Alon Levy wrote about the USDOT’s airfare database and the implications for potential high speed rail markets. Alon’s broad conclusions are that the regional markets suitable for HSR haven’t changed all that much in the last decade – the NEC is the most promising corridor, California is second, and others also make sense – but short of a true national network.
His post sparked me to ask – what are the best practices for integrating HSR with airports? High Speed Rail networks have different characteristics from airline networks. Some elements of the services are competitive, while others are complementary. A big part of the political sell for HSR in California was the counterfactual of required airport investment to facilitate travel in the state – but what’s the real impact on airports and airline service?
HSR Corridors vs Airline Hub Networks
Even modest rail systems can effectively compete against air travel in given corridors. Amtrak often brags about the barely-high-speed Acela earning an 80% share of the air/rail market (a notably smaller portion of the overall travel market) between NYC and DC. American and Delta still offer their Shuttle flights between DCA and LGA, but they don’t dominate they way they used to.
Amtrak has the advantage along a linear corridor, where one train making multiple stops serves multiple markets – If United Airlines wanted to match service along the NEC, they’d need a lot of feeder flights from DCA, BWI, and PHL into EWR – and they’d have to run them far more frequently to offer as many departure times.
Airline hubs are a different beast – the spokes all feed traffic into the hub. They’re not flying a route solely for the local traffic, they’re also feeding traffic into their hub to connect to other routes.
International Comparisons
Despite the track record for HSR in other countries, there are surprisingly few examples of excellent air-rail integration. Only a handful of big airports have quality connections to HSR networks: Frankfurt, Paris-CDG, and Amsterdam are most prominent. I found this paper exploring the history of both FRA and CDG:
In Frankfurt, the addition of HSR service allowed Lufthansa to decrease domestic flights and increase international flights without losing domestic passenger feed:
With the opening of the Frankfurt–Cologne and Frankfurt–Stuttgart HSR lines, Frankfurt Airport increased its catchment area by 10 million people who suddenly lived within 2 h of the airport…
Whereas Frankfurt Airport may be limited in terms of capacity, it has managed to support an increase in international passenger traffic and to maintain dominance in Germany as the major long-haul international airport serving the country.
Conversely, at CDG, Air France retained a small number of domestic flights to feed their long-haul flights, despite the TGV’s dominance in markets to/from Paris:
The other two routes, Paris–Lyon and Paris–Montpellier, experienced a relatively flat trend in passenger traffic, although there are competitive HSR alternatives that link these cities directly to CDG. In particular, the capacity on these routes has remained relatively stable, supporting research claims that airlines are likely to maintain a certain number of flights at their hub airports to maintain their network, even with the presence of fast and reliable HSR–air connectivity (7).
The paper’s conclusions about the characteristics of a successful HSR/airport integration:
Infrastructure. To provide feeder or transfer service between HSR and air transportation, the rail station should be located at the airport. If the HSR connection at the airport is constructed as a detour from the primary network patterns on the rail system, it is unlikely that the airport will be served with enough frequency.
Schedule and frequency. Rail operators and airlines often have the same goal of optimizing their networks, but they are separate networks. Coordinating timetables to ensure that rail service meets banks of connecting flights is an important consideration.
Market characteristics of the airport. In the two successful cases in this study, the primary airports with HSR links were the dominant international hubs of each country. For both CDG and Frankfurt International Airport, domestic traffic declined and international passenger traffic increased. Two key factors may have influenced this growth: partial alleviation of congestion at the airport by decreasing domestic flights and success of the HSR lines as feeder service for international flights.
In other words: lots of similar patterns to good transit planning on the rail side of the equation: Be on the way; frequency is freedom, etc. The key difference is about integration: is HSR service best thought of as a means of ground transportation – just extending an airport’s catchment area? Or is it best considered like a true connecting flight – integrated into an airline’s ticketing, loyalty programs, baggage handling, etc.
For airports and airlines, things are a bit more complicated. Both CDG and FRA are huge hubs, and the primary international connecting hubs for their respective countries.
In the US, the airport/airline situation is quite different. As a much larger country, no airline operates a single hub that dominates international traffic the way that CDG and FRA do in France and Germany. Far more medium-distance air travel is domestic travel; US visa rules and unfavorable geography make international-to-international connections less common – meaning domestic feed at big connecting hubs is even more important.
Future US HSR/Airport Links
Alon’s conclusions:
In the Midwest, the core lines remain strong, but more peripheral Midwestern lines, say a bypass around Chicago for cross-regional traffic or improved rail service due west toward Iowa, are probably no longer worth it. The Cleveland-Columbus-Cincinnati corridor may not be worth it to build as full HSR – instead it may be downgraded to an electrified passenger-primary corridor (as I understand it it already has very little freight).
Now, Alon was using this air travel data as a proxy for overall travel demand. Assuming the demand is still there (even if the recent growth has been mdoest) Chicago strikes me as a market with great potential as an air/rail hub – a huge hub airport, located near existing candidate HSR lines, and wouldn’t require a large detour to serve. The full Midwest HSR network centered on Chicago could extend ORD’s catchment area (connecting MSP, MSN, MKE, DTW, STL, CLE, and others) and potentially free up airport capacity for longer-haul, higher-value flying.
