» A proposal for New York regional rail improvements.
This is the first post in a two-part series by guest author Alon Levy. This part explains the reasoning behind the proposal and the general principles underlying it. The second part, coming tomorrow, will describe the regional rail network that would emerge from this proposal.
Most proposals for commuter rail expansion in New York focus on the ends: extending lines further out, extending electrification, extending multi-tracking. The only improvements at the center, East Side Access and the ARC tunnels, are stub-ends, in line with the current thinking that commuter rail exists to get people from the suburbs to the CBD.
Although this form of operation seems natural in New York, many other cities have instead improved their regional rail system by combining stub-end lines to form lines that go from one suburb to another, through the central city, making for convenient travel between inner suburbs. Almost all subway lines work like this: nobody expects the A to run from Upper Manhattan to Midtown and discharge all riders, with a separate line running from Midtown to Queens.
A better framework for regional rail improvement in and around New York should focus on revamping the LIRR, Metro-North, and New Jersey Transit along the lines of the Paris RER, the SEPTA regional rail system, or any S-Bahn system—that is, with through-routing and a priority placed on easy transfers from each line to the others, and on serving multiple city center destinations from each line. At a time when the current thinking about urbanism is to have seamlessly integrated regions, with central cities and suburbs working together rather than competing, it is ideal to run a region’s rail system like a single city’s.
Ideally there might also be integration of governance, under one agency, but this is not necessary. After all, some subway systems, such as Tokyo’s, are run by multiple companies. And the RER is run by two separate agencies, the local transit agency RATP and the national rail SNCF; the RER line B is run half by the RATP and half by SNCF, and yet commuters can travel from one agency’s section to another on a single train, without a significant wait at the boundary. No such through-running is offered today in New York.
The Need for Through-Running
The New York metro area has many stub-end terminals—Flatbush Avenue, Grand Central, Hoboken, Long Island City, St. George—as well as one station, Penn Station, which is a through-station by layout but a terminal by use, except by Amtrak. Such a configuration works in getting people to take commuter rail from the suburbs to Manhattan, but is inherently limited for all other functions.
To get from Westchester to Midtown Manhattan is easy: one need only take Metro-North to Grand Central. However, to get from Westchester to Brooklyn, or Newark, or Queens, one needs multiple transfers. The other suburbs are just as their limited in their access to secondary downtowns. Thus, at their most transit-oriented, New York City’s secondary job centers have about a 50% rail modal share. Suburban job centers are even worse: Stamford and White Plains have a rail modal share of only a few percent.
Manhattan acts as a barrier to transportation, both by auto and by rail. By train, one needs to transfer. By car, one needs to cross jammed roads and pay multiple tolls. Through-running is a way of breaking this barrier by enabling people to live in North Jersey and work in Queens and Brooklyn, Long Island, or Connecticut, and vice versa. Though some people live on one side of Manhattan and work on another today, the current stub-end use of Penn Station lengthens those commuters’ travel time and restricts their number.
Worse, the stub-end layout reduces track capacity. A rapid transit train can dwell at a through station for under a minute, even if it is crush-loaded with passengers trying to enter or exit. At a terminal, the minimum dwell is about five minutes, and mainline trains discharging all or most passengers at the terminal typically dwell more. This clogs the tracks, leading to the absurd situation that while the RER’s central transfer point, Châtelet-Les Halles, serves 500,000 daily passengers on 6 tracks, Penn Station strains to serve 300,000 riders on 21 tracks.
Such clogging increases costs down the line: the main reason the ARC tunnel is being built to feed into a new deep-level station rather than into the existing Penn Station tracks is that the existing station is unable to handle the additional load without through-running. Thus the project’s budget is billions costlier than would be an extra two-track tunnel connecting New Jersey with Penn Station.
The RER Example
The current situation in New York is similar to that of Paris until a few decades ago. Paris famously has six major train stations, each serving trains to a different part of France, with no easy interconnections. The Métro proved impossible to expand outward after World War Two: its average interstations were very short, and there were no express tracks for trains, which kept average speeds down. Thus, as the city sprawled, it found itself relying on mainline rail for its rapid transit needs. The existing SNCF suburban lines proved inadequate in getting people to the emerging job centers, and had difficult transfers.
