Green Line Reconfiguration

One transfer access for Everett/Chelsea that should not be overlooked is taking the Orange Line at Sullivan. In a scenario where Sullivan->Lechmere can't be built and thus my pattern (2) can't be run, I wonder if Everett/Chelsea riders would rather transfer to OL (or BL from Chelsea) than going all the way to Kenmore and backtracking into the Central Subway on (1).

Probably depends on where they're going. The Sullivan transfer would be useful to anyone going downtown (or, obviously, past the core on the Orange Line's route itself), but annoying if it meant a second transfer at North Station (i.e. anyone going somewhere on Green). Depending on GJ station siting, there could well be a Red transfer at Kendall as well. I'd be interesting to see whether the projections would favor the fastest ride (i.e. the transfer) or the most convenient (one-seat via GJ/Kenmore even though it's longer).
 
Having spent way too much time down this particular rabbit hole (upthread), I'm gonna try to limit my current contribution (emphasis on "try", we'll see how succinct I can be...)

Loops

The thing to keep in mind about any sort of looping service on the Grand Junction is that it rapidly becomes inefficient once you start to close the loop. The inefficiency isn't necessarily a problem for the Grand Junction service itself, but it becomes a problem when it starts taking up slots at Park St. Even at reduced headways, a combo of Harvard, Commonwealth, Beacon, Nubian, and Seaport -- to say nothing of at least one of Needham, Riverside, and/or Huntington -- requires capacity increases at Park St. Once increased capacity is available, there will still be competition for which destinations get OSRs, and it's hard to imagine a circuit service beating out any of the alternatives.

Northside capacity is greater, so if you wanted to do a wraparound service that ran GC-Lechmere-Grand Junction-Kenmore (with some sort of creative solution to loop at Kenmore, that could potentially be feasible, but through-running seems hard to justify.

Chelsea and Everett

I mentioned this somewhere else recently I think -- maybe the GLX thread or Crazy Transit Pitches -- but I don't think a Green Line branch that runs from the Central Subway to Sullivan to Chelsea makes sense. It's circuitous, and still misses the densest parts of Chelsea. I think the stronger solution for Chelsea is threefold:
  • BRT-ifying the 111 hard (I'm talking dedicated lanes on the Tobin, high platforms, pre-payment, priority signaling)
  • Circumferential LRT/BRT service to Airport and Sullivan/Cambridge
  • RER service via the NSRL to Back Bay and Ruggles/Lansdowne with limited stops
The RER's higher speed can offset the circuitous path better than the Green Line can.

Everett is potentially different -- but a major question for that calculation will be how far into Everett you can bring the one-seat-ride; if it's just Sweetser Circle, that'll be much less compelling than managing an extension all the way to Ferry St, which I think would be worth an OSR into the Central Subway.

Ah to heck with it... I recently gamed out my latest thinking on potential full-build service patterns. Let me see if I can dig that out and polish enough for primetime.
 
Northside capacity is greater, so if you wanted to do a wraparound service that ran GC-Lechmere-Grand Junction-Kenmore (with some sort of creative solution to loop at Kenmore, that could potentially be feasible, but through-running seems hard to justify.

I mentioned this somewhere else recently I think -- maybe the GLX thread or Crazy Transit Pitches -- but I don't think a Green Line branch that runs from the Central Subway to Sullivan to Chelsea makes sense. It's circuitous, and still misses the densest parts of Chelsea. I think the stronger solution for Chelsea is threefold:
  • BRT-ifying the 111 hard (I'm talking dedicated lanes on the Tobin, high platforms, pre-payment, priority signaling)
  • Circumferential LRT/BRT service to Airport and Sullivan/Cambridge
  • RER service via the NSRL to Back Bay and Ruggles/Lansdowne with limited stops
The RER's higher speed can offset the circuitous path better than the Green Line can.

Everett is potentially different -- but a major question for that calculation will be how far into Everett you can bring the one-seat-ride; if it's just Sweetser Circle, that'll be much less compelling than managing an extension all the way to Ferry St, which I think would be worth an OSR into the Central Subway.

It's worth noting that having the ability to run, for instance, Lechmere->GJ->Kenmore->Park and/or Chelsea/Everett->Sullivan->Lechmere->Park doesn't mean that all cars have to run those specific service patterns. If there's capacity issues as a concern in the Central Subway (even with the Brattle Loop for services fed from the north), some of that can be mitigated by intelligent route planning. Sending everything coming off the GJ or from Sullivan to Park (or even GC) might well not make sense, but say, sending one-in-two or one-in-three might be more feasible; it'd allow direct OSR interlining between Cambridge and Chelsea/Everett (and, ideally, Airport as the end of the line for a nice alternative to transferring downtown) while also providing OSRs at a more manageable capacity into the Central Subway. (Basically, the point is, build the connections and loops that maximize flexibility and figure out the service plans in practice later, instead of leaving us with a half-baked system because of target fixation on certain services or ultimately-out-of-scope capacity constraints elsewhere).
 
