[Alon Levy’s] post on The Transport Politic about through-routing commuter rail in New York brought up a question I’ve had for several years regarding transit systems. Why isn’t through-routing more common? This applies to rail, BRT, regular bus, etc. It seems that through-routing all or most of a city’s lines via a central transit center provides all the benefits of the “hub-and-spoke” model but also eliminates the need for transfers for a significant minority of people. Is there a downside or cost that isn’t apparent at first?
Through-routing means that a route designed to carry people to/from a downtown doesn’t end in the downtown; instead, it flows across downtown and out the other side as a different route. In Alon Levy’s vision of New York area commuter rail, for example, the trains flowing into Manhattan from Connecticut (red on the map) would flow through the city and out onto southern lines in New Jersey, and vice versa. Through-routing means that downtown is in the middle of a line, not at the end. You can do the same thing with any kind of radial (i.e.to/from downtown) transit service, bus or rail.
Through-routing has these colossal advantages:
- Fewer “transfers.” Some people are actually going from Connecticut to New Jersey, for example, and through-routing lets them make this trip without changing trains. More commonly, a lot of people from the north (i.e. Connecticut) are going to southern parts of New York City, while a lot of other people from New Jersey are going to northern parts of the city, and a through-routed system serves both groups, which are briefly on the train at the same time in the city. Because of the decentralised structure of Paris, lots of Parisians are riding across the city to the far side, in both directions, so the RER’s through-routed structure is absolutely essential to avoid forcing huge masses of people to change trains.
- Reduced need for terminus facilities on expensive downtown real estate, thus potential for higher frequency. Ending a line downtown means having facilities to store a bus or train for at least a few minutes, consuming expensive space. Trains typically reverse direction at an end-of-line station, so the driver needs to close her cab, walk the length of the train, and get herself set up on the other cab; she may also be entitled to some break time. A train occupies one of a limited number of rail tracks while this is happening, so this function becomes THE limiting factor on the frequency of the whole line. Downtown, there are lots of physical and cost constraints on station design, so you almost never have as many rail tracks as you’d like. Buses need space to turn around. Their drivers, too, are entitled to some break time at the end of a trip, so end-of-line stations on frequent services need space for a number of buses to pile up in a first-in-first-out queuing arrangement. All this takes a lot of space. This space is a lot cheaper at the end of a suburban line than it is in the middle of downtown.
- Fewer line ends for reduced operating and capital cost. The time it takes to turn a bus or train around, and provide the driver break, is usually not related to the length of the line. Through-routing two routes eliminates two ends-of-lines, which reduces the cost, both operating and capital, of those inefficient turnaround movements. Often, through-routing two lines actually reduces the number of buses or trainsets required by one or two.
- Fewer vehicles downtown providing the same service. Sometimes, the pre-through-routed lines overlap in downtown. Through-routing eliminates that overlap. Instead of having a bus dropping off passengers interacting with another bus picking up passengers, you have one bus dropping off passengers and picking them up at the same time. For buses especially, downtown street capacity is a very limited resource in big cities. Through-routing helps economise on it.
So why isn’t there more through-routing?
- Unbalanced markets on the two sides of downtown. There’s never an exact one-to-one match between routes approaching from one direction and those approaching from the opposite direction. For example, San Francisco’s downtown is on the bay at the northeast corner of the city, so there are no routes extending north and east that could be paired to routes flowing south and west. To the extent that such through-routes have been created (e.g. San Francisco’s Muni Metro T line) the result is a circuitous approach to downtown for one of those lines.
- Excessive line length. The probability that your train or bus is delayed is directly related to how long it’s been running since it last had an end-of-line break. (When a vehicle arrives late at the end-of-line, its break time is reduced so that it can leave on time or at least not as late.) Through-routing makes lines longer, so it can compound this problem. This is obviously more of an issue in services that are exposed to more causes of delay, such as services in mixed-traffic and services with driver-administered fare collection.
