Boston South Station’s Supposed Capacity Limit

I don’t have much to add to Yonah Freemark’s post about Boston’s proposed Fairmount Line infill; as Yonah correctly notes, this is a good idea in principle, but in practice it also requires operational integration, especially unified fares. The current federal aid system gives agencies a large incentive to install concrete, some incentive to install electronics, and none to improve organization.

What I want to discuss in this post is the myth that Boston’s South Station has capacity problems, a myth almost as pernicious as the same myth about New York’s Penn Station. While South Station can’t immediately solve all of its capacity problems with through-running (P.S. note the cost estimates for 2.4 km of tunnel in Boston are $3-9 billion), it still has enough tracks for service increase. Thus the 20-minute frequency limit mentioned in the comments to Yonah’s post is not as binding as the MBTA may think.

South Station has 13 tracks. These naturally separate into a group of 4-5 to the east and a group of 8-9 to the west. The eastern tracks are fed by a four-track bridge serving the Fairmount Line, the Old Colony Lines, and the Greenbush Line. The western tracks curve 90 degrees (with radius, I believe, 250 meters, limiting approach speeds) west and become a four-track line reaching Back Bay, and fanning out to the Worcester, Providence, Needham, and Franklin Lines;  the Providence Line also hosts Northeast Corridor intercity trains, while the Worcester Line hosts a single daily Amtrak train.

For all intents and purposes, the two sets of tracks should be treated separately, for the following reasons. First, any train, any track is good to have as a contingency, but should not be done regularly, in order to make service as predictable as practical. Second and more importantly, the capacity of a terminal is far higher when the trains are completely interchangeable, as they are to the east. If slight schedule irregularities create conflicting terminal moves, the run can be done from any track.

In the simplest case, that of a two-track line hitting a two-track terminal with (short) tail tracks, the turning capacity can approach 30 trains per hour, the same as that of a running line; see for example the schedule, satellite view, and station map of the Chuo Rapid Line. This is uncommon, but many other commuter lines in Japan turn 12-15 tph on two tracks.

The four-track eastern segment of South Station can be split without revenue conflicts into two western tracks serving Fairmount and two serving the other lines, and such capacities become realistic. Since total peak traffic on the Old Colony and Greenbush Lines is currently 6 tph, and total peak traffic on the Fairmount Line is 2 tph (should be 6 tph for good urban service), capacity there is a non-issue. Although there are no tail tracks at South Station, all platform tracks except the easternmost are long enough that they could attach to platforms a few tens of meters longer than an eight-car commuter train, which with modern rolling stock should suffice.

The western tracks pose a bigger problem, for two reasons. First, the trains are not perfectly interchangeable, and do not separate neatly into two two-track lines running alongside each other. Second, Amtrak should be planning on 400-meter trains, and although the platforms could be lengthened to accommodate them, tail tracks become impossible, forcing even slower approach speeds than required by the curve.

Regardless, South Station has enough capacity even for trains serving Back Bay. With completely non-interchangeable intercity trains and dwells that are long by regional rail standards, the Tohoku Shinkansen turns a peak of 14 tph using four station tracks at Tokyo. While the Tohoku Shinkansen does not have the sharp turn of South Station, the MBTA can turn trains faster (trains already turn in about 5 minutes at the outbound terminals), and all services but one use the same equipment. So the capacity for South Station West is at a minimum 28 tph; current peak traffic excluding Amtrak is 12 tph.

It goes without saying that the operating assumption I’m using is that service is run well, better than is currently possible under the FRA-regulated regime. Among the FRA’s sins is brake tests at every terminal, forcing longer dwell times than are routine in Japan, France, and other countries with a much safer rail record than the US, to say nothing of American rapid transit (which outside Washington D.C. is very safe). While all of the above examples of high turn capacity use EMUs with high acceleration and deceleration, the separation between maximum capacity and current MBTA traffic is high enough that large service increases are possible without either more concrete or more electronics; with better electronics, even more increases are feasible.

