I guess I’m a transportation blogger now, which was certainly not my intent, but here we are. Last Monday, September 17, I took the day off from work and went to the joint MassDOT and MBTA board meeting at the State Transportation Building in Boston. I wrote up a bunch of notes from that meeting, which I tweeted out Monday night, and I thought that was it, but I spent a lot of my free time this past week reading the documents and notes from prior meetings that I hadn’t attended, and ended up thinking of even more stuff to say. For reference, here’s my crazy long thread about the September 17 meeting:
Garrett Wollman (@garrett_wollman) September 18, 2018
One of the documents I looked at after the fact was the “commuter rail vision” presentation, and I also went through a slide deck that was presented at a previous commuter-rail meeting. As with the “Focus40” twenty-year investment plan, the “vision” so far is very, well, “visiony”, with lots of generalities and not much in the way of useful information. It does introduce a taxonomy of possible service improvements, which is something to at least think about, and it actually got me to look a bit more at commuter rail schedules for the Framingham/Worcester line, which I’ll get to in a moment. One particularly notable take-away from the board meeting was that the Central Transportation Planning Staff, the group at MassDOT that is actually responsible for modeling transportation demand and transit ridership, has been tasked with collecting updated commuter-rail ridership data. If you’ve followed my previous posts on this subject, you’ll note that I lamented the fact that the available data is from 2012, before a number of important service changes, and anecdotal evidence is that there have been substantial changes in ridership on some routes including Worcester. Another take-away was that the projected cost of South Coast Rail is so awful, even the Federal Transit Administration’s “New Starts” program won’t touch it. (This really suggests that Phase 1 should just be canceled outright, and Phase 2 should be implemented as a part of Regional Rail after the Stoughton Line is converted.) I can think of a very large number of things I’d rather the Commonwealth spent a billion of our dollars on before SCR Phase 1, which hurts more people on the Middleboro/Lakeville Line than will ever ride it from New Bedford and Fall River. Take a little more time and do it right (and work on those ridiculous inflated costs while you’re at it).
You could get more transportation value out of an east-west intercity route connecting Springfield for $1b than SCR.—
Garrett Wollman (@garrett_wollman) September 18, 2018
And there you have the next thing I started thinking about: once you are doing reasonable frequencies with Regional Rail on the Framingham/Worcester Line, how exactly do you slot in intercity rail without disrupting the nice clockface headways? The Lake Shore Limited is bad enough, but at least that’s only one train daily in each direction; if we really want to connect Boston to Springfield, never mind Pittsfield, that’s something that we’ll have to figure out, because the right of way isn’t wide enough for a passing track in a lot of places. If the line to Springfield is electrified, then maybe you can do like the Finnish network does with Helsinki–Tampere trains, where they are just integrated into the commuter rail schedule as express or super-express trains on the route that they would follow anyway, with full fare integration at stops within the Helsinki commuter zone. (Of course, VR does have plenty of tracks heading into Helsinki to run a variety of express services, so it’s not an exact comparison.)
Then on Wednesday, the Democratic candidate for governor of Massachusetts, Jay Gonzalez, who I have supported, released a very unfortunate proposal to tax university endowments to generate more revenue for transportation. Yay for the admission that our transportation system needs more funding, but taxing what is our largest and most globally competitive industry, higher education? No thanks. This isn’t bad enough to make me reconsider my vote (Gonzalez is still better than Baker, and this plan has zero chance of making it through the General Court) but it’s still a real disappointment. (If the polls are correct, Baker is going to win in a landslide anyway. Maybe it will be OK if he cancels South Coast Rail and we can spend the $1b on something worthwhile, although I think it more likely he’ll cancel SCR and refuse to spend the money on things we actually need.)
If we had more rail service heading into Boston, especially on Framingham/Worcester, these could be installed all t… twitter.com/i/web/status/1…—
Zachary Agush (@zagush) September 19, 2018
In response to Zach’s tweet, I noted that one problem with doing this in Massachusetts is that the Turnpike really only follows the Boston & Albany right-of-way in Newton and Boston; the railroad turns sharply to the southwest after crossing the Charles, where the Turnpike runs more westerly, and from that point they diverge until Westborough — but there’s no exit in Westborough or Grafton near enough to the stations in those towns to make such a thing work, and then the B&A turns north to get around Lake Quinsigamond and enter downtown Worcester from the northeast. But, if you double-tracked the Agricultural Branch to Framingham Technology Park, and ran four trains per hour to a new station and layover facility there, you could capture traffic right off the Turnpike at exit 12 (the Tech Park is practically at the exit, and there’s already adequate highway capacity on Route 9 because of the employment density there). In the same conversation I also noted that there ought to be an infill station at Newton Corner, which is also directly accessible from the Turnpike, but because of traffic bottlenecks there would not be useful for traffic diversion. (On the other hand, it could replace the express buses that currently serve Newton Corner!)