- Jurisdictional barriers. Finally, there are plenty of cases where through-routing would be in order but the two sides of downtown are different jurisdictional turf. They may be different states, as in New York, or different transit agencies. In the UK and Australia, they may be different private operating companies. Although both are government-subsidized, UK/Australia governments tend to defer to private companies about network design, so you often get designs that reflect company turf boundaries rather than efficient use of subsidies or meeting the needs of the customer. (London is an exception. Some Australian states are also working on this problem.)
- Infrastrucutre barriers. Your great-grandparents’ jurisdictional barrier is often your infrastructure barrier. New York City is the obvious example. Because each commuter rail line was designed by a separate entity, and each of these entities was thinking only about getting people into Manhattan, the terminal stations are not physically connected in the way you’d need in order for trains to flow through. This is the cause of most of the capital expense in Alon Levy’s proposal for New York above.
- They just haven’t thought of it. If your transit agency has non-through-routed lines, and you can’t figure out why, send them a link to this post and ask them!
One related phenomenon are U-shaped lines, such as the DC Metro Red line, which starts in Maryland northwest of downtown, heads southeast into DC, hits key tranfer points like Metro Center… then turns back north and heads into Maryland again but further east.
Not to pick too much on the DC Metro, but is this a useful through route? It’s worth noting that the Red Line is the only line in the DC Metro system which doesn’t share its tracks with any other line; but if you want to get from Shady Grove to Glenmont, there are more direct routes. If and when the proposed Purple Line gets built; you’ll be able to take the train with a fraction of the distance. I’ve actually ridden the Red from the AU/Tenleytown area to Takoma Park, but I was a transit-dependent tourist who didn’t know any better way(s) to make that trip.
In the full historical context of Metro construction, the route of this line might make sense; but I’ve always thought that the Red and Blue lines east of Metro Center ought to be swapped. (Doing so would require significant track rework; as the Red enters the station on a different platform/track than does the Orange/Blue/Silver multiplex. Of course, the serpentine route taken by the Blue Line is another interesting debate…)
The reason for the Red Line’s route is probably more a matter of balancing service levels on the two branches than anything else. When the Washington Metro was being designed, I strongly suspect the goal was to facilitate suburb to downtown commutes and little thought was given to anything else.
By the way, another problem with through-running is that sometimes the system is set up for asymmetric running. For example, the Park Avenue line on Metro North has 3 tracks in the peak direction and 1 for reverse-peak. There’s also potential issues with delay propagation with through running. With a 24tph scheduled service through the Hudson tunnels, there’s no room for recovery at all, and the East River tunnels are pretty close to capacity too. Through running with the layout of stations that we have most likely means recovery happens at Penn Station, which means you still need platform space to hold trains. Running them through to either West Side Yard or Sunnyside is actually more efficient in some respects, as inbound trains in the morning can be sent off as soon as they’re empty and there’s a slot available, and afternoon trains have much less potential to run into delays on the way to the station. And of course there’s the whole power supply issue.
Looking at this from a bus perspective as well, I would say that overall route length and the inevitable cumulative delays are THE primary reason more bus routes are not through in large cities without a comprehensive regional or rapid transit system.
This is the case in my home town of Baltimore. The feeder buses that connect with our one-radial Metro are well-liked and timely, because their routes are never more than 5 miles across. On the other hand, there are routes that start from one suburb, cross the City and Downtown then enter the suburbs on the opposite side. Consequently, they are the most hated, yet among the heaviest lines in the system. It certainly justifies with my research expansion of a grade-separated rail for our City.
(WRT through implementing rail lines, this is also why I oppose any regional/rapid projects to use the rump end tunnel curiously placed beyond the west “knock-out” walls at the Lexington Market Metro Station. This would only force everyone going beyond that point to transfer to the Metro and overload the system as well as wasting significant amounts of time.)
