Saturday, August 22, 2015

How the North South Rail Link could work

North South Rail Link schematic showing current train volumes for Commuter Rail
lines, potential future RER-style connections and new and existing stations.
Discussion about the North South Rail Link has bubbled up again, with Mike Dukakis and Bill Weld writing in the Globe about how it should be a priority, Commonwealth magazine calling them out on where the money would come from, and Dukakis replying that if you wanted to find the money, you certainly could. A thread started on Universal Hub with a lot of people wondering how the rail link would function. Here are my thoughts/explanations.

Efficiency. The crux of the issue is that MassDOT wants/needs more space at South Station (and eventually North Station) and is limited by the number of tracks. Turning trains around is time-consuming: you have to unboard (deboard?) the train, board the train (even at rush hour, 50-100 or more reverse commuters may be waiting), have the crew change ends, and then thread out the same slow trackage you came in. Even with the best of practices, a terminal station track can only accommodate a train every 15 minutes or so. But if trains run through, boarding and alighting is faster and easier, there's no need for the crew to change ends, and with more destinations,  fewer passengers get off at each stop. You could route 20 trains per hour per track, meaning that four tracks in the rail link could do the work of 20 at South Station (currently there are 13; the expansion would add about a half dozen more).

Electrification. You need to electrify the whole system (and maybe Amtrak to Portland). Maybe you could get away with dual mode engines, but that's lipstick on a pig. Electrification of the Commuter Rail network would cost somewhere in the neighborhood of $1 to $2 billion dollars. Not cheap. But huge operational efficiencies: electric trains are cheaper to run, cheaper to maintain, and have faster acceleration meaning that trip times (and staff costs) are shorter, while ridership is higher (faster trains = higher ridership).

North/South imbalance and the Grand Junction. The idea behind the North South Rail Link is that trains coming in to South Station run through the tunnel and out one of the North Side lines. This is the Philadelphia model. This works well in theory (some passengers would get a one-seat ride; others would have to change), but in practice more trains run in to South Station than North Station (although this was not always the case; in 1972 there were twice as many passengers at North Station). Several pair off relatively well:
Old Colony Lines-Eastern Route
But falls apart when you realize that all the North Side routes are taken and you haven't accounted for the Worcester or Fairmount Lines. You could route them through to other terminals, say, to Anderson Woburn, but this seems wasteful; you don't need a train every 8 minutes inbound from Woburn. This is where the Grand Junction comes in. If it were grade-separated in Cambridge (or maybe even if not; there are grade crossings in downtown Chicago with 20 trains per hour at peak times), Worcester Trains could split off at West Station (hopefully) and make a loop (much like lines in Melbourne, where a similar imbalance existed), serving Yawkey, Back Bay, Downtown and Kendall. As for the Fairmount Line …

Potential for RER(/S-bahn/Crosslink)-style service. Having the rail link in place with the Grand Junction connection would not only allow you to route trains through the city, but it would also allow you to implement RER-type service filling much of the urban ring and providing more frequent service than Commuter Rail (although Commuter Rail could run more frequently as well). If all the paired lines shared one set of tunnels, the Worcester Line could share the other with the following RER-type services:
Allston to Assembly via Cambridge (Kendall) and Sullivan
Readville to Allston via Downtown and Cambridge (Kendall)
Allston to Chelsea via Downtown and Sullivan
These would operate frequently all day, every 10 to 15 minutes, allowing transfers between subway and Commuter Rail. This leverages the capacity of the rail link to create a complementary urban rail network to provide new connections and take some pressure off the existing hub-and-spoke network.

Get rid of North Station. Most ideas about the Rail Link include three downtown stations. But you only need two. Where North Station is today is convenient to very little, and the half mile to its north is taken up by highway ramps and water. A better idea would be a station between Aquarium and Haymarket. It still allows for transfers to all subway lines, and there is very little that is less than a 10 minute walk from North Station but more than a 10 minute walk from Haymarket. It would save money, too. (Obviously, the existing Green/Orange line station would remain.)

Open new land to development. Much of the land near North and South Stations is taken up by trackage, and the Rail Link, obviating the need for these stations (a couple of tracks to use in emergencies and for trains like the Lake Shore Limited might make sense, at South Station especially, although these trains could operate to West Station and provide connections there instead), and this would be very valuable land. In addition, it would render Boston Engine Terminal obsolete as maintenance would be for electric vehicles, and if maintenance facilities were rebuilt on less valuable land (say, in Billerica or next to the Anderson/Woburn station). Based on what Northpoint just sold for, it would be worth maybe half a billion dollars in development rights. A lot of this money could be captured to help pay for the rail link itself.