Finally, I ate something I shouldn’t have and couldn’t get to sleep on Thursday night, so I used the time to make some more changes to my R script for modeling boardings on the Framingham/Worcester commuter rail (GitHub repo). As I noted above, CTPS hasn’t published the 2018 ridership data (I suspect they’re still collecting the data, since September represents some significant startup transients due to the school year starting, and it’s a manual count due to the antiquated fare collection system on the commuter rail network). The major change this time around was to simulate a “zone express” structure (to use the “rail vision” terminology) under various scheduling scenarios and see if it makes any difference. (Well, obviously it makes a different to the Worcester riders if they’re arriving eight minutes faster, but I’m interested particularly in cost differences.) I discovered, somewhat to my surprise, that it actually does make a difference: all of the scenarios that I had simulated previously required at least 26 EMU trainsets, and I was able to find a schedule that required only 24 trainsets. (If you build my Ag Branch proposal, you end up needing those two trainsets back, and there are a few other caveats that I’ll mention below. Of course with full Regional Rail, you’re running identical equipment on all 12 lines, so with a small reserve of extra trainsets you can slosh around equipment as demand warrants.) You can download PDF of the revised spreadsheet with full details, including inbound and (for the first time) outbound schedules and an equipment plan.
As before, this simulation assumes a full Regional Rail implementation, but not North-South Rail Link, and train capacity is based on the 75-meter configuration of the Stadler FLIRT delivered to Helsinki’s commuter rail system in 2015 (which cost then about EUR 7m each for 75 trainsets). These articulated EMU trainsets hold 232 passengers in fixed seating, but I’ve assumed that 300-passenger loads are acceptable for short distances when you add standees; as I noted in a previous post, TEXRail is buying a diesel version of this train with 230 seats and they are advertising a passenger capacity (including standees) of 447. In my simulation, there are five “local” trains (making all stops from Framingham to Boston during AM rush) that have more than 232 passengers, but in all cases every one who gets on in Natick and Wellesley gets a seat, so the standees are exclusively Newton passengers who will either get a seat or get off within 12 minutes, and in my view this is a reasonable compromise to keep the equipment utilization down.
(It would be easy enough to make these trains use double-trainset consists, but that would require at least four more trainsets. Alternatively, you could schedule extra inbound trains from Riverside Junction, but this would require construction of a new turnaround facility and additional platform space at South Station, and you would have to add the same four trainsets anyway, and the operating costs would be higher because you’d have to pay operators in addition to conductors. I looked at this idea briefly but concluded it was a bad one, and I oppose the “urban rail” service on the “rail vision” menu — you really want the extra service all the way to Framingham, and you want your layover facility at Framingham, not Riverside.)
The other assumptions I’m making are that you can build a second platform at Worcester (this is in planning, and my schedule absolutely requires a second platform for at least one inbound/outbound conflict), and that you can (re-)build a layover facility at the Fountain St. yard in Framingham. Finally, I assume that a 12-minute turnaround time is sufficient for almost all turns, and furthermore, one turn at South Station can be done in 10 minutes (this is easily fixable if you let the incoming train occupy a third platform, avoiding an inbound/outbound conflict), and one turn at Framingham can be done in 7 minutes (to fix this requires another trainset, although by the time it’s needed, there are already trainsets in midday storage so you don’t need more capital investment, just another hour of train crew, but it’s fairly late morning where a few minutes of lateness doesn’t ripple through the schedule). One final issue is that there is a late-rush (8:45) reverse-commute train to Worcester that goes out of service on arrival: I didn’t want to take it out of service at Boston or Framingham because part of the promise of Regional Rail is better reverse-commute service, but there’s no place to stick it in Worcester, so unless you can find 150 meters (~500 feet) of siding west of Union Station, you have to turn the train and deadhead to Framingham.