It’s not just the infrastructural or jurisdictional barriers, it’s also the lack of a champion to push to break through them. An example: Back in 1999, there was a study done of MARC/VRE throughrunning. The study said there was reason to run MARC trains from Maryland through Washington Union Station at least to Alexandria in Virginia. A substantial proportion of MARC riders detrained at Union Station and continued on public transportation to offices south of The Mall or in Northern Virginia. Those would be served by the trains continuing to L’Enfant Plaza (for offices south of The Mall), Crystal City and Alexandria (for Northern Virginia), all VRE stations.
What stopped this? The fundamental barrier was a need for additional trackage and crossovers in Washington Union Station, plus some platform changes. Throughrunning trains would have to take a different route through the station (not envisaged when the station was originally laid out; different railroads), and use different platforms. Cost estimated at $17M in 1999. There were a couple of nice-to-haves: L’Enfant Plaza also has a Metrorail station where four lines meet, it would be nice to build a direct connection from the rail platform to the Metrorail concourse; a couple of layover tracks at Alexandria would obviate the need to deadhead back to the Union Station yards.
$17M is not a lot of money. But making this happen involves an awful lot of players: two states and the District of Columbia (and therefore the US Congress), MARC and VRE, the tracks are owned by CSX and Amtrak, the nice-to-haves involve WMATA and Norfolk Southern. Who cared enough to push these players even to meeting?
The study was done in 1999. Nothing has happened since.
The jurisdictional barriers are smaller than they seem, given agencies that are willing to cooperate. In Japan rapid transit is run by multiple private companies as well as public ones, but there is still through-running of services, even across different owners. Some of the lines of the Tokyo subway are through-run with private commuter rail lines; on its journey, a train on such a line runs first on one company’s suburban track, then on the subway, and then on another company’s suburban track. Likewise, in Paris one RER line is owned half by rapid transit operator RATP and half by national railway SNCF; trains have no difficulty running on both the RATP-owned section and the SNCF-owned section.
Unbalanced markets are only a big problem in cities like San Francisco, whose development is constrained by water. Most other cities suburbanize in a uniform circular or semicircular pattern, which makes it relatively easy to match lines.
Are you as optimistic about jurisdictional aspects of your NYC proposal? I understand similar ideas for MRC+VRA were discussed in DC. Don’t you quickly run up against attitudes within each agency that see such integration as an “existential threat” to their turf?
When the Regional Connector project is completed in Los Angeles, it will allow for this sort of travel to take place. There will probably be trains from Pasadena to Long Beach and from Santa Monica to Whittier going through Downtown.
Most Los Angelenos haven’t even heard of this project, but it will go a long way towards creating an actual “system” here.
I’m not very optimistic about jurisdictional aspects in New York. This is because New Yorkers of all social classes hate New Jersey, and have come to believe that all Port Authority does is subsidize New Jersey at New York’s cost. This makes public acceptance of a unified operator improbable.
Through-routing trains while letting each agency keep its turf is possible, but is somewhat more difficult than elsewhere, as the two agencies that would share the initial line in the system, Metro-North and NJT, don’t even share the same station.
While convincing people on the New Jersey side of the system’s advantages should not be difficult, on the New York side one would need to get Metro-North to serve a new station, or else to get the MTA to tell Metro-North to do so. On the other hand, now that the MTA chair previously headed the LIRR, which would see immediate benefits from reduced congestion at Penn Station, the agency might be more amenable to this proposal.