Will the North South Rail Link happen any time soon? Probably not. But hopefully this will help people understand its potential (which is huge!) and what we're talking about when we talk about the NSRL.


  1. Alternatively, a single station with one end of the platform under the blue / green intersection at Government Center and the other end under Hawley / Summer at Downtown Crossing might work, at the cost of losing the direct SL1 / SL2 connection.

    You should be able to buy enough Tesla Powerpack capacity for $2 billion to store more power than 80 P42 locomotives can produce in 24 hours of running full throttle, which I think demonstrates that batteries and not overhead lines are going to be the way forward if the T is open to actually buying current technology. Weight distribution may require battery EMUs instead of locomotive hauled trains, and door configuration is going to be a major factor in dwell times if everyone is going to get on and off at one or more expensive underground stations; single level cars with doors at the middle of the cars in addition to the ends may end up offering a better system cost per passenger per hour capacity than multi-level cars with narrow doors.

    Pairing lines may not be optimal. Giving south side residents a one seat ride to Porter and Sullivan and Chelsea may be useful, and giving north side residents a one seat ride to Back Bay, JFK/UMass, Quincy Center, West Station, etc may also be valuable.

    1. To clarify, if you can get 15 minute or so headways from an outer suburb into the NSRL, providing hourly one seat rides from the outer suburb to various far side employment centers has potential to increase the value of the system, and may suggest that Fairmount Line stations could become strong employment centers in the future, since north side, Worcester, Providence, Franklin, and Stoughton (and possible extensions beyond Stoughton) lines would be able to provide one seat rides to the Fairmount Line.

      However, the NSRL may not support 15 minute headways into the NSRL itself for every branch. But 15 minute headways from Middleborough to Quincy Center might work if the entire Middleborough to South Station route were double tracked, and then perhaps Middleborough could have hourly service to each of the South Station surface platforms, NSRL, West Station, and Track 61, with Quincy Center or JFK/UMass acting as a transfer point to a train from Kingston or Greenbush for Middleborough riders who just miss the hourly one seat ride to their preferred downtown destination.

    2. A mode of motive power that must carry its energy source will always be inferior to a mode that can off load its energy source. Trains that draw their energy from electric wires overhead or third rail have the full might of a major power plant behind them. Batteries can never compare to such power/weight ratios.

      The trains will need very strong traction power to make the steep slopes of the link tunnels navigable in reasonable time. Let's not pretend otherwise.

      Fast acceleration out of stations is also important for making schedules work.

    3. Tesla has no problem achieving 0-60 in 3.5 seconds or less with a battery powered four door sedan that weighs significantly more than gasoline powered sports cars with similar 0-60 times, and transit generally shouldn't be accelerating anywhere near that fast.

      If a 280,000 pound railroad car is 70 times heavier than a probably roughly 4,000 pound Model S with probably a roughly 1,000 pound battery pack, and you're content to have the train take five times longer than the model S to do 0-60, the train car probably needs roughly 14,000 pounds of battery (5% of its mass) to get adequate acceleration (adequate range may require a bigger battery, however).

  2. In particular, the Tesla Powerwall has a peak output of 3.3kW. An ACS64 has a peak power of 6.4MW and if you include HEP and such you can round it up to 6.6MW, which makes a nice even 2000 Tesla Powerwalls to make sure the locomotive can haul itself up to speed. At a cost of $7 million, that more than doubles the price of the locomotive. And that's not even mentioning the 200 metric tons that it would add to the mass... even spread throughout a train of MUs, that's not exactly a small amount of mass to be carrying around.

    1. The 3.3 kw was in Tesla's original announcement, but Tesla announced an increase in peak power to 7 kw at no cost increase after potential customers complained they wanted more.

      By the time you account for the actual power Tesla is delivering, and scale to PowerPack pricing of $250/kwh which I assume is basically a bulk purchase discount, you might be looking at more like $3 million or less per locomotive if we're just looking at the acceleration and not worrying about range, and don't forget that the elimination of the diesel prime mover and/or overhead power lines offsets at least some of the battery cost.

    2. Don't forget the 100 metric tons, though.

    3. Just, no. We have perfectly good existing technology (overhead catenary) to deliver power to rail vehicles. It works very well. It's relatively cheap ($5m/mile). It doesn't degrade over time. It is proven. Why would we use something that is unproven technology from a company which doesn't yet make a profit?

  3. Don't forget that electric trains are also much cleaner than diesel. Diesel trains emit a LOT of ground-level pollution, especially when they start up again after stopping. (Ask anyone who stands in Back Bay Station for any period of time.) Converting to electric would be a huge improvement for public health in addition to being quieter and more efficient.

    1. I hear VW is making some very efficient diesels these days. Oh, wait.