the province-to-province market.

Third. Most of the Parisian railway stations (Gare de Lyon, Gare du Nord and Gare Montparnasse) are at capacity or will be in the future when the planed network expansions are going to open. The SNCF clearly needs new stations in the area.

Four. Like you said this plan can actually get more people in trains in Paris and the French provincial cities by offering new services.

But the amount of new HST users in the Paris metropolitan area should be quite small. In fact, I’m not sure the SNCF will offer enough interresting services in the suburbs stations to attract many people there. Most of the actual stations (CdG Airport, Massy and Marne-la-Vallée) are not well served. This can be reinforced by the difficulties one have to travel from ‘burbs to ‘burbs in the Paris region. It’s still far more easier to enter the city than to move around it.


But enough with that. The French HSN is very peculiar and might not fit the urban structur of the US. I kind of think that the US will be best served with Japenese style services (all-stop, semi-express and express trains) and graps of stations in the urban areas. The sprawl are too big and the mass transit network usually not effective enough. I will nonetheless concedes that every station doesn’t need to be build directly but can be implemented after the opening of the line. It will save some millions.

A ring tunnel linking the lines leading away from the old steam-era head-end stations within Paris would be shorter but stupendously expensive to build. The present proposal almost creates the next-best thing, a grand peripherique ferroviaire (GPF), but the connector between the LGV Normandie to the LGV Atlantique is missing.

The airport connector strikes me a superflous, why not just add some turnouts to support a shuttle service out of Gare de l’Est?

Rafael –
I should point out that the majority of the CDG Express line would actually not run at high-speeds… it would parallel RER services on its own tracks, so it’s not as big of an investment as it appears. It will also be privately funded.


It’s not possible to “add turnouts to support a shuttle service out of Gare de l’Est”.

You have to create some new rails near the airport and excavate a tunnel between Paris Est and the RER B right-of-way.

Anyway the CDG Express project is actually in very bad shape. Nobody knows if it will attract enough people to be economicaly viable with the M14, RER B and metro orbital there. There is just one bid, the other companies have withdrawn, and the bid’s due date was push back yesterday.

I use the TGV/Eurostar to travel from Geneva to London a couple of times a year, and I have to say that for me the transfer in Paris is actually a highlight of the trip. I think I’d miss it.

La Defense-Massy is a missing link on that map. How long until they figure out the obvious advantage of extending north trains from La Defense around to Massy and Orly, west and southwest trains around to La Defense, Pleyel, and CDG, south trains to La Defense, and east trains to CDG, La Defense, Massy, and Orly? Utilize the western side of the proposed loop to not only directly serve a majority of the suburban areas with each line, but also provide a single transfer between any major HSR line. Perhaps simplify into Belgium-Lyon, London-Bordeaux, and Strasbourg-Le Mans through routes with enough bundling and reach to offer all-day hourly service on each paired route.

If SNCF can submit proposals for Texas TGV, perhaps I can submit a bid proposal to Paris for such. LaHood is a French name, yes?

Washington Union Station is a through station, and the only major station in Washington. The southern approaches to it aren’t terribly good — the Long Bridge is only double tracked and the viaduct is slow — but it would be easier and cheaper to fix the approaches than to build a bypass.

The Union Station movement in the US saved us from the Paris or London mess. I can think of only the two cities where there’s two major stations: New York (Penn and GCT) and Boston (North and South). It’s conceivable that true high speed rail to the north and northwest of New York could be based on GCT, in which case there would be an argument for running a bypass through Jersey. It’s conceivable that true high speed rail to Maine and New Hampshire could be based on Boston North Station, in which case a bypass along rt 128 might be necessary to avoid the MBTA. But we are a very very long way from that.

@Mark: I don’t think you are going to miss it anytime soon. There is allready a physical link between London and Geneva. Eurostar could have open route if they wanted but they didn’t. Here two posisbilities:

– journey is too long and the trains will not see enough clients.
– the fact that you have to clear customs and security check before boarding is very inconvenient and requires to isolate a part of the departing station. For a few trains a day that’s not a viable option, space is too scarse in stations.

@All oui oui’d:
I’m pretty sure the plan you describe exists. But the connection between Massy and La Défense is the less interesting link. You could be able to connect the LGV Nord to the Altantique one by going through the LGV Interconnexion Est and Sud.

Jim, AlexB: Chicago Union Station is a single station, but the tracks leading to it from the north and from the south are not connected. That’s why the city of Chicago is proposing connecting the two approaches via a Clinton Street subway, and why SNCF is proposing to build a large bypass serving O’Hare.

In addition, LA Union Station is a dead-end terminal, but California is about to convert it to a through-station.

