Just about a year ago, I went to Worldcon 75 in Helsinki. Because Worldcon 76 in San Jose (which I’m not attending) is about to start, I was reminded that I still have a whole bunch more photos from Helsinki that I’ve never finished editing, and the research that I’ve done in the last few weeks has definitely given me a bit more to say about them (and Helsinki’s brownfields redevelopment and transit construction costs), so I’m going to try to push a final set of Helsinki posts out over the next few days, if I can get a few hours clear for photo editing. (I say “final” because I have no plans to go back, but if someone’s willing to subsidize my travel I would happily return next summer.)
Beyond that, I wanted to wind up my discussion of Regional Rail and the future of the commuter rail service between Worcester and Boston. In addition to the analysis in the previous two posts, I did a lot of modeling of service patterns that I haven’t shared, because they were not sufficiently interesting, or at least sufficiently different from what I’ve already presented. In some cases these models had insurmountable technical problems, mostly due to the lack of storage space in the right places, and in other cases in order to make the model work I had to assume an impractically short turnaround time. (Impractical under current FRA regulations and railroad operating practices, at any rate — railroads elsewhere in the world don’t seem to have a problem turning trains at rapid-transit-like frequencies.) In a few cases it’s not storage space but platform length that is the biggest issue, like the four-trains-per-hour service pattern that requires trains longer than the current MBTA construction standard. Here are a few of my major take-aways:
- The F/W Line is incredibly busy for a small number of AM peak runs and fairly easy to service at whatever headway you choose the rest of the time. I based my simulations on the 2012 CTPS riderahip audit, and there’s every reason to think that actual ridership (despite the well-publicized problems of 2015–2017) has increased since then.
- The lack of sufficient storage capacity in Worcester is a serious limitation on any service. In the Regional Rail model, additional storage capacity would be needed elsewhere at least for middays and probably at night.
- I believe that additional storage should be at Framingham, not Allston or Southampton. The reason it should be there is a simple one: equipment positioning from those other places has no transportation value, even if for some reason you made it “revenue”; scheduled early-morning and late-evening runs from/to Framingham would have substantial transportation value, and there are multiple existing yards at Framingham that either have room or could be expanded (or the freight railroad could be incented to consolidate operations).
- Extending passenger service via the Agricultural Branch to Framingham State University and Framingham Technology Park would have actual transportation value, but the costs and RoW limitations are such that it is unlikely to happen without capital support from MWRTA, the City of Framingham, FSU and/or the Tech Park tenants (especially Bose and Sanofi, the largest employers). Such a service might be operated as the tail end of Framingham-terminating Regional Rail trains, in which case there’s room in the Tech Park to construct a layover yard, or it might simply be a diesel shuttle from South Framingham that met each train at the existing station. (Either option would require substantial station construction, as there are no historic stations along this four-mile segment of the line.)
- None of the schedules I came up with can work with fewer than twenty-four 75 m (250-foot), 232-seat articulated EMUs; most of the usable ones require 26 or 27 (plus one more for a Tech Park shuttle, if you want to build that). That’s an equipment cost of about $216 million, in addition to the capital construction (which I’ll detail below).
- These schedules and car requirements would get substantially easier with a longer trainset. The MBTA’s standard platform length of 800 feet puts a substantial limitation on the number of passengers you can accommodate with a fully accessible, level-boarding, single-level design. Plus there are issues with platform length at Yawkey that I know about and possibly other stations. (And I have assumed that the storage facilities, which the T reports in “consists”, reflect this maximum length and not the actual lengths of trains being stored for the schedule current when the “State of Commuter Rail” report was written.) For 800-foot platforms, with three-trainset consists you’re limited to 80 m trainsets, which depending on the layout could get you another 20 seats, but if you’re willing to accept a maximum of two-trainset consists, you could go as long as 120 m — and then you have a different problem, because then you either have to manage two different EMU equipment types, or waste a lot of energy hauling around a 350-passenger train on the majority of runs that have fewer than 150 passengers. I’d probably spring for the 80 m length if there are no other constraints, just to have the capacity.
- Another option is to forget about running articulated EMUs and just do New York-style EMUs, which are the same length as regular passenger coaches. Depending on clearances, you might be able to run bilevel cars under wire, although this is probably a bad idea because of steps and fewer doors leading to slow boarding and alighting and taking away the dwell-time advantage of modern articulated EMU designs.