I am going to return to this issue, specifically the Providence Line, because one way to save some money on Northeast Corridor improvements is to speed up the Providence Line, using existing electrification and new rolling stock; this would permit the line to remain two-tracked with one mid-line four-track passing segment around Sharon, obviating the need for Amtrak’s proposed third track, even with large increases in ridership.


  1. Stephen Smith

    What other FRA “sins” are there in this instance preventing capacity increases beyond the brake test? And what about the electronics upgrades you speak of? Are they also prohibited by the FRA, or is that incompetence at a more local level?

  2. Chris G

    The fact they Boston can’t get that tunnel between north and south stations is as bad as not extending the NEC to at least Richmond with electrification.

  3. Pingback: This, That, and the Other Thing | Worcester
  4. BBnet3000

    The Penn Station issue is ridiculous. This whole idea about “Penn Station South” is utterly unwarranted. The station complex is already as large as it should get, and the capacity is not an issue by international best practices (not to mention basing the idea for a huge construction project on current use when a year from now, a slight majority of LIRR trains will be going to Grand Central).

  5. Alon Levy

    @Stephen: the other sins I’m talking about are the by now infamous rules about buff strength, cant, and cant deficiency.

    The electronics upgrades the MBTA should be thinking of are good signaling, especially on the Providence Line. If I’m not mistaken the line already has ACSES, an overlay clone of ETCS, and this may be good enough for the capacity needed for more commuter trains. The FRA isn’t preventing that, but it’s not providing any guidance to railroads about which PTC system to use, and they may well get a freight-based system rather than an off-the-shelf passenger-oriented system. At least, Caltrain is trying to come up with a globally unique signaling system.

    @Chris: you’re right with regards to intercity service (though, if the goal is to have high-speed service, it’s better to cut off the Virginia line on I-95). But the primary use of the North-South Rail Link is for regional service: together with electrification and infill stations, it would create a Boston S-Bahn.

    @BBNet3000: yes, exactly. Bear in mind, Boston’s existing concrete is somewhat lower-capacity than New York’s. In addition to what I said in the post, Penn Station’s eastern access tracks are two separate pairs of tracks, whereas South Station’s western tracks are a single four-track line. Four-track lines are better for cross-platform transfers and for operational flexibility, but they create more station throat conflicts.

  6. emdx

    I look at purported station capacity issues, and I always laugh.

    Move back to the time those stations were built (100 years ago for Pennsylvania Station and GCT) and see how much more trains were run, under less elaborate signalling systems (interlocking, with 3-4 towers per station).

  7. Pingback: Bad US Rail Practices, and What It Means for FRA Regulations | Pedestrian Observations
  8. Pingback: Followup on the FRA and Amtrak | Pedestrian Observations
  9. Jeremy Steinemann

    You fail to consider train storage needs. Part of the problem at South Station is the lack of adequate yard storage. Tracks must be used for train storage even in the short term. Without additional lay-over tracks, it’s difficult to get trains in and out during the peak.

    • Alon Levy

      Train storage can and should be done in places where there’s room for it, i.e. not a CBD with expensive land. Like Metro-North and perhaps other Northeastern commuter railroads, the MBTA has the unfortunate practice of turning trains quickly at the outer end and dwelling for a long time at the CBD end instead of the reverse.

      • Richard Mlynarik

        Using CBD platform tracks as train parking lots — along with parking expensive depreciating equipment out of service 80% of the day at platforms or yards or otherwise — is deeply engrained in Olde Tyme US Railroading.

        Trivial commuter rail lines with a peak of five trains per hour (plus trivial amounts of third party/freight on top) will undertake multi hundred million dollar taxpayer-funded projects to expand terminal stations to 9 platform tracks just so they can park trains there all day. Incredible, inexplicable, incomprehensible — but true, and this is pretty much universal, uncontested and unexamined US Olde Tyme practice.

