Commuter rail schedules for the Plague Year

At a couple of recent MBTA board meetings, the authority’s staff have made it clear that they expect the passenger volume on MBTA commuter rail to remain depressed for significantly longer than that of the subway, light rail, and bus networks, as demand recovers from the COVID-19 pandemic and related business closures. These conversations have been happening in the context of the MBTA’s financial situation and the fiscal year 2021 budget, which the authority is about to approve at a special board meeting on Thursday, because reduced income from sales of expensive commuter-rail passes and parking fees has an outsized impact on the operating margins of the service (much of the operating cost of which is set by the T’s contract with Keolis, which has another two years to run). Much of the discussion centers around how the commuter-rail service is built for 9–5 office workers who live in the suburbs and drive to a station where they park all day, and whether some significant fraction of those workers will continue working from home for the rest of the year, or their employers will otherwise arrange for them to no longer commute into the urban core, especially considering that suburban homeowners are more likely to be older. But all of the commuter-rail lines serve cities where a significant number of residents are service workers, work in research or medicine, or otherwise have schedules that the current commuter-rail service fails to address.

Currently, the commuter rail is running on the “reduced service” schedule, which is a modified Saturday schedule, but in the governor’s “reopening plan” announced on Monday, it was suggested that the commuter rail might not resume its previous operating schedule — and a lot of people raised some eyebrows about that. I want to argue that this is a good thing, and that this crisis provides a tremendous opportunity to fix the damn schedules at a time when ridership is already reduced.

What do I mean by “fix the damn schedules”? Well, we could start by looking at what the MBTA board voted in favor of last fall: all-day, bidirectional service on clockface headways, as proposed in the Rail Vision report. (What’s a “clockface headway”? In essence, it means that the train schedule at every station is the same every hour of the day: a train that arrives at 6:07 also arrives at 7:07, 8:07, 9:07, and so on. Having a memorable schedule all day long is important to reduce the cognitive load and stress for riders, regardless of what Stephanie Pollack may think, and is also key to the “Swiss model” of scheduling, because you only need to make the schedule for a single hour and just repeat that across the entire day.) Rail Vision proposed 15-minute and 30-minute headways, which are today a substantial challenge due to long-standing facilities constraints, garbage rolling stock, the lack of electrification, and the lack of full-length high platforms at many stations. This all takes time and money to fix (although not as much of either as the MBTA’s management is trying to make it take) — but the question naturally arises, in this pandemic year: What could we do now, within the constraints of the facilities and equipment that we already have, given that demand is substantially reduced?

The answer, as it turns out, is quite a lot.

For starters, nearly all lines (Old Colony excepted) could run all-day bidirectional service on 60-minute headways using significantly less equipment than is required for “status quo ante” schedules. The five two-track main lines can run all-day bidirectional service on 30-minute headways, with careful scheduling of meets on a couple of lines that have single-track sections. These are all local trains — no expresses or other non-standard service patterns — so they’re easy to schedule uniformly; some expresses would be possible but I haven’t run a constraint solver to actually see which ones. (As a practical matter, the MBTA currently only runs expresses on the Providence, Worcester, and Fitchburg lines.) I’m making relatively few assumptions beyond the availability of track slots from Amtrak and Pan Am on the various routes they control dispatching for. Here’s a table showing my thinking:

Route Cycle time Consists required
Status quo ante 60′ headway 30′ headway
Fitchburg 240′ 6 4 8
Lowell (NHML) 120′ 5 2 4
Haverhill (Western Route) 180′ 5 3 x
Reading (Western Route) 90′ * *
Rockport (Eastern Route) 180′ 4 3 x
Newburyport (Eastern Route) 180′ 4 3 x
Beverly (Eastern Route) 120′ * o
Worcester 210′ 8 4 7
Framingham 150′ * *
Needham 120′ 3 2 *
Franklin 180′ 3 3 x
Providence 180′ 7 3 6
Wickford shuttle 90′ 2 *
Stoughton 120′ 2 2 *
Totals 47 31 43

* Not studied.
x Service precluded by infrastructure constraints.
o Main line of the Eastern Route receives 30-minute service as a consequence of 60-minute headways on the two branches, although they may not be evenly spaced due to the need for scheduling the single-track through Salem.

Cycle times were calculated by taking the current reduced schedule, which makes all local stops, and adding at least 15 minutes turnaround time at each end. It turns out that for some services, like the Haverhill and Fitchburg lines, single-track sections restrict the feasible schedules such that the turnaround time is much longer at one end of the route than the other. (I assumed for the Fitchburg line that westbound trains would hold at Fitchburg until the eastbound train clears the Wachusett crossover, about five minutes on the schedule, to ensure that as few passengers as possible are inconvenienced; adjusting the schedule earlier in the trip would make the meet in Waltham much chancier and delay far more passengers. Unfortunately, both Wachusett and Fitchburg are single-platform stations, even though the Fitchburg Line is double-tracked all the way from Waltham.) Finally, implementing 30-minute headways on the Worcester Line might be impossible with the current configuration of Worcester Union Station; in that case, most of the benefit can be had by simply running short-turns to Framingham (which wouldn’t reduce the equipment requirement — note that cycle times have to be padded upward to make a multiple of the headway, and ideally you would like the extra padding to be at the out-of-town end of the line, except that you then need layover space that won’t foul the platform tracks, which isn’t always available).

What does this give us? Well, obviously, it provides “less” service than the status quo ante as the MBTA would normally measure things. But it provides vastly more service at times when non-office workers are more likely to be traveling, and serves the state’s public-health goals by supporting a more spread-out, less peaky commuting pattern for everyone. And, of course, it completely eliminates the (false perception of a need for) midday layover facilities. On the other hand, it does potentially increase staffing costs; I have not studied this and don’t know what the actually required staffing levels are, particularly if trains can run with shorter consists (meaning fewer conductors per train, but balanced out by the overall increase in trains per day). I’ve completely ignored the Foxboro pilot service. I also haven’t seriously looked at scheduling optimizations that could be done, such as running all Franklin service via the Dorchester Branch or running all Haverhill service via the New Hampshire Main Line, or other low-cost interventions that might reduce cycle times and improve equipment utilization such as increasing speed limits or eliminating some very lightly used and inaccessible stops.

But if we did do this, then we would establish a different, in my view superior, baseline — call it a “new normal” — to which commuter rail passengers could grow accustomed and expect to see maintained or improved as passenger volumes return and we move forward with the “Rail Transformation”.

This entry was posted in Transportation and tagged , , , . Bookmark the permalink.