Ive assumed that Seaport and/or Nubian would either turn around at Park(never using the inner track) or that B would become one/some of both of them and vice-versa
 
Ive assumed that Seaport and/or Nubian would either turn around at Park(never using the inner track) or that B would become one/some of both of them and vice-versa

Can't turn anything coming from the Tremont lower tunnels without touching the inner track at Park Street unless the long-gone outer loop is rebuilt. I've occasionally wondered if that might be preferable just for the gains of eliminating the weaving you'd need to do to turn Nubian/Seaport (and/or D/E via rerouted Huntington), but that'd require a study or at least a solid picture of the operational conditions and capacity impacts to determine.

B itself (or anything from Kenmore) can't serve Nubian or Seaport (well, not without the agony and expense of trying to tunnel down Essex). Cars certainly could run Kenmore->Park Loop->Tremont->Wherever, but that probably wouldn't be a preferred service pattern (though more quasi-run-as-directeds or overlapping/complementary service patterns wouldn't be bad), and it wouldn't be practically any different than just turning Nubian/Seaport at Park (at least from a passenger perspective).
 
Can't turn anything coming from the Tremont lower tunnels without touching the inner track at Park Street unless the long-gone outer loop is rebuilt. I've occasionally wondered if that might be preferable just for the gains of eliminating the weaving you'd need to do to turn Nubian/Seaport (and/or D/E via rerouted Huntington), but that'd require a study or at least a solid picture of the operational conditions and capacity impacts to determine.

B itself (or anything from Kenmore) can't serve Nubian or Seaport (well, not without the agony and expense of trying to tunnel down Essex). Cars certainly could run Kenmore->Park Loop->Tremont->Wherever, but that probably wouldn't be a preferred service pattern (though more quasi-run-as-directeds or overlapping/complementary service patterns wouldn't be bad), and it wouldn't be practically any different than just turning Nubian/Seaport at Park (at least from a passenger perspective).
Yard location is also worth considering. Seaport and Nubian are unlikely to ever have their own yards, so running them through Lechmere starts to look appealing.
 
Can't turn anything coming from the Tremont lower tunnels without touching the inner track at Park Street unless the long-gone outer loop is rebuilt. I've occasionally wondered if that might be preferable just for the gains of eliminating the weaving you'd need to do to turn Nubian/Seaport (and/or D/E via rerouted Huntington), but that'd require a study or at least a solid picture of the operational conditions and capacity impacts to determine.

B itself (or anything from Kenmore) can't serve Nubian or Seaport (well, not without the agony and expense of trying to tunnel down Essex). Cars certainly could run Kenmore->Park Loop->Tremont->Wherever, but that probably wouldn't be a preferred service pattern (though more quasi-run-as-directeds or overlapping/complementary service patterns wouldn't be bad), and it wouldn't be practically any different than just turning Nubian/Seaport at Park (at least from a passenger perspective).
Unclear why anything from Kenmore cant turn at Park but switch to the outer track leaving the station. Restoring outer loop capability should not be that tough
 
Unclear why anything from Kenmore cant turn at Park but switch to the outer track leaving the station. Restoring outer loop capability should not be that tough

Oh, if you restored the outer loop it would be possible to run Tremont->Park->Tremont without touching either fence track or run Tremont->Park->Kenmore using only the inner track from as wall-track westbounds do now, switching before Boylston. My point was that anything coming from Tremont (and Nubian and Seaport would be via Tremont, there's no meaningful path for them to reach Kenmore even if that was desirable) currently cannot loop at Park without using the inner track (because that's the only operable loop).

Technically I don't think it's possible to get from the current Park loop to the outer (wall) track at any point, so at the moment anything coming from Kenmore or Tremont can't turn and go out to Tremont, so a "B" Kenmore->Park->Nubian/Seaport would require one or both of a crossover from the inner westbound fence track to the wall track (probably not hard) and/or reactivation of the outer loop (probably harder, but might be more valuable to allow Tremont->Park->Tremont looping without weaving).

Still not really sure why we'd be specifically talking about a Kenmore->Park Loop->Tremont (Nubian/Seaport) service pattern. I could more easily see cars running that at times (when needed) than a regular pattern with passengers over the loop, but that's second-order consideration to fixing up the infrastructure to allow it in the first place.
 
Here's my quick stab at service patterns:

Harvard-North Sta [Lechmere/Reservoir] This should really be cut back to Kenmore, but that's not possible. Given it'll be the best performing of the 3 lines running through Kenmore, I ran it the farthest, but could be convinced other lines are better choices for that.
Boston College-Park Street [Lake]
Cleveland Circle-Park Street [Reservoir] Cut back to free up Reservoir cars for other services.
Riverside-Kenmore [Riverside] Assumes Indigo to Riverside.
Needham-Seaport via Huntington [Riverside/Needham] Maintains Needham OSR to South Station.
Hyde-Government Center [Riverside] Extension mitigates 39->T39 change
Nubian-Medford [Lechmere]
Seaport-Union [Riverside] Faux North-South Connector. I think Seaport would warrant two lines, but might be wrong.
Sullivan-West via GJ [Lechmere] Sullivan has OSR to North Sta, Downtown, and Back Bay, with half a cross-platform transfer at North Station.
West-Government Center via GJ (Brattle Loop) [Lechmere/Reservoir]


This pretty clearly requires expanding Lechmere or Riverside yards. Lake Street and Reservoir are are big as they will ever get and maxed out already. Needham you could maybe put a small Forest Hills/Caddigan-style yard in. Riverside has provisions to expand, but it's a very long way from anywhere it might be useful, except to obviously feed Needham and maybe take over the Lake Street supplement from Reservoir, and there is constant discussion of ToD here too, not to mention how to squeeze in Indigo.