        US train stations are uniformly primarily treated as locations to park idle trains, not as high throughput passenger facilities. We have all this idle equipment, better build some idle infrastructure to match!

        • anonymouse

          Richard, that’s not actually what the San Jose station project is about. It’s partly about reworking the layout of the station throat, to reduce congestion and increase speeds, and partly about building some decently wide platforms and hopefully rebuilding the existing obscenely narrow ones (look at the lack of space between the tracks and the ramps on tracks 3 and 4). And of course that station is not used just by Caltrain but also by ACE (1 tph peak direction) and Amtrak (1 tph at best). Much as the three operators try to ignore each other’s existence, they’re all still sharing the same tracks.

          • Richard Mlynarik

            Dear Anonymouse,

            You’re wrong. Unfortunately. I wish I were making this up, but reality is worse than any rational personal could imagine.

            The insane Diridon Pangalactic “South Terminal Improvement Project” adds four platforms tracks (T6, T7, T8, T9) around two island platforms (ength 955 feet 8 inches = 291.3m).

            ALL of these four tracks connect, in a simple ladder, ONLY to the western-most mainline track (designated MT-3 north of Cahill, MT-2 south.) There are NO parallel moves of any type possible to these spanky new expensive platform tracks given this insane configuration. (The people responsible for this, as for the equally purposeless and wasteful and capacity-neutral “North Terminal Improvement Project” didn’t even understand the concept of “non-conflicting moves”. I’m not making this up!)

            The only parallel moves possible from the north are:
            MT2-T2/3/4/5 vs MT3-T6/7/8/9
            MT2-T2/3/4 vs MT3-T5 (ie the only non-trivial move is via existing crossover 13, not part of this idiotic project.)
            That’s it. Everything else causes a conflict.

            The only parallel moves possible from the south are:
            MT1-T2/3 vs MT2-T4/5/6/7/8/9
            That’s it. Everything else causes a conflict.

            This would be a laughable concept for a through station. For a nine (count ’em, none!) platform face terminal at which 3/4 of trains terminate it is disaster, a self-inflicted, $26m disaster. (And 15mph design speeds! 80kmh, 50kmh at a pinch, is standard approach speed for anybody who knows or cares in the slightest about throughput.)

            Moreever, there is no need for any of this platform capacity. None. Zero.

            Moreoever, it is guaranteed incompatible with both any sort of HSR project AND with any sort of future level boarding.

            In short, it is a total waste of $26 million of your and my dollars. It will ALL need to be redone from scratch in the near future, and will serve no purpose in the interim.

            America’s Finest Transportation Planning Professionals At Work. They simply do not get it — in fact, they don’t even understand that there’s anything to be gotten. (I know several of those responsible — they’re dumber than fenceposts.)


          • anonymouse

            Wait, are they really designing the whole business for 15 mph? I suppose that’s technically a 50% improvement in speed, but still pretty absurd. I would have hoped for at least 35, which ought to be possible even with the curve north of the station. Do you happen to have some documents that show the design speed and intended final track layout? I thought there was going to be a more thorough remodeling of the station throat, but so far it looks like that won’t be happening. I’ll have to check on the progress of the trackwork sometime soon, it’s only a few minutes’ walk from my house.

            I’m also curious about the people who didn’t understand the concept of non-conflicting moves. Did you actually speak (or write) to them about this? I think this calls for some quotes and serious public shaming. Also, off topic, but I love how you make the Mozilla devs look like such obvious fools with respect to their attitude to users.

          • anonymouse

            So I went and had a look, and it does look like you’re right about the track layout. They probably could have squeezed in another facing crossover between the southbound and northbound to remove conflicts between T6/7 and T8/9, but the current (still under construction) layout makes that impossible. That’s a level of stunning absurdity I was honestly not expecting. By the way, there has actually been one fairly recent improvement to the capacity at the San Jose station, and that’s the construction of the yard south of Tamien, which moves the ACE trains going to and from their midday layover out of the north station throat and probably also moves the Capitol Corridor train that sits at San Jose all day out of the station. And that was both cheap and quick to build.