To deal with the problem, RATP bought two unprofitable commuter lines from SNCF and connected them with new subways. In 1977 the first line, RER line A, was complete, with an east-west connection through the La Défense business center and the heart of Paris. It was built with much longer average interstations than the Métro, so that its average speed was double, enough to attract suburban commuters. The system was an immediate success, spawning four more lines. Nowadays, Paris has a billion commuter rail riders a year, mostly on the RER; New York, with twice the metro population, has 250 million among its three commuter rail agencies.
The RER’s primary innovation over traditional commuter rail, besides through-routing, is that it treats itself as rapid transit. It keeps dwells short even at crush-loaded Châtelet-Les Halles. It is fully electrified and double-tracked. It is signaled as rapid transit. Altogether, its maximum frequency is one train every two minutes in the peak direction on line A, or 30 trains per hour, on a par with a typical subway system. Traditional commuter rail can’t reduce frequencies under two and a half minutes, or 24 tph. An increase in frequency would alleviate some of the capacity issues on the LIRR and NJT, which are hard-pressed against their 24 tph limit.
Another way the RER looks more like rapid transit is that it provides some local service, on a similar level as the NYC subway’s express lines: it serves major neighborhood centers and local job centers, rather than just downtown Paris and the suburbs. In New York such local commuter service exists on the LIRR, whose lines in Brooklyn and Queens are much like super-express subways without fare gates. However, Metro-North barely has any local service, and the NJT has none. Expanded service increases ridership; New York’s regional rail system should aim to create infill stations on existing lines in Hudson County and in Queens, as well as ensure future city center construction serves major non-CBD destinations such as 14th Street. It should also avoid skipping important transfer stops such as Jamaica, Harlem-125th Street, and Secaucus Junction.
Not all through-routed systems have worked so smoothly as the RER, and none is as large. The Berlin and Munich S-Bahn are successful; if they have lower ridership, it’s because they serve smaller metro areas than Paris. However, in Philadelphia, through-running on the SEPTA did not lead to higher ridership; the problem was that shortly after the new through-routed system opened in 1982, the rail workers went on strike for seven months, causing ridership to halve. Notably, despite the SEPTA’s failure, other American cities are trying to engage in similar construction: Boston would like to connect North Station and South Station, San Francisco is planning rail service to Transbay Terminal in a way that will enable a future extension across the Bay into Oakland, and Los Angeles is planning to reconfigure Union Station to have run-through rather than stub-end tracks.
Future Connections and Development
The current market for cross-Manhattan travel is small, due to the difficult transportation. Initially, running trains through Penn Station will serve more to simplify operations and reduce costs. It will increase ridership because currently Penn Station is not configured well for transfers—New Jersey Transit and LIRR don’t even share the same waiting concourses—but the increase will not be large immediately. A greater increase in riders will come from new connections, and eventually more development.
First, connections: the Metro-North New Haven Line could already run trains to Penn Station, if only the station had more track capacity. Through-running will create this capacity: for instance, the NJT and Metro-North could agree that every Northeast Corridor Line train will continue to the east as a New Haven Line train. New Jersey Transit is planning to provide direct access between the New Haven Line and Secaucus starting this summer for Jets and Giants games in the Meadowlands.
Simultaneously, the MTA should build new stations along the Hell’s Gate approach to Penn Station from the east; current plans call for a few stations in the Bronx, and there should also be stations in Queens as far as the grade permits. Such work would make it easy to get not only from the New Haven Line to Penn Station, but also from Newark to Queens, the Bronx, and Westchester, and vice versa.
As the system expands, more connections would be created, both across Manhattan and into new neighborhoods. Such expansions, especially an extension from Atlantic Terminal to Lower Manhattan and tunnels from Lower Manhattan to Grand Central and New Jersey, will do a lot to shorten Outer Borough commutes, which are the longest in the nation. The authorities may have to build a few more stations than its plans currently call for, especially in Sunnyside and Inwood.