If you wanted to do another extension, you pretty much have to build another yard. This makes Everett, Chelsea, Watertown, Arborway, and Mattapan look the most appealing.
 
Last edited:
Riverside-Kenmore [Riverside] Assumes Indigo to Riverside.

Running a permanent branch only to Kenmore is unlikely to go over well. I can see a case for overlapping service patterns such that some cars turn at Kenmore (and some at Reservoir in the other direction for that matter), but it's kind of cutting the legs out from under if it's a forced transfer. (Mitigated if Blue reaches Kenmore via Riverbank, though.)

Needham-Seaport via Huntington [Riverside/Needham] Maintains Needham OSR to South Station.

Similar to the last, I think it'd be better off with a mix of Needham->Seaport via Bay Village and Needham->Park (or points north); South Station as an OSR's less useful than hitting Red, Orange, and Blue if you reach GC, and a built-in forced transfer's not the best at attracting passengers. (How much of the mix is weighted towards Park or towards Seaport would be answered by demand patterns on the one hand and capacity constraints on the other.)

Nubian-Medford [Lechmere]

Unless Washington Street is getting nuked to become a trolley reservation, this seems infeasible. That much street running or quasi street running on the south end would make me nervous about having that be the big Medford feed line.
 
Here's my quick stab at service patterns:

Harvard-North Sta [Lechmere/Reservoir] This should really be cut back to Kenmore, but that's not possible. Given it'll be the best performing of the 3 lines running through Kenmore, I ran it the farthest, but could be convinced other lines are better choices for that.
Boston College-Park Street [Lake]
Cleveland Circle-Park Street [Reservoir] Cut back to free up Reservoir cars for other services.
Riverside-Kenmore [Riverside] Assumes Indigo to Riverside.
Needham-Seaport via Huntington [Riverside/Needham] Maintains Needham OSR to South Station.
Hyde-Government Center [Riverside] Extension mitigates 39->T39 change
Nubian-Medford [Lechmere]
Seaport-Union [Riverside] Faux North-South Connector. I think Seaport would warrant two lines, but might be wrong.
Sullivan-West via GJ [Lechmere] Sullivan has OSR to North Sta, Downtown, and Back Bay, with half a cross-platform transfer at North Station.
West-Government Center via GJ (Brattle Loop) [Lechmere/Reservoir]


This pretty clearly requires expanding Lechmere or Riverside yards. Lake Street and Reservoir are are big as they will ever get and maxed out already. Needham you could maybe put a small Forest Hills/Caddigan-style yard in. Riverside has provisions to expand, but it's a very long way from anywhere it might be useful, except to obviously feed Needham and maybe take over the Lake Street supplement from Reservoir, and there is constant discussion of ToD here too, not to mention how to squeeze in Indigo.

If you wanted to do another extension, you pretty much have to build another yard. This makes Everett, Chelsea, Watertown, Arborway, and Mattapan look the most appealing.
Looks like Nubian-Union Square and Seaport-Medford are more feasible. Possibly add some Seaport-Union Square or Seaport-Sullivan-Chelsea trains. Such arrangements will take away the Somerville-Longwood OSR, but I don't think there's a good way around it.

This will change with an extension to Porter and even further west.
 
Does anyone know if there is an upper "safe" speed for a vehicle with rapid transit profile wheels vs FRA?
 
Does anyone know if there is an upper "safe" speed for a vehicle with rapid transit profile wheels vs FRA?
Nothing that the FTA or FRA is likely to allow in revenue service. For non-revenue tows...probably 10 MPH. For articulated Green Line trolleys (esp. Type 8)...probably no truly safe speed.
 
Nothing that the FTA or FRA is likely to allow in revenue service. For non-revenue tows...probably 10 MPH. For articulated Green Line trolleys (esp. Type 8)...probably no truly safe speed.
Im sorry, I meant for instance, how fast a train could you run on the Red Line
 
Im sorry, I meant for instance, how fast a train could you run on the Red Line
As in...run traditional RR rolling stock like an EMU on Red Line trackage?

Probably 10 MPH. But I'm not sure the feds would allow it to carry passengers.
 
Im sorry, I meant for instance, how fast a train could you run on the Red Line
Oh...you mean, like a "high-speed" Red Line.

The wheel profile really wouldn't matter, but third-rail friction does set a practical upper limit of about 90-110 MPH on the equipment before you're just grinding the 3d rail shoes to cinders. You'd need really beefily-built cars to do that kind of propulsion, though...so probably something you wouldn't bother custom-designing with HRT rolling stock.
 