        • Richard Mlynarik

          Anonymouse: sorry I didn’t see your request earlier.
          Here is a diagram of the final SJ Hypergalactic configuration.

          I agree with you re the Amtrak/ACE turnback/storage yard south of Tamien. That is the SOLE worthwhile and even marginally operationally useful capital project Caltrain has undertaken in the last 20 years. Incredible and sad, but true. One exception from three score or more, and over a billion dollars gone. Where can they even find the people to come up with this stuff?

      • Stephen Smith

        Richard, get a blog! You have so much insight and you’re so quotable, but you leave these nuggets as comments on months-old blog posts that are never gonna get read by anyone!

        • Richard Mlynarik

          Stephen, Nobody cares! Plus I have (aside from throw-away comments) one of the worst cases of writer’s block that anybody in the world has ever had — it’s extraordinarily difficult for me to assemble far too many diagrams and too much run-on text into anything that anybody else would choose to read.

          At one stage a few years back I paid (on the order of $10k) professional people to write things up and make nice presentations to Influence Decision Makers, but it turns out that has as much effect as making blog comments. (If you want to change public policy, write a check for $1k to some dim-witted penny ante local official! Really. Most of the people you need to get at are cheap.) I really did try to go about things the right way before giving up completely and just throwing bombs and impoliticly speaking my mind about stupid people doing inexplicably stupid things using astronomical amounts of public money.

          Thank you for for the nice words.

  10. MobilMan

    Citing Japanese examples is a bit much. They have a quasi-military operating culture (delays measured in seconds) and the necessary infrastructure. The practical limit in the West is approx. 3 trains per hour per track for termini and slightly above 4 trains per hour per track for through stations. But yes, even with less than ideal approaches a capacity improvement should be doable.

    • Alon Levy

      Go to the RER A at Chatelet-Les Halles. That’s 30 trains per track per direction at the peak.

      For terminal capacity, the RER B turns 8 tph at Terminal 2 of Charles de Gaulle, on two tracks. Though, the next station on the line, Terminal 1, is four-tracked, making it possible to rearrange trains if necessary (and, bear in mind, RER timetable adherence is about on a par with Metro-North’s, i.e. around 90% by a 5-minute standard).

      • MobilMan

        Ah yes, I forgot to mention: capacity for intercity and regional trains.
        RER, S-Bahn and assorted can of course be operated at subway frequencies. I assume MBTA operates regional trains that are not S-Bahn-like (?).

        • Alon Levy

          There isn’t much about the MBTA that’s inherently non-S-Bahn-like, as far as turnaround times are concerned. Remember, fast turnarounds happen on American commuter rail every day, just at the outbound terminal. While two-track terminal throughput higher than 20 tph requires perfectly interchangeable trains or perfect schedule discipline (or both, in the case of the Chuo Line), 8-12 tph is doable even with poor punctuality.

          Intercity trains usually turn much more slowly, yeah. The reason is that they offer a higher quality of service, and make longer trips each time, which means that they usually need to be cleaned every time they arrive at a terminal. In other words, regional commuter traffic should be counted together with S-Bahns and subways, rather than with intercity traffic.

          • anonymouse

            For what it’s worth, the standard timetabled turn time at outer terminals on MBTA is 10 minutes, so you can count on at least 4 tph per platform at BOS, maybe 5 or even 6. I was told by an Amtrak worker that they could do an engine change in 8 minutes if they really wanted to, and a change of direction should take less, especially if they’re just taking the train to the yard.

  11. Pingback: MBTA-HSR Compatibility | Pedestrian Observations
  12. Pingback: The Urbanophile » Blog Archive » Bad US Rail Practices and What It Means for FRA Regulations by Alon Levy
  13. Pingback: Improving the MBTA: Electronics and Concrete | Pedestrian Observations

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.