Green Line Reconfiguration

Simplifying the Reconfigured Green Line

The brilliance of the Reconfigured Green Line concept lies in its use of Aldgate Junctions to enable increased service without requiring increased capacity in core areas. The downside is that the Reconfigured Green Line quickly becomes complex to map and to explain. (Consider my previous posts on this subject.)

This has operational implications as well. The most extreme version of the Reconfiguration (which no one here is seriously advocating for) creates a highly interlined network, with all the ensuing sequelae: complex dispatching, cascading delays, and difficult navigation. But even more modest visions of the Reconfiguration face these same problems, though in milder form.

Additionally, several routings we’ve discussed in this thread are tempting on paper, but end up either being roundabout or indirect, including:
  • Chelsea - Downtown via Sullivan
  • Kendall - Kenmore via BU Bridge
  • Downtown - Seaport via Bay Village
  • Kenmore - Huntington via Riverway
So, what’s a good way to take maximal advantage of the Reconfigured network while avoiding the pitfalls above? Here’s my take:

Emerald Line

Running via the Boylston Street Subway, with 30-35 tph on the trunk, short-turning at Park Street.

A: Oak Square - Park Street via West Station
B: Boston College - Park Street via Commonwealth Ave
C: Boston College - Park Street via Beacon Street
R: Riverside - Kenmore (with occasional through-runs during peak)

Boston’s continued legacy streetcar-subway network, relieved of as many burdens as possible and largely isolated from other services.

Green Line

Running through the Central Subway from Lechmere to Bay Village, with 25-30 tph on the trunk:

D: Needham Junction to Medford (or points north) via Huntington
E: Heath Street or Hyde Square or Jackson Square or Forest Hills to Porter (or points west) via Huntington
F: Nubian to Government Center via Washington
G: Nubian to West Station via Washington and Grand Junction

Washington and Huntington each get doubled branches with layered short-turns.

Excess capacity remains available on the northside; this can be used in the short-term for extra Brattle Loop services to modestly increase frequencies on northside branches from 7.5 tph (8 min headways) to perhaps 10 tph (6 min). In the long term, the capacity can be reallocated to multiple branches diverging at Porter (e.g. Waltham and Watertown). I maintain that service to Chelsea is not an efficient use of Central Subway capacity, though I could be convinced by additional data.

The reality is that there are a dozen ways to slice-and-dice Green Line pair-matching, so it’s worth keeping key principles in mind:
  • High frequencies on Washington
  • Layered high frequencies on Huntington
  • Short-turning Huntington trains at Heath frees up slots for Kenmore-Riverside shuttles to keep frequencies high to Reservoir
  • Per-branch frequency demand on the northside branches will be lower than the southside trunks along Washington and Huntington
  • Chelsea-Downtown service is hard to justify
Gold Line

Running via the Grand Junction and dedicated lanes south of the Charles River, with 6-10 tph on this service, doubling similar frequencies on other routes sharing the ROW, such as the G and the Q, providing sub-5-minute headways for local journeys.

J: Airport - Nubian via Sullivan & BU Bridge

This is the Urban Ring, straight and simple.

By necessity, this route sees the most interlining, but with limited scope:
  • Northeast quadrant, shared with BRT services
  • Grand Junction in Cambridge, shared with one (lower-frequency) branch of the Green Line, which itself keeps a smaller footprint in the Central Subway
  • Southwest quadrant, shared with Aqua Line
What’s more, the Gold Line would benefit from the flexibility of the Aldgate Junctions, enabling temporary diversions to West Station, Lechmere, or into one of the subways from the south.

(Also – and this couldn’t be less important – this means that the Grand Junction in Cambridge would be served by the G and J lines. Is this coincidental? I assure you not. Can it be justified for reasons other than transit nerd perfectionism? Barely, but I think that’s enough!)

Aqua Line

Running to the Seaport via Back Bay, with anywhere from 6 to 15 tph on the trunk.

Q: Seaport to Harvard via Huntington and West Station

This line will be a monster, linking together the Seaport, South Station, Back Bay, the Prudential, Northeastern, Longwood Medical Area, Boston University, West Station, and Harvard.

In principle, the E (or the A, or the R) could be shifted over to the Aqua Line to run to the Seaport. However, I believe that this primary corridor will likely merit full frequencies end to end; additionally, I believe legibility is improved with no branching.

