Category: Good Transit

Philadelphia Link, or Organization Before Concrete

Pedestrian Observations commenter Steve Stofka has a blog post treating Amtrak’s $117 billion high-speed rail proposal for the Northeast Corridor with all the criticism for extravagance it deserves. Focusing on his hometown of Philadelphia, he explains how Amtrak’s proposal for new urban tunnels under the city and a new stop at Market East is insane, and how using mostly existing rights-of-way and stopping at the existing 30th Street Station is a vastly cheaper alternative.

Criticizing Amtrak’s plan is like shooting fish in a barrel. The reason I’m linking to Steve’s post is that it underscores a general theme in transit cost overruns. He explains the reasoning behind Amtrak’s choice of new tunnels:

How expensive is freaking expensive? The kind of bore being proposed is the single most expensive type of tunnel possible: it runs through a soft geological environment with zero tolerance for surface subsidence. It would cost more, mile by mile, than even the Gotthard Base Tunnel. The expense of this tunnel is so great that it amounts to about a tenth of the total budget of the plan (about $10 billion, or a billion a mile, out of a budget of roughly a hundred billion). When a single budgetary item commands that much expense, one must analyze and ask why: why do we need to spend a ludicrous amount of money in Philadelphia for what amounts to marginal access improvements? Knowing SEPTA, politics–and SEPTA’s “get-off-my-lawn” attitude–is most likely to blame.

The relevant answer is the slogan Organisation vor Elektronik vor Beton: organization before electronics before concrete. Getting agencies to cooperate is hard on the managers, but cheap. Electronics, for example modern signaling to increase train capacity, costs more, but is affordable in a rich country. Concrete requires labor-intensive construction and is expensive.

The existing right-of-way in Philadelphia has no capacity constraint. It has four tracks, and a peak commuter rail frequency of six trains per hour. In contrast, the S-Bahn tunnel in Munich has two tracks and 24 30 trains per hour (thanks to ant6n for the correction); the above German link is concerned with cost overruns on a project to construct a second S-Bahn tunnel, currently estimated at half the per-km cost of the Philadelphia extravaganza. And Munich is far more advanced on organization than Philadelphia, where Amtrak and SEPTA have separate tickets, station staff, and schedules.

The same could be said about the LIRR/Amtrak grade separation. From a technical perspective, it is unnecessary. From a political one, it requires Amtrak trains to use the Penn Station’s lower concourse, currently monopolized by the LIRR; said concourse has better passenger flow and has station staff and ticket vending machines, but because of artificial separation into LIRR and Amtrak turf, New York State has to fork over $300 million for concrete.

Overperforming Rail Lines

Amtrak’s latest addition to the Northeast Corridor network, the once daily Lynchburg extension, is overperforming. Both Amtrak’s press release and local reporters brag that this train has overperformed ridership expectations by a factor of 2.5 and revenue expectations by a factor of 3. As a result, it has been consistently operationally profitable, in fact the only train to have this distinction other than the Acela.

The remarkable thing about it is that service levels aren’t high. The average speed south of Washington is mediocre, about 80 km/h. NARP talks about the importance of frequency; but the train is once daily, and is offset by only two hours from the Crescent, a long-distance train covering the same route. There were weak signs of pent-up demand on the Crescent – it sometimes sells out due to limited capacity, but even then it loses money like all other long-distance trains.

The best explanation for this success is that, although the route is slow, so are the competing highways. There are no Interstates that realistically compete with this train; I-81 is too far west. Google Maps gives a Washington-Lynchburg travel time of 3:32, versus 3:46 on the Regional and 3:30 on the Crescent. Add in traffic and the train can beat the car.

A more general point is that bad service that is failing could become more successful if it were improved. German regional trains that were closed due to low ridership when they ran just a few times per day are now flourishing after reopening on an hourly clockface schedule. And several Amtrak corridor runs improved their ridership and finances after more than daily or twice daily frequency was added; they just have to compete with faster roads, so they still lose money.

The next issue is then what other gaps there are in the Interstate network to be filled by trains. I’d say the biggest is Chicago-Kansas City, on which the Southwest Chief takes 7:11 and, since the only all-freeway route detours through St. Louis, driving takes 8:33. But this is a much longer distance, and the route is served by air. At shorter range, some other options I’m thinking of are Chicago-Fort Wayne and New York-Albany-Burlington. Any other suggestions?

Edit: for a similar view on frequency, see this rant, sourced to, I believe, the URPA. There are a lot of things in there that are just insane, but the point about financial performance improving with service levels is true. Too bad the implication is that those extra frequencies belong on long-distance rather than medium-distance trains. With the same equipment as just one extra long-distance run, Amtrak could run 4-5 times daily frequencies on an important corridor run.