Alright, well this is gonna be a hot mess, but I think the math adds up. This is by no means intended to be definitive or final in any way, but is an example of what a full-build LRT/BRT network could look like. I use a new nomenclature system just for the sake of simplicity on this post. And yes -- the colors probably would be user-hostile enough not to deploy IRL, but they are useful for our purposes here, particularly since they reinforce how the different "subnetworks" interrelate.

Line​
Terminal​
via​
Terminal​
tph​
LRT Lines
Green (GR1)​
Riverside or Needham​
Back Bay & Park​
Medford​
10​
Each southern terminal get 5 tph​
Green (GR1x)​
Riverside or Needham​
Brookline Village​
Back Bay or Kenmore​
10​
part-time​
Green (GR2)​
Nubian​
Park & Grand Junction​
Allston​
10​
Green (GR3)​
Seaport​
Park​
Porter & points west​
10​
Emerald (EM1)​
Harvard​
Kenmore​
Park​
10​
Emerald (EM2)​
Oak Sq​
Kenmore​
Park​
5​
supplemented by SL57​
Emerald (EM3)​
Boston College​
Kenmore​
Park​
7.5​
Emerald (EM4)​
Cleveland Circle​
Kenmore​
Park​
7.5​
Emerald (EM4x)​
Cleveland Circle​
Beacon St​
Kenmore​
2.5​
part-time​
Emerald (EM5)​
Heath, Hyde, or Jackson​
Brookline Village​
Kenmore​
10​
Aqua (AQ1)​
Seaport​
Back Bay​
Heath, Hyde, or Jackson​
5​
Aqua (AQ2)​
Seaport​
Back Bay​
Boston College​
5​
Gold (GO1)​
Chelsea & Airport​
Sullivan & Grand Junction​
Kenmore​
10​
Gold (GO2)​
Chelsea & Airport​
Sullivan & Union Sq​
Porter & points west​
10​
I like Watertown/Newton Corner
for this service,
but it can be flexible​
BRT Lines
Navy (NV1)​
Nubian​
Post Office Sq & Tobin Bridge​
Chelsea​
10​
Surface BRT
(all destination pairs are
highly provisional)​
Navy (NV2)​
City Point​
Post Office Sq & Rutherford Ave​
Everett​
10​
Surface BRT​
Navy (NV3)​
Airport Terminals​
Post Office Sq & Summer St​
Charlestown​
10​
Surface BRT​
Silver (SL57)​
Newton Corner or Watertown Yard​
Oak Sq​
Kenmore​
5​
57A gets converted to LRT
and extended to Park;
57 is upgraded to full BRT​
Silver (SL39​
Forest Hills​
Huntington​
Copley​
10​
upgraded 39​
Silver (SL3)​
South Station​
Seaport & Airport Station​
Chelsea​
10​
Silver
Nubian, Kenmore, Ruggles​
Dorchester​
Mattapan, Ashmont​
Full-build BRT network​
Silver (SL32)​
Forest Hills​
Hyde Park​
Readville​
Diamond
Kenmore​
LMA​
Nubian​
10​
honestly this probably won't be one service
but a constellation of services,
but either way, this is covered​
Diamond (DM1)​
Harvard​
Boston Medical Center​
Nubian​
10​
upgraded 1​
Diamond
Nubian​
South Boston Haul Road and Airport Station​
Chelsea and points west​
distinct from the T12 proposal,
this service is limited stop
and is meant to provide a fast connection
to the Seaport and East Boston from Nubian​
Diamond
Chelsea and points east​
Wellington​
Davis​

And a terrifying diagram (which I guess I'll put in the next post because the forum software seems to be mad at me), meant to illustrate primarily what the cumulative frequencies look like on key stretches. Figures in parentheses add in the buses per hour along that stretch: for example, Washington St sees 10 tph to Park and beyond, and 10 buses per hour to Post Office Square and beyond, for a total of 20 vph and 3-min peak headways.

Key stretches
  • Huntington Ave to Back Bay: 20 tph that can increase to 30 tph at peak, plus another 10 bph on surface
  • Washington St: 20 vph
  • South Station to Seaport: 20 tph, 30 vph total
  • Grand Junction: 20 tph
  • Union Square to Porter: 20 tph
  • Sullivan to Chelsea to Airport: 20 tph, plus a number of BRT layer-ons at various points
  • Beacon maintains current frequencies, though with a (limited) use of short-turn service
  • Commonwealth sees a slight reduction in frequencies on some stretches, but is buttressed by increased freqs east of Packard's Corner, sees increased service just to the north on the old A Line route, and enjoys at its western end a new direct link to LMA and the Seaport
Honestly some of the cumulative frequencies are higher than likely needed -- this exercise was to demonstrate what could be possible. (Also I used multiples of 5 because the math was easier.) But for instance it's not clear to me that Grand Junction needs a train every 3 minutes; on the other hand, I don't necessarily want riders waiting that much longer than 6 minutes if they are looking for a specific branch. So it's a balancing act.

Comments and diagram in next post.
 