Silver Line

Providing circumferential BRT service via a layer of services running between Kenmore and Nubian to the south and Airport Terminals, Airport Station, Chelsea and Wellington to the north, via the LMA Transitway, dedicated BRT infrastructure on Melnea Cass Boulevard, Albany St, Summer St, the Ted Williams Tunnel, and the Silver Line Gateway.

Map

(Dashed lines indicate potential future expansions; dashed lines with an arrow indicate feeder services.)

Grand Mini Map - Simplifying.png
 
A rather simple alternative from B C and D trains to the Seaport. All Huntington cars go Prudential - Back Bay - Bay Village - Boylston . . . A new Green Line from Copley, under Dartmouth Steet to Back Bay. Thence to Bay Village, Marginal Street tunnel . . . No double transfers for Brighton, Brookline and Newton passengers
 
(And @The EGE — to be clear, I am by no means trying to one-up your very elegant system design. We may be coming up with somewhat different solutions, but I think we’re tackling the same problems — and I definitely think they are the right problems to tackle! It was a real pleasure to read your posts.)
 
A rather simple alternative from B C and D trains to the Seaport. All Huntington cars go Prudential - Back Bay - Bay Village - Boylston . . . A new Green Line from Copley, under Dartmouth Steet to Back Bay. Thence to Bay Village, Marginal Street tunnel . . . No double transfers for Brighton, Brookline and Newton passengers
,

Forcing the entire Boylston Street contingent into the Bay Village alignment completely negates the benefits of multiple trunk lines for this stretch (and would put a capacity ceiling on both Copley-Kenmore and Bay Village-Huntington, well, maybe unless it was quad-tracked).

Moreover, while it might be technically possible to shiv a tunnel under Dartmouth to meet the Boylston Street Subway at Copley (the square itself and the wide sidewalk in front of the BPL mean it's blissfully unconstrained on Dartmouth in that block), it's completely impossible to grade-separate a junction there without obliterating the existing station (and on the outbound side, there's nowhere else you can build a replacement), meaning that you'd either need a flat junction (and that's worked so well at Copley, hasn't it) or you just eliminate Copley as a stop (at least outbound)...which doesn't seem likely to be feasible.

Also, the Aqua Line (I think, @Riverside while I'm fascinated by the ideas, the colors continue to confuse me:) has transfers from the "Emerald"/current Green Line branches, just outbound of Kenmore rather than in the core, so it'd be a single transfer (from the B, for example, at BU Bridge) rather than a double. Emerald's lack of a connection to the Teal? BRT? thing is mildly awkward and would make for extra transfers. I assume the rigidity of turns at Park Street is a capacity issue on the Park-GC tube (though in a real-world implementation I'd probably throw in some run-as-directed cars to be able to say that there's direct connection...even if it isn't much of it.)

Running to the Seaport via Back Bay, with anywhere from 6 to 15 tph on the trunk.

Q: Seaport to Harvard via Huntington and West Station

This line will be a monster, linking together the Seaport, South Station, Back Bay, the Prudential, Northeastern, Longwood Medical Area, Boston University, West Station, and Harvard.

I probably missed it, but is this as LRT? I thought I recalled some serious (i.e. Crazy Transit Pitches and/or God Mode-level complicated) impediments to tunneling the southern chunk of the Urban Ring, but obviously if this is surface LRT (even if that means eating some lanes...which would be nice) that isn't a concern.
 
@Brattle Loop the colors are the best part! :LOL: But yes — the Aqua Line (“Teal Line” would sound too much like “T Line” over a PA, you see :p) is full LRT for maximum capacity. No tunneling through Longwood:
  • Surface: Falcon Terminal - World Trade Center
  • Subway: World Trade Center - Northeastern
  • Surface: Northeastern - BU Jct
  • TBD: BU Jct - Harvard
Using a surface transitway definitely isn’t a cure-all — even with signal priority, I have to think the number of intersections to traverse will slow you down at least a little bit. But Park Dr is very wide and so could fit dedicated ROW and Fenway has relatively few abutters — mainly colleges, who would probably be willing to negotiate about the loss of a street for a massive gain in transit connectivity.

Transfers between Emerald (Kenmore) LRT and Navy