Second, in the long-term, transit should encourage more transit-oriented development. Plans for neighborhood improvement should be articulated for new stations, such as those included in the aforementioned plans to bring New Haven and Hudson Line service to Penn Station; at inner-suburban stations, which will suddenly be accessible from a far larger section of the region, making them more attractive for office and retail space; and around stops in existing edge cities, which will be rail-accessible for more people, potentially spurring more TOD.
The extra development near new stations is the easiest to pursue. Most of those stations will be in areas that are already dense and walkable. Those proposed for Hudson Line service into Manhattan—125th Street and 62nd Street—are already undergoing gentrification. The Bronx stations proposed for the New Haven Line are not so gentrified, but Hunts Point is located next to a run-down industrial zone. An extra stop at Sunnyside could come with air rights over the adjacent Sunnyside Yards, which one hopes the MTA would sell more competently than it has Atlantic Yards in Brooklyn.
More difficult is upzoning near other stations, in anticipation of greater traffic from new connections; inner-suburban stations would gain the most from these links. It would be straightforward to create a high-density transit-oriented residential district around Secaucus Junction. But it’s less straightforward to gauge accurately whether there is demand for a mixed-use district with intensive retail and office space. We can assume that increased regional rail ridership will require more commercial development in Newark and Jamaica, but elsewhere it involves guesswork. Would Yonkers suddenly attract commuters from lines in Long Island that are joined to the Hudson Line? Would Secaucus Junction become a hot destination in itself? If Hoboken were connected to Lower Manhattan by commuter rail, how much extra ridership would there be on the Hoboken Division?
A libertarian would suggest that the government should leave this analysis to the market. However, in practice the market begs for zoning regulation, and if the government doesn’t provide it, as in Houston, then it will resort to restrictive covenants, which are far harder to change. In fact, Jane Jacobs argues that developers build what they’re used to, which is single-use auto-oriented development—hence, Houston’s urban form. To ensure this doesn’t happen, local governments will have to upzone rather than dezone.
Transforming existing edge cities to TOD is the most difficult form of development here. It can be done: Tysons Corner, a Northern Virginia edge city so auto-oriented that it has a secondary rush hour during lunchtime, is planning to convert itself into a series of walkable neighborhoods in anticipation of future Metro service. It plans to punch through-streets through subdivisions to create a grid and build sidewalks for new mixed-use development.
However, to exploit the regional rail system’s full potential, more is necessary. White Plains is walkable by edge city standards, as are New Brunswick and Mineola. The greatest problems are in edge cities without walkable cores, such as Stamford, the Edison-Woodbridge-Metropark cluster, Princeton-Princeton Junction, and the string of cities on the Hudson centered around Tarrytown. Metropark is a gigantic park and ride located next to I-95, with plenty of office parks a short drive but a long walk away. Stamford is somewhat more urban, but its street layout is barely more pedestrian-friendly than Tysons’; employers there have to run shuttles from the train station to their buildings. Such cities will have to engage in Tysons-style street grid conversion and replace parking with TOD; both actions will be contentious among suburbanites.
New York should follow the example of Paris and reconstruct its commuter rail system along the lines of rapid transit, which include through-routing first and service to multiple urban neighborhoods second. It should exploit the fact that Penn Station already has a run-through configuration, enabling through-routing without new construction, and at the same time engage in additional construction to create a similar station in Lower Manhattan.
The obstacles to implementing the above program are more political than technical. But even politics is less intractable than it seems, given the right impetus. After all, a year ago it seemed inconceivable that the US would ever entertain a $50 billion high-speed rail bill; now the debate is whether to introduce it now or in 2011. The same may be true for expanding regional rail, which, despite its lack of glamour, carries far more people than HSR.
The best time to construct such a system was 50 years ago, when the region was just beginning to suburbanize. However, as more and more suburbs try to adopt a traditional urban layout, the next decade could be a reasonable second best.
Read the second part of this post.