Unholy diagram:

LRT BRT Buildout June 2022.png


Comments:

Grand Junction

The eureka moment for me was realizing that there really are two corridors within the Grand Junction: a north-south corridor between Cambridge and Sullivan, and an east-west corridor between Downtown, Cambridge, and Allston. Hence why there are two services, both vaguely S-shaped, and layered on top of each other like this: >---<

So, services from Lechmere should head to Allston, and services from Sullivan should head to Kenmore.

Central Subway capacity

Soooooo this should in theory actually require not that much increase. The Green Lines stay on the outer tracks from Boylston through Park and beyond; Boylston-GC pre-pandemic saw 30 tph, so this could probably be doable. The Emerald Lines would stay on the inner tracks, which currently see 40 tph, until the interlockings between Boylston and Park; the question then becomes whether the Inner Loop can actually handle 30 tph shoveling through it.

That is where we'd probably need to lean on some additional "relief valves", including a reconstructed Outer Loop and potentially running some services through to loop at Government Center, or rerouting more Huntington Line services out to the Seaport -- the operational reality may be a bit less tidy than the idealized version discussed here.

Kenmore

I think you can swing a small LRT/BRT surface terminal, with a modestly expanded footprint. There's a lot of surface space at Kenmore Square that is given over to automobiles; building a true intermodal transfer center seems like a good reason to reclaim some of that space. If you truly cannot, then it's not the end of the world for the GO1 to terminate at West Station instead.

Riverside and Needham

Boy howdy was this an annoying part of this exercise. My self-imposed insistence in using 10 tph base headways worked pretty well everywhere else in the map, but makes a headache here. 20 tph east of Newton Highlands is excessive, and even 10 tph to both Riverside and Needham probably is a bit more than needed... Riverside probably could suffice with 7.5 tph, and Needham could probably do 6 tph. But 7.5 and 6 don't divide nicely into 10, so obviously (he said sarcastically and self-deprecatingly) I had to come up with a "better" solution.

What I landed on was a "base" frequency of 5 tph to each spur, with a layer-on service that increases the frequency using short-turning trains terminating at a new Back Bay Green Line station and/or at Kenmore. (Assuming the new Back Bay station could be built to accommodate short-turns.) On the map, I put the layer-on service at a full-on additional 5 tph on each spur, which may be a little excessive, but again is meant to explore the "maximum" possible.

Branch Pairings

I've felt for a while that the branches out of Kenmore should terminate in downtown, likely at Park but easily also at Government Center, and possibly at North Station. For the most part, they will be street-running, with surface transit stop spacing rather than rapid transit stop spacing. That's a different kind of transit than most of the rest of the network, and I think is best treated as its own thing.

Assuming we leave radial Chelsea service to BRT, that gives us three northern trunk-branches:
  • Medford and points north
  • Porter and points west (e.g. Watertown, Belmont, Waltham, Weston)
  • Grand Junction & Allston (West Station)
And three southern trunk-branches:
  • Huntington
  • Nubian
  • Seaport
(Porter, Grand Junction, Huntington, and Seaport can all be supplemented by services that bypass downtown -- the northern branches being fed from Sullivan, and having Huntington and Seaport feed each other on the south.)

Huntington basically has to pair with Medford, I would argue: running Huntington to Porter would mean likely running Newton to Waltham, which basically no one will ride end-to-end; running the Huntington to Grand Junction would encounter a similar problem, plus creates asymmetry with a much shorter northern leg than southern leg. Newton - Huntington - Medford pairs similar legs together and avoids doubling back on itself.

I would pair Nubian with Grand Junction. They are similar lengths, and both would have the most grade-crossings of this "Green Line subnetwork". Moreover, pairing them would allow both legs to short-turn at Government Center: if Washington St is seeing bunching, short-turn the next inbound Nubian train and do the same with the next inbound Grand Junction to balance things out.

That leaves Porter and Seaport, which actually works pretty well -- if you draw a straight line from the Seaport to Park St, and then keep going, you eventually end up... in Watertown and Waltham. So this pairing also maximizes direct crosstown trajectories.

Washington St

I believe that Washington St needs headways better than 6 minutes. Currently the SL4 and SL5 combine to offer irregular frequencies that average just under 5 minutes. Pre-pandemic, these frequencies were closer to 4 minutes. As I’ve written about previously, Washington’s large appetite quickly has an outsized effect on capacity in the Tremont Street Subway, given that it branches off so close to the core.

Layering on LRT and BRT services allows for increased frequencies to downtown (and OSRs to a larger variety of locations downtown while still maintaining transfer access to most other core lines). This would also open the door to “express LRT, local BRT” service pairings to speed up connectivity between Nubian and Downtown.

In this particular build-out, I sent Washington St service north to downtown, with SL3 service running from Chelsea into the Transitway, and SL1 converted to Navy Line service running via Summer St and continuing into downtown. There’s a lot of opportunity for playing around here. Washington service could instead turn east down Summer St, or branch to serve both Seaport and Downtown. Airport service could be rerouted back into the Transitway, or could instead run via I-90 into the South Station Bus Terminal itself (an underrated idea, in my opinion).