SFO is another airport that stands to benefit from HSR. SFO’s airfield is constrained and unable to expand; United’s trans-pacific hub depends on feeder flights often limited by poor visibility; the airport’s location is immediately adjacent to the planned HSR service.
Better planning for CAHSR would add even more value to the connection at SFO – particularly had planners used Altamont Pass, the airport would have a faster connection to Sacramento, increasing SFO’s catchment area.
The Northeast Corridor has lots of potential air-rail connections. The rail line itself passes plausibly close to lots of major airports: DCA, BWI, PHL, and EWR. DCA lacks international flights and is can’t grow; BWI’s hub carrier is almost entirely domestic and doesn’t have many international flights to feed.
Both PHL and EWR are interesting candidates for American and United, respectively. Newark in particular meets all of the criteria for success described in the paper. United already uses EWR’s proximity to code-share with Amtrak in lieu of flying short connecting flights from Philadelphia (though the business practices of Amtrak and United are quite different and can make for a challenging passenger experience) and United recently announced cutbacks to flights within the NEC to EWR – dropping the 4x daily BWI-EWR service due to airport constraints.
Philadelphia would require an airport station along the NEC that doesn’t currently exist. The airport’s proximity to New York and the larger travel market there means a bit of a precarious existence, but PHL’s hub carrier (American) has recently been shifting connecting traffic to PHL and away from their greater New York operations, which are split across LGA and JFK and increasingly focused on serving local traffic.
The potential is there for these airports – the big questions will be a) if HSR service ever exists (or improves), and b) what form the airline/railroad partnership takes – as a true connection, or as extended ground transportation?
Actually, Lufthansa reserved space on the regular IC trains between Frankfurt Airport and Stuttgart way before the HSL was opened.
They also leased a very distinctive class of EMUs from DB to run trains from Frankfurt Airport to Bonn, Köln (and beyond), via the scenic Rhine valley route, way before the HSL to Köln even got built.
Things are easier in this case, because the Frankfurt – Mainz line runs along the airport, and it was easy to establish a stop.
Max – thanks for that, helpful context.
Integrating air tickets with rail tickets seems to be a big challenge, in part because the gap between airline and railroad practices is so big in the US. I’ve read of bad experiences with folks using the Amtrak/United codeshare, where their first leg is on Amtrak, their ticket isn’t properly scanned, and the United bumps them off of their connecting flight – that’s a big oops!
If SFO is constrained, then you want to steer people away from it or reduce air traffic to it, rather than increasing its catchment area. The high and growing level of LA-SF air traffic suggests HSR would mostly be a replacement for air travel in California, rather than a feeder.
The Midwest is a different situation. The issue there is that you can use HSR to extend O’Hare’s range, but not by all that much. You get Milwaukee and Indianapolis really easily, and probably also St. Louis and Cincinnati, but beyond that range cities get too far, and Detroit and MSP are also giant hubs in their own right.
The only place in the US where HSR can plausibly feed airports is the Northeast, by connecting Washington, Philadelphia, and Boston to New York better. However, there is zero reason for any HSR train ever to stop at Newark Airport, so something will have to connect Newark Penn with the airport; it’s plausible to send some surplus NY-DC trains to Jamaica for the JFK connection, but the scheduling around regional trains is nontrivial.
SFO’s airfield is constrained. The runway layout works OK in clear weather, but when the fog rolls in, the capacity drops. And when an airport enters flow control, the airlines have to pick and choose which flights to prioritize. They usually opt to cancel the smaller planes flying the shorter feeder routes.
I think the broader point is that for a strong international hub like SFO, replacing traffic or feeding traffic isn’t an either/or – HSR would serve both roles.
The reason I picked ORD over MSP or DTW is a) ORD would be closer to the center of the network, and b) ORD has much stronger international demand (requiring feed) than DTW and MSP combined, which makes the complementary roles stronger in Chicago – at least in theory.
Alex – Lufthansa did prevent the ticket handling issues by having their own check-in counter in Stuttgart, Bonn and Köln (maybe a few other places too). Because baggage check-in had to be done too, this assured accordingly trained staff.
I don’t know how it is handled in France, where certain TGV stations also have an IATA code (such as QXB for Aix-en-Provence TGV or XZN for Avignon TGV). It is possible that they hand out a voucher for the train ticket, which is exchanged at the station. (I used ebookers to see what shows up between Avignon TGV and NYC, and the trains do shop with an Air France flight number … although it is considerably cheaper to buy the tickets separately).
Max – interesting.
I’m not aware of any airlines in the US offering remote check-in facilities, either for rail or bus. I do know that United runs a connecting bus to EWR from Allentown, PA (the short connecting flight was often cancelled), but I’m not sure how they treat it.
I’ve been interested in the prospects for the equivalent of a city air terminal (such as in Seoul) in the US, using buses, but because most flights don’t need passport control, I’m not sure what the Korean version would get you. You don’t clear security remotely, nor check bags remotely (thought I suppose this could be done with the right kind of facility – tbh a coach bus is probably easier than a train).
You might find the International Air Rail Organisation a useful resource. It has produced reports on off-airport check in and competition and cooperation between air and high speed rail, among others. http://www.iaro.com