The way I’ve structured this, all Framingham-to-Boston trains are single-trainset consists (250 ft), and all Worcester-to-Boston trains are two-trainset consists (500 ft). As a consequence, all Boston-to-Worcester trains also have two trainsets, as do some Boston-to-Framingham trains which are turns of inbound trains from Worcester. However, if you’re willing to split consists on the Worcester-bound platform at South Statuin, you can send a single trainset (and save the variable costs of the second trainset) back to Worcester, and then when the next inbound train arrives with two more trainsets, send them both (coupled) to Southampton St. for inspection and maintenance, or just storage. The caveat, of course, is that you have to reverse the process to put them back into service at midday. Since this is more complicated I didn’t write it into the equipment plan.
This service takes only 24 trainsets, and requires only a small layover yard at Fountain St. in Framingham (capacity 7 trainsets or 1750 feet, 9/2250 if you can’t store that one Worcester train), and sends eight trains per hour at the peak through Framingham, West Natick, and the Boston stations; it’s my current favored plan. Now time to go to bed, because there’s another FMCB (MBTA Fiscal & Management Control Board) meeting on Monday.
UPDATE 2018-09-25: I didn’t end up going to the FMCB meeting, after looking at the agenda and deciding it was unlikely to be very interesting. I did, however, think a bit more about the equipment plan, and came up with a revised schedule that cuts back more of the early-morning service, deleting some inbound runs from Worcester entirely (the trains shown in the simulation as arriving with zero passengers at 5:08, 5:23, 5:38, and 6:08); this still leaves one early-morning train from Worcester. In order to start the service, I return to having some trainsets stored overnight at Southampton St. yard, and these run in revenue service to Worcester. Also, one two-set consist which I had previously sent to Framingham for midday layover I instead keep at Southampton St.; different equipment could be sent each day, and the layover time used for inspections and light maintenance. The overall plan has 13 trainsets in all-day service (three singles for the Framingham service and 5 doubles for the Worcester service (although as noted above the doubles could be decoupled after peak to reduce the mileage on those trains, so long as you have a midday layover facility close to South Station); that makes a base-to-peak ratio of 54%. Overnight storage is then 6 units at Southampton St., 6 units at Fountain St., and the usual 12 units at Worcester; midday layover requirements are 7 units at Fountain St., 2 units somewhere in Worcester, and 2 units at Southampton for inspection and maintenance.
As for the East-West Intercity service, I spent a bit of time thinking about the schedule. Amtrak’s Lake Shore Limited takes an hour and a quarter between Springfield and Worcester, and if you look at the circuitous route you can understand why it’s so slow: the famous Worcester Hills mean that there just aren’t any direct routes at grades compatible with 19th century railroad engineering, and since 1900 we’ve seen only abandonments anyway. To get a meaningfully better service between Springfield and Boston you’d need to straighten the route (the Turnpike follows a much more direct path, but runs about 10 miles south of downtown Worcester), but there are probably some upgrades (passing tracks, superelevation, even electrification) that could reduce the time to an hour even with station stops. For now, I’m suggesting an hourly DMU service with timed transfers at Worcester to a regional express train at peak periods, and there’s enough track time off-peak to run straight through to Boston. That would require 7 or 8 DMUs, but they could be ordered from the same vendor and family as the Regional Rail EMUs for economies of scale and commonality of maintenance procedures. Double that order and build the maintenance base at Worcester, rather than Springfield, and you have enough equipment to run a north-south DMU commuter service serving Spencer, Holden, Webster, and Millbury, even down the Blackstone Valley into Providence if there’s actually any demand. Add another six or eight for eventual service to Pittsfield, which is also an hour and a quarter from Springfield. (All told, the rolling stock purchase for the combined East-West Intercity, Worcester North-South commuter line, and Regional Rail looks like 20 FLIRT3 DMUs for $250m and 300 EMUs for $2.4bn, delivered over about a decade as the lines were upgraded. For the East-West line, which serves longer trips, you probably want a slightly different fitment with more comfortable seating and a vending area at one end, which you don’t need in commuter service.)
Another note about Zach Agush’s tweet, seen above: while the Framingham/Worcester Line is not a good candidate for integration with GoTime, there are a couple of stations that would be fantastic candidates: Anderson RTC in Woburn (from both I-93 and I-95, but especially from I-93 where an exit was built to directly serve the station), Route 128 in Westwood (right off I-95 and close to the south end of I-93), and Forge Park/495 in Franklin off I-495. All of these stations are close to freeway exits, and in addition to travel times, the signs could show the number of parking spaces remaining at the station. (I think Littleton/495 and South Acton would meet this criterion as well, for Route 2 travelers.) Other edits to Sunday’s post for clarity.