- My demand model is surprisingly sensitive to the exact timing of the new schedule being simulated, relative to the arrival times of the high-demand trains. There are three things going on here: first, the demand model considers only the arrival time to be important, not the departure time. Many commuters may have a deadline to arrive, hence the high demand trains, but also have nothing in particular to do with the time they would gain from a faster schedule, and so would choose a departure time closer to their current one rather than the latest train that gets in before their deadline. Second, there is no feedback in this model: in the real world, passengers’ expectations of comfort and reliability have an impact on what services they choose, and I don’t know, for example, what fraction of passengers would shift to an earlier or later train in order to be sure of getting a seat on, say, the half-hour trip from West Natick to South Station. Third, the demand model simplistically assumes that the passengers’ desired arrival times are uniformly distributed over the interval between trains (actually slightly offset by a few minutes to attempt to account for satisficing). To get a better demand model would require Actual Survey Research, which I’d be very interested in but am not prepared to fund out of my own pocket — I assume that CTPS has such a model already, and groups like TransitMatters should be regularly surveying commuters in order to justify their policy proposals to Beacon Hill.
- In particular, one of the scenarios I simulated for multiple schedules was short-turning trains at Framingham. This always makes the loading much worse, but exactly how much depends to a surprisingly great degree on the order of Framingham and Worcester trains, because so much of the model depends so specifically on the exact timing of the few arrivals provided by the 2012 schedule. Even in 2012 the MBTA was operating a local/express pattern during peak periods; this service pattern isn’t compatible with clockface headways because the time gained by the express over a local causes an express to catch up to the previous local. If you’re limited for other reasons to 15-minute headways (four trains per hour), then the local/express pattern could potentially work, but I haven’t simulated any expresses because I think 4 tph is a pretty weak-sauce service on this line, and there’s no practical place to turn a supplementary short-turn train short of Framingham without fouling the line. However, during middays there does not appear to be enough demand to run all trains to Worcester — even at Framingham the current schedule actually has a huge gap that makes it impractical for me personally — so you’d probably want to short-turn at least half of your trains at Framingham (another reason to put the layover facility there and not in Allston).
What is to be done?
So supposing we (the people of Massachusetts) decided that we really wanted to have a modern, reliable, frequent rail service between Boston and Worcester. What would we need to pay for to make this happen?
- Expansion of Amtrak’s Sharon substation, which powers both the Northeast Corridor and the “Boston terminal district”. The substation was built with the expectation that it might eventually be expanded for commuter service, and because the terminal district is chained off the NEC that additional power will be needed for electrifying the F/W Line. (For other reasons, of course, the capacity at Sharon should be maxed out — you also want to serve Stoughton/Providence and Fairmount trains, and probably at least part of Franklin.) I don’t know how much this costs, but you can search the web for informed estimates and the original Amtrak NEC electrification plans.) This also gives you sufficient capacity for electrification through Newton and Brighton.
- Construct a new substation in MetroWest. Exactly where depends on the location of sufficient high-voltage distribution capacity.
- Correct superelevation on the line to allow for increased speed limits. There’s a machine that does this without excavation or rail replacement.
- Replace obsolete signaling and switch machines. A lot of this is already programmed as a part of the existing state-of-good-repair program and will serve equally well for Regional Rail.
- Construct full high platforms for both tracks at all stations outside Boston. An upgrade for Natick Center (which is the busiest station on the whole system to not have any accessibility) to a high-level island platform is in process.
- Obviously, construct the actual electrification infrastructure. In order to maintain freight clearances under the wires, it may be necessary to depress the tracks or alter some overpasses in some locations. (Beacon Street, Boston, is a noted clearance issue: it would be fine for single-level cars but is too low for bi-levels, and so long as there are bi-levels anywhere on the south side system, the MBTA needs to be able to move them under that overpass to get to the Commuter Rail Maintenance Facility on the north side.)
- Currently, Worcester Union Station is limited to a single platform on the Main Line. (The other platforms at the station serve other railroads that do not currently have any passenger service, although a private operator had proposed to run Worcester-Providence service from one of them.) Currently this is under study with no firm funding or construction date, but in order to support reasonable schedules at Worcester with sufficient recovery and turnaround time for the train crews, this is a firm requirement.
- Most importantly, the MBTA has to decide to do it. There are several related planning processes doing on right at the moment: the “Focus 40” plan is the T’s long-term plan, and in order for anything worthwhile to be done before we’re all dead or retired it has to get prioritized in that plan. There is also a “commuter rail vision” process, about which there doesn’t seem to be much public information, and the people responsible for that will have to be pushed very hard to make a plan that’s actually forward-looking and not repeating the same defeatist, stuck-in-the-past thinking (which the MBTA is exceedingly good at). And there is a procurement process starting to ramp up now for the new private contractor to actually operate the MBTA Commuter Rail.
- Finally, we have to have a governor who actually uses transit and is responsive to the needs and aspirations of the communities in the MBTA district — who are responsible for the vast majority of the Commonwealth’s jobs and tax revenue.