Assuming a viable and robust BRT corridor between Washington Street and South Station could be built (and I believe it could), then the need for a southeast leg of a junction between the Nubian Branch and a Marginal Rd subway becomes less acute.

In Conclusion

For the most part, this is really just meant to be a thought-exercise illustration of what a full-build could look like. I do like the relative simplicity of this schema: Green has three branches north and three branches south, Emerald has basically the same shape as today’s Green Line, Gold has two branches, Aqua has two branches, and Navy has three northern branches and three southern branches. (The point about colors being too similar is well-taken, though.)

And again, to review, these are the infrastructure enhancements that enable this:
  • D-E Connector to an extended Huntington Subway to Back Bay
  • Creation of a three-way junction at Bay Village
  • Bay Village to South Station subway, with conversion of Transitway to enable LRT
  • Conversion of Grand Junction to LRT
  • Creation of a multi-way junction at Brickbottom
  • Creation of a three-way junction at BU Bridge
  • Creation of an LRT branch from Kenmore to Harvard
  • LRT/BRT transitway along the Eastern Route in Everett and Chelsea
  • “Center City Link” BRT spine through downtown
  • BRT infrastructure for Washington St, Everett, and Chelsea
None of those enhancements lock you into a single service pattern in perpetuity. All increase the flexibility of this system, meaning you can make something that looks like what I’ve done in this post, or something that looks quite different.
 
The eureka moment for me was realizing that there really are two corridors within the Grand Junction: a north-south corridor between Cambridge and Sullivan, and an east-west corridor between Downtown, Cambridge, and Allston. Hence why there are two services, both vaguely S-shaped, and layered on top of each other like this: >---<

So, services from Lechmere should head to Allston, and services from Sullivan should head to Kenmore.

The lack of (direct) Sullivan->downtown (be it Park and points west/south, or GC-Brattle) bugs me. There's no viable location I can find for a transfer station, meaning riding LRT from Everett/Chelsea would require a reverse transfer at the GJ or a transfer to Orange at Sullivan. Not the end of the world, of course, but I do find myself wondering what that might do to ridership projections versus an option with an OSR (even if it was only a minority of cars).

Soooooo this should in theory actually require not that much increase. The Green Lines stay on the outer tracks from Boylston through Park and beyond; Boylston-GC pre-pandemic saw 30 tph, so this could probably be doable. The Emerald Lines would stay on the inner tracks, which currently see 40 tph, until the interlockings between Boylston and Park; the question then becomes whether the Inner Loop can actually handle 30 tph shoveling through it.

That is where we'd probably need to lean on some additional "relief valves", including a reconstructed Outer Loop and potentially running some services through to loop at Government Center, or rerouting more Huntington Line services out to the Seaport -- the operational reality may be a bit less tidy than the idealized version discussed here.

Uh, which one's Green and which one's Emerald again? (For unrelated projects of my own I've taken to referring to a darkish green as emerald, so I'm probably confusing myself.) Color-related confusion aside, it makes sense to run Boylston Street cars on the inner tracks and Tremont Street cars on the outer tracks as much as possible, given where they need to be for the Boylston split. I'm increasingly in agreement that the Outer Loop might well need to make a return, as well as the long-wished-for crossover from Track 3 to Track 4, the latter of which would help improve the efficiency of GC and its loop with better train staging. (Ah, the days where I'd be on a GC-terminating train behind a Lechmere train. The stupefied "huh" expression when they dumped us at the Park Street end of the eastbound platform was almost as amusing as the inevitable confusion whenever a car offloads at the Brattle Loop 🙃 )

One minor quibble, or maybe a question: would those numbers be proper TPH, or more CPH (cars per hour) if we assume monster Type 10s? I can see a relatively like-for-like TPH->CPH conversion rate given that a two-car train is about the size of one T10, but I find myself doubting that anything in the subway could necessarily handle 30+ T10 doubles per hour. (Maybe this is just me wondering whether it's fair to call a one-car GL vehicle a train or not.)

I think you can swing a small LRT/BRT surface terminal, with a modestly expanded footprint. There's a lot of surface space at Kenmore Square that is given over to automobiles; building a true intermodal transfer center seems like a good reason to reclaim some of that space. If you truly cannot, then it's not the end of the world for the GO1 to terminate at West Station instead.

Surface? I suppose a Gold LRT surface terminal solves the problem of B's lack of access to the Kenmore loop, but if there's not room for some tail/pocket tracks there, that seems like it'd do...unpleasant...things to Gold's reliability (...like after Red Sox games...) unless you nuked half the cross roads for a reservation to the square. Not a reason to outright reject the idea, but it'd be one of the things I'd stick a pin in to scrutinize in any study as a potential danger spot.

What I landed on was a "base" frequency of 5 tph to each spur, with a layer-on service that increases the frequency using short-turning trains terminating at a new Back Bay Green Line station and/or at Kenmore. (Assuming the new Back Bay station could be built to accommodate short-turns.) On the map, I put the layer-on service at a full-on additional 5 tph on each spur, which may be a little excessive, but again is meant to explore the "maximum" possible.