I’m not sure that convincing the New Jerseyans will necessarily be easier. There’s definitely a contingent pushing “NJ Transit for New Jersey” and is trying to oppose what they perceive as its transformation into “Metro West”. Anyhow, even if you manage to get NJT and MNR to cooperate, what about the technical issues? MNR’s trains can’t currently run into NYP, and the future trains won’t be compatible with 25Hz, but will at least be able to get to NYP on the LIRR’s third rail. There’s absolutely nothing that’s compatible with both the NJT and LIRR side electrification, at all, so it would have to be dual-mode diesels. Perhaps as an interim measure, MNR can pay NJT to extend 1 train per hour from NYP to New Rochelle, for a relatively transfer to the New Haven Line. This mostly keeps schedules decoupled, keeps union work rules more or less intact, and builds on the inter-agency cooperation of the MNR-to-Secaucus special service (which is run with NJT trains).
The existing Metro-North and LIRR rolling stock can probably be converted from one electrification to another with little difficulty. The voltages are nearly the same, so the only change that’s needed is the physical location of the shoe, which I’m sure can be done at a fraction of the cost of new rolling stock.
Once that’s done, you can move catenary-only trains from the Morris and Essex Lines to a combined Northeast Corridor from Trenton to New Haven, and dual-mode trains to a combined M&E-LIRR system.
As for the politics, I know that there are people who have a Jersey-centric approach to things. But as far as I know there’s nothing like the raw hatred you find on the other side of the Hudson.
This is how the SEPTA Regional Rail works. Unfortunately, too many visitors get confused because of the numbering system, which requires you to know both the destination and route number when you are boarding a train. Only 33% of trains are throughrouted with the original destinations at each end anyway. http://www.philly.com/philly/news/local/20090722_SEPTA_seeks_input_on_Regional_Rail_name_changes.html
I find the confusion weird, from the perspective of trying to run regional rail more like rapid transit. Subway lines don’t change names in the center of the city – people are expected to know which direction to take them in. When I first took the SEPTA, I checked which station I had to get off at. Then I checked the line number and destination, as I would whenever I needed to use a new subway system. To guard against mistakes you post individual maps of each line at its stations, and systemwide maps everywhere. At each station you also post clear directions; for Philadelphia it’s more difficult than elsewhere, but there’s nothing that says the three center city stations can’t have signs like “To West Trenton, Doylestown, …, Chestnut Hill” and “To Trenton, Thorndale, …, Airport.”
The obvious solution to route delays as a result of excessive line length is to split the route in two on either side of downtown, but provide an overlapping section through downtown (where demand is typically higher). This is the planned solution to the chronic delays and streetcar bunching on the 501 Queen streetcar route in Toronto, Canada, which is currently one route which crosses downtown – on weekdays, it will be split into two routes with a long overlapping section – http://stevemunro.ca/?p=2539 – to reduce delays. This is much more of a problem for surface routes, it shouldn’t be much of a problem at all for rapid transit routes.
The problem with splitting everything downtown is that dwells at a terminal are much higher than at a through-station, which means the downtown station will be clogged with traffic.
Yes, Alon is right. We often get the same effect by through-routing a long route to a short one, but there’s always a tradeoff between a too-long running time and the high cost of terminals in or near the CBD.
You mentioned Muni through-routing in San Francisco, which it barely does, but you forgot BART, which totally does. Several lines go from the East Bay, through the city and down the Peninsula.
I would just like to add my thoughts on this matter:
The problem with uneven loading on both sides of the downtown area can be solved by only having some trains go through the downtown area, while others simply terminate.
For example, if, at Penn Station in the AM rush hour, there are 6 inbound trains per hour from the Port Washington Branch of the Long Island Railroad and 2 outbound trains from the NJ Transit Northeast Corridor Line, that means that 2 trains run the full route and 4 trains go out of service.
The problem is, as Alon Levy said, that customers might be confused as to which trains run the full route, but people on each end of the route can think of it as a short-turn (in this case, people in Long Island see trains running the full route to New Jersey or the short-turn to Penn Station.
Speaking of NYC, a lot of subway lines (13 out of the 19 that enter the Central Business Ditrict (CBD)) enter Manhattan from the south and then exit through the north, but rapid transit is different than commuter rail in terms of demand for through-service.
A bird in the hand is worth two in the bush.