In New York there’s no reason to base a line to Boston out of GCT. The Park Avenue tunnels leading to GCT have worse capacity problems than the East River Tunnels leading to Penn from the east, and East Side Access will relieve whatever capacity problems do exist in the East River Tunnels. Besides, traffic through New York is an important part of the NEC’s ridership and will become even more important if the line is sped up.

Washington Union Station is a through station, and the only major station in Washington. The southern approaches to it aren’t terribly good — the Long Bridge is only double tracked and the viaduct is slow — but it would be easier and cheaper to fix the approaches than to build a bypass.

Two tracks can easily handle 20 trains per hour in each direction. Washington doesn’t have to begin to think about more than two tracks south of Union Station until VRE goes over 12 or 15 trains per hour at peak. That’s far far off in the future. 2 and half minute headways, which is doable, gives them 24 trains an hour in each direction. . Far far in the future if ever.

It’s conceivable that true high speed rail to the north and northwest of New York could be based on GCT, in which case there would be an argument for running a bypass through Jersey.

Again far far in the future if ever. Tunneling through New Jersey, to develop a high speed right of way through suburban New Jersey would mean lots of tunnels, would probably be as expensive as tunneling through the mountains of eastern Pennsylvania. Through Penna. would get people from Toronto and Buffalo to Philadelphia Baltimore and DC faster and give high speed service to places like Scranton and Binghamton,

It’s conceivable that true high speed rail to Maine and New Hampshire could be based on Boston North Station, in which case a bypass along rt 128 might be necessary to avoid the MBTA. But we are a very very long way from that.

Why do they need to avoid the MBTA? It’s not like they will ever be running 20 trains an hour into Boston from the north. If they connect North Station to South Station they can easily handle 20 trains an hour in each direction. If two come in from Maine and two come in from New Hampshire the people destined for Boston can get off and the people headed south can watch the train fill up with people from Boston who are headed south. Leaves the MBTA with 16 trains an hour to fill. Means people headed to Boston have more trains to choose from and the people headed south of Boston have more trains to choose from. ( instead of one train to Boston leaving Portland or Manchester and then one leaving for New York they both go to Boston and New York )

Because I’m interested in mobolity and access rather than infrastructure for its own sake, I find it hard to assess this proposal without seeing an operating plan that would show how trains actually flow through this network.

The La Defense terminal for the TGV-NW is clearly a winner.

But in the case of the TGV-Atlantique, what exactly will trains from Bordeaux do when they use the new southern connection? Will they go to Gare de Lyon, or bypass the city and head out the TGV-SE line to, I don’t know, DIjon or something? I suppose if the real purpose here is to construct an east-west path all the way across France, something like Brest-Rennes-LeMans-Orly-Dijon-Basel, I could see how that might be a coherent and logical market, so that Bordeaux trains would continue into Montparnasse and the two lines would have a transfer station on their common segment. But other than that I’m at a loss.

Branching ALWAYS dissapates frequency, which is why, to service-oriented planners, new branches aren’t always good news. The addition of these circumferential links requires either (a) doubling frequencies on each entire TGV line or (b) reducing service from each line into its original terminal station. I believe that the eastern bypass happened in the context of a major frequency increase, tied to a fast-growing market. In the east, we were talking about connecting huge cities (London, Brussels) to the whole south of France and Riviera, with connections onward to Italy and Spain. Can the same intense service really be expected east-west, where populations are so much sparser, and if not, what will be left of frequencies into Montparnasse when Atlantique trains start bypassing Paris through this new southern link?

I’m sure SNCF has answers to all these things, but they weren’t in your post so I just thought it important to mention them.

The way the Long Bridge is managed today is at peak one track is for passenger trains going in peak direction, the other is for freight in both directions and passenger counterflow. Freight gets 10 minute slots, passenger five (freight trains are longer and slower and it is a long bridge). Right now there is little passenger counterflow, one VRE turnback and an occasional delayed Amtrak long distance train. Through running MARC trains would add to the counterflow. But even 2030 planned SEHSR service can fit into this management scheme.

But there would be a problem with genuine HSR south of Washington with multiple trains per hour levels of service. Unless freight magically disappeared.

Based on what you’re saying, Long Bridge is managed poorly. The normal way of running things is to have one track in each direction, with both a minimum and a maximum speed. European railroads have to accommodate slower freight trains all the time; what they do is schedule multiple fast passenger trains one right after the other followed by multiple freight trains – see explanation here.


I also don’t know what will be the use of the south bypass. Linking Bordeaux to Lyon and Marseilles through Paris hardly make sense. It’s much easier and as quick to pass through central France… but not as money making ofr the SNCF.