Would a new Back Bay need any provision for short-turns? Wouldn't it be possible to configure Bay Village to allow cars coming from BBY to turn back there? (Even as a flat junction it wouldn't be the end of the world for an extra-helping service pattern.) That'd at least save on having to design-build an additional turnback, and in a more-constrained environment to boot.

I've felt for a while that the branches out of Kenmore should terminate in downtown, likely at Park but easily also at Government Center, and possibly at North Station. For the most part, they will be street-running, with surface transit stop spacing rather than rapid transit stop spacing. That's a different kind of transit than most of the rest of the network, and I think is best treated as its own thing.

As a transit nerd, I agree with you on this. As someone who spent the entirety of my high school career commuting from Copley to North Station on the Green Line, I'm extremely happy you weren't designing surface patterns at that time :D. (I know, I could have taken the Orange Line, and I did in fact in the mornings, but the noxious BBY fumes weren't worth the trouble when I actually had to wait there.)

I believe that Washington St needs headways better than 6 minutes. Currently the SL4 and SL5 combine to offer irregular frequencies that average just under 5 minutes. Pre-pandemic, these frequencies were closer to 4 minutes. As I’ve written about previously, Washington’s large appetite quickly has an outsized effect on capacity in the Tremont Street Subway, given that it branches off so close to the core.

Layering on LRT and BRT services allows for increased frequencies to downtown (and OSRs to a larger variety of locations downtown while still maintaining transfer access to most other core lines). This would also open the door to “express LRT, local BRT” service pairings to speed up connectivity between Nubian and Downtown.

In this particular build-out, I sent Washington St service north to downtown, with SL3 service running from Chelsea into the Transitway, and SL1 converted to Navy Line service running via Summer St and continuing into downtown. There’s a lot of opportunity for playing around here. Washington service could instead turn east down Summer St, or branch to serve both Seaport and Downtown. Airport service could be rerouted back into the Transitway, or could instead run via I-90 into the South Station Bus Terminal itself (an underrated idea, in my opinion).

Don't send anything BRT from Nubian to downtown, don't even whisper that you might send it anywhere that LRT could conceivably go. You'd just be asking for trouble. "Sorry, you have to take a bus to South Station/Seaport or reverse-transfer at Park? Boylston? because we don't want the hassle of building another leg of the Bay Village interchange at Marginal" is going to raise eleven different kinds of hell with the people who had the Elevated stolen and replaced with a crappy silver bus. I don't disagree with you with respect to the better flexibility of BRT, and I share the concern about the fact that Nubian's branching so close in magnifies its effect on Tremont capacity: I just think it's worth considering that accepting that outsize impact and potentially some service-pattern compromises might be the cost of business in avoiding some potentially-very-nasty political problems. We've all seen specific constituencies kill projects before because they weren't satisfied, let's tread carefully lest we lose more.

(Most of that's just a messaging issue, or at least manageable as one. BRT as a legitimate add-on to places otherwise unreachable from Nubian, fine. So much of a whiff of it as the way out of a harder LRT option, that's gonna be a problem. Even if the actual plan is good and realistic, its sales pitch has to be modified to account for a repeatedly-screwed-over neighborhood's understandable wariness.)

Now that I've nitpicked the hell out of things, I'll say this: bloody hell, Riverside, this is an amazing piece of work. Absolutely fascinating, well-reasoned with logical justifications for the decisions, and an excellent summation of both the infrastructure work that maximize system flexibility and an incredible illustration about what that flexibility could give our city. Is there some way I can give you like a thousand more likes for these posts?
 
Thanks for the kind words, @Brattle Loop! All of the points you raised are fair ones -- I think all have potential solutions, including in some cases by putting the pieces of the puzzle (i.e. the infrastructure improvements & expansions) together in different combinations. Some specific thoughts below.
The lack of (direct) Sullivan->downtown (be it Park and points west/south, or GC-Brattle) bugs me. There's no viable location I can find for a transfer station, meaning riding LRT from Everett/Chelsea would require a reverse transfer at the GJ or a transfer to Orange at Sullivan. Not the end of the world, of course, but I do find myself wondering what that might do to ridership projections versus an option with an OSR (even if it was only a minority of cars).

Yes -- this is an open question, and for this particular build-out, I made an intentional decision on this point. Like I've said, I'm increasingly convinced that we need to look for radial transportation solutions for Chelsea and to a lesser extent Everett.

But -- as you've alluded to -- there are options. If Lechmere to Government Center (Brattle Loop) can accommodate 40 tph like Copley-Boylston can, then the solution can be simple -- add a fourth branch. If you do that, I'd suggest through-running the Sullivan/beyond branch to Nubian, and cutting back the Grand Junction branch to the Brattle Loop instead, under the theory that the loss of the Red Line transfer will be less acute for a Grand Junction service. F-Line has noted the possibility of quad-tracking Haymarket, so there may be ways to increase capacity to meet 40 tph.

On the other hand, if you are capped at 30 tph, then you can do the slightly-less-elegant-but-still-relatively doable trick of dividing three southern branches into four northern branches, reducing northern headways from 6 minutes to 7.5 minutes, which wouldn't be the end of the world. Additionally, Sullivan has excess capacity in this model, so you could feed extra Sullivan-originating trains on to the Grand Junction, Union (and even Medford) branches to bolster within-branch frequencies as compensation.

Finally -- and I am less fond on this idea, but it does have potential-- Lechmere-Sullivan could be fed as a "half-branch": for example, the Grand Junction 10 tph branch I have in my diagram could be split into a 5 tph Lechmere-Grand Junction branch and a 5 tph Lechmere-Everett branch. Given that the Everett segment would be supplemented by layer-on bus service feeding in to Sullivan, the 12-min headways could still be useful (similar to what I proposed for a resurrected A Line/57A + "SL57" combination); I'm less convinced that 12-min headways on a Lechmere-Grand Junction branch are viable -- I think that wait time is too long. (Riders will plan for different routes with less brittle itineraries, and the lack of transfer options make it less useful for opportunisitic journeys.)

As you said, it merits study. To me, I think the case for Lechmere-Sullivan service is much stronger if there is a dedicated Everett branch, so I think that's a variable that plays into this as well. But either way, there should be sufficient slack here to play around with this.
Uh, which one's Green and which one's Emerald again? (For unrelated projects of my own I've taken to referring to a darkish green as emerald, so I'm probably confusing myself.)

Haha yeah, as much as I love the poetry of an "Emerald Line", particularly given the connection to the Emerald Necklace, I will grant that it's probably more confusing than it's worth. (I'm not sure I made this explicit, but part of the schema is that the "less common" color names like Emerald, Silver, Gold, and Aqua are supposed to suggest the "light metro" tier of service I proposed a few months ago, distinguished from the "heavy metro" of Red, Orange, Green, and Blue.)
Color-related confusion aside, it makes sense to run Boylston Street cars on the inner tracks and Tremont Street cars on the outer tracks as much as possible, given where they need to be for the Boylston split. I'm increasingly in agreement that the Outer Loop might well need to make a return, as well as the long-wished-for crossover from Track 3 to Track 4, the latter of which would help improve the efficiency of GC and its loop with better train staging. (Ah, the days where I'd be on a GC-terminating train behind a Lechmere train. The stupefied "huh" expression when they dumped us at the Park Street end of the eastbound platform was almost as amusing as the inevitable confusion whenever a car offloads at the Brattle Loop 🙃 )

That last parenthetical point is actually one reason why I'd like to give the Kenmore branches a separate color and line identity altogether. A separate trunk, a separate service pattern, very little overlap...

Yes, in an ideal world, the Kenmore ("Emerald") trains would run completely isolated on the inner tracks, and the Tremont ("Green") trains would operate independently on the outer tracks. In practice, I think the flexibility of two tracks in each direction available to each service and two loops available to Emerald trains would be important for reliability, particularly since the loop makes it harder to hold trains out of service as needed, etc etc.
One minor quibble, or maybe a question: would those numbers be proper TPH, or more CPH (cars per hour) if we assume monster Type 10s? I can see a relatively like-for-like TPH->CPH conversion rate given that a two-car train is about the size of one T10, but I find myself doubting that anything in the subway could necessarily handle 30+ T10 doubles per hour. (Maybe this is just me wondering whether it's fair to call a one-car GL vehicle a train or not.)

Yeah, this is a good question that I have not contended with. I've been thinking in terms of TPH, not CPH, and really more specifically in terms of "occupied signal blocks". I'm not too concerned about Type 10 singles, but agree that Type 10 doubles are potentially a different beast.

For what it's worth, I do think that this is a case where my "heavy metro" vs "light metro" distinction becomes actionable. The Emerald Line out of Kenmore sees close stop-spacing, a high number of branches, and a lengthy shared trunk that acts like a line unto itself with a genuine within-corridor market. The Green Line out of Tremont Street sees more sealed ROWs, longer stop spacing, and has a trunk that is more focused on transfers and journeys into the core, and branches which themselves act more like the Boylston St Subway between Kenmore and Boylston.

In a system like the Emerald Line, a "light metro" is preferable because it usually means you can shovel more trainsets with distinct destinations through a core, allowing all branches to maintain reasonable frequencies. That might mean sacrificing capacity within each trainset in order to accommodate a greater number of trainsets and thereby keep wait-times viable for all journeys.

By contrast, the Green Line system I'm describing here fits a "heavy metro" model much better. The longer distances out to the suburbs, the longer stop spacing, the presence of feeder bus services to individual branches -- this is more like the Red Line or Orange Line, where longer trains become more worth it in a potential trade-off between frequency of trainsets vs size of trainsets. (For example, being able to accommodate three feeder buses' loads of passengers boarding simultaneously at a place like Nubian.)

All of which is to say: I'm a bit less skeptical that Type 10 doubles are the most urgent on the Kenmore branches. Yes, more capacity is needed, but I think it'll be important to keep branch frequencies near where they are now, even if it means reducing the number of supertrains.
 

Back
Top