It makes a little more sens when you think about Rennes (in Britanny), Nantes or Tours (wich also are on the LGV Atlantique). They are big markets and can easily profit from a direct connection to the South East.

The problem is the connection on the LGV SE. This part of the network is at the moment allmost to the capacity and the opening of the LGV Rhin-Rhone East is going to use the remaining slots. It will be then hard to add new trains on it… only if someone decides to four track the line between Pompadour (where the LGV Interconnexion Sud will connect to the LGV SE) and the spur to Dijon or if the LGV Centre (which will also connect to Lyon) will open.

I think the real interest here is to have a station at Orly Airport and to connect Gare d’Austerlitz (Gare de Lyon sister station) to the HS network. Trains from the LGV SE or LGV Atlantique could then end up in this underused Parisian station.

Anyway, Paris will allways be the center of France HS Network. It’s the center of the country, a fith of the overall population and a fourth of the GDP. I do think the bypassing traffic will be minimal. A small number of train each day (in the morning and the afternoon for daily business trips).

Las thing, don’t overestimate the importance of the foreign markets. The real pax numbers are from within France (this is known as the border effect). Connecting Lille (fourth french metropole) to Lyon (second fench metropole), the Rivierra (wich is home of four of the most important french metropoles) and beyond the French Alps is allready a huge achievement. It was really a multipurpose link. With it you can interconnect the North of France and Europe, the East (Strasbourg, Germany), the SE (Lyon; Marseilles, Nice, Montpellier, the Alps), CdG Airport and Disneyland.

The south connection will be an important part of the plan to open TGV service to Austerlitz station in the next few years; because Montparnasse and Lyon stations are at capacity, a renovated Austerlitz will be able to absorb some of the service from the Atlantique and Sud Est lines. In addition, it will be the hub of LGV Centre service, which will duplicate Sud Est service to Lyon by routing through Clermont-Ferrand.

It will also, as dist put it, allow people from Bretagne and Normandie to get to the eastern side of the country via HSR, currently not possible.

Chicago hurts itself by its reluctance to make drastic changes to the Union Station site. At present, there’s only one through platform in the entire station; everything else stub-ends, as it has since the station’s opening in 1925. Aggravating this is the more than one-million square foot Riverside Plaza tower built over the station in 1969, with the former CMEX commodities exchange, now converted into a Multiplex fitness center, next door.

Union’s problems cannot be fully fixed without addressing the Riverside/Multiplex barrier. The Multiplex building should be removed, and while I’m not optimistic about the prospects of getting rid of more than one million square feet of prime office space, the Riverside building should be at least gutted out to about the fourth or fifth floor so that a proper rail terminal can be built in the place of those first few floors. This would enable passenger services such as departure lounges and baggage reclaim to be located where they belong, at or above street level, and through platforms to be built in their place. Mayor Daley’s vision of a West Loop Transportation Center, as exciting as it is, won’t offer this kind of capacity for through service, although it expect it to be an important component of the transformation I have outlined.


Gutting out of the 222 S. Riverside Bldg is not an option due to its caissons/columns on 9 foot centers. The building would have to be removed.

The 2 West Loop Transportation Center through tracks could handle 12 trains in each direction, which should accommodate most needs

The structural columns are at nine foot intervals, which is what I take that to mean? I don’t think so. Not every divider between a window in that building is structural, which is the only way you could possibly come up with nine feet. I suggest you go and take a look around both the basement level where the Amtrak ticketing and departure area are, and the lobby of the building. You will not find a maze of columns at nine foot intervals, I can guarantee you that.

As for West Loop, it is not physically possible to get 12 tracks through there even with the multi-level design being discussed. The plan calls for a busway with a lane in each direction, two through subway tracks and two through HSR, each on its own level and each served by island platforms. That poses a significant issue with HSR dwell time chewing up slots what with the large numbers of passengers likely to get on and off at Union Station.


Incredible as it may seem, there really are structural columns every 9 feet on the perimeter of the building. They are clearly visible penetrating through the concourse level. Amtrak had to do their best to work around them when they remodeled, but they were a major impediment.

12 trains, not tracks. With good discipline a 5 minute headway can be operated on a through track. I’ve used a watch to clock this at Atocha in Madrid. Dwell obviously has to be strictly limited at peak times. Trains should have lots of doors (as in Europe)!

Alon: RER A riders are mostly commuters, who aren’t carrying luggage, are experienced with the boarding/alighting procedure and dwell times on that line, and are probably in a hurry anyway. Families coming from Toledo or Madison for a weekend of shopping in Chicago will obviously alight more slowly.

Is it so important for the majority of trains to be through-routed anyway? Chicago is a far larger destination than any of the other Midwest HSR cities, and seems to have plenty of terminal capacity. Transferring between the north and south platforms of Union Station is hardly comparable to schlepping across Paris. Berlin Hauptbahnhof has through routing, but most ICE trains go no further than the suburban station opposite their other endpoint (e.g. trains from Hamburg terminating at Sudkreuz).

Okay, then don’t think of it in terms of RER riders. Shinkansen dwells are on the order of 50 seconds, even at major stations. Shinkansen headways are longer than 2 minutes but that’s because the trains have high top speed; for low-speed tracks, as the terminal tracks are (even high-speed trains will have to slow down), capacity can be anywhere from 20 to 30 tph.

It’s not important for the majority of trains to be through-routed. However, it’s a good idea to let people from Milwaukee and MSP get to the Midwest south and east of Chicago without changing trains, and once you do that, there’s no reason not to through-route every train that comes to Chicago from the lesser-used north tracks.

once you do that, there’s no reason not to through-route every train that comes to Chicago from the lesser-used north tracks.

There is a reason, namely that it would require hundreds of millions of dollars of construction and disruption in an urban area. Through-routing one or two HSR trains per direction per hour for the relatively small number of onward passengers could easily be accomplished using the existing single through track. This isn’t an ideal solution, but surely limited funding is better used on straightening curves etc further out on the lines than on the money pit that is city-centre construction? (Recall that a single elevated mile of the Circle Line is supposed to cost a billion dollars.)

It wouldn’t be useful just for HSR, but also for Metra. Commuter rail requires through-routing much more than intercity rail, because the transfer time is proportionally a larger share of the total trip time.

Institutionally, Metra thinks of itself as a suburb-to-Loop service. It doesn’t see a need for any through-running. I think part of the problem is that it inherited service from many independent operators, who didn’t through-run because they had trackage rights on just one side of the station, and never thought to change things.

I don’t think there’s a single commuter rail operator in the US that sees itself as a through runner — it isn’t just Metra. The public transit culture in the US still doesn’t get regional transportation. Boston — what else can I say? New York — NJT, MTA and LIRR in the process of symbolizing their institutional separation with tunnel stations that dead-end against water mains. LA MTA, a little better but too much still centers on downtown. The last time we had real vision on a regional rail project in the US was probably Frank Rizzo with the Center City tunnel in Philadelphia, and even that had been languishing for 15 years as a good idea from planners at Penn until Rizzo decided on being elected mayor that he wanted to establish a legacy.

I have hopes for Frontrunner in Utah, though. They certainly seem to be starting off in the right way — a service that’s frequent all day and all evening long and, once the Provo extension is open, connecting the entire Wasatch front without an arbitrary break in Salt Lake.

Looking back at Chicago, if as D Phillips (is he THE “D Phillips”, BTW?) states, 222 N. Riverside is built in that way (the same construction method as the World Trade Center was, or the AON Center is, by the way — I was not aware of another major building done like this), there really is no way of enhancing through capacity at Union other than removing the building. You can appropriate the first few floors for a dramatically enhanced passenger concourse that would solve the pedestrian congestion but nine feet is, of course, narrower than a train.

Removing the building, however . . . . you’d have opportunities like you have at Toronto Union Station. Absolutely massive potential capacity. Build a new tower on the air rights that actually respects what’s going on underneath if that’s what it takes economically. I am convinced it would be cheaper than West Loop and provide more capacity — but only if Metra can be persuaded to sign up for run-through service.

With regard to West Loop and dwell times, would anyone care to comment how 30th Street Station in Philadelphia is doing these days with passenger ingress and egress, seeing as how it’s probably the only current major US station that requires continual rapid-fire boarding of large numbers of passengers to through trains at present? But I think even D Phillips’ five-minute idea is optimistic in practice for Chicago — with a high speed train you’d be talking several hundred on, several hundred off with each train. Perhaps ten minutes is more realistic. And then you have to start asking if two tracks will be sufficient in the long run. We could be barking up a very expensive tree for very little return with West Loop when the answer of removing 222 S. Riverside, even if only to replace it with something that allows trains through the basement, is staring us in the face.

Before I get too carried away on 222 N. Riverside I need to also admit a weakness in my suggestion. That would be the “throat” at the north end of Union Station, which is currently only three tracks wide and is stuck between almost brand-new high rises and can’t be widened further. Toronto has something like nine tracks out of each end of the station; Chicago only has that to the south.

Well, after the fine digression from @17 through @29, I do begin to understand why SCNF’s proposal to operate HSR in the Midwest features a bypass around Chicago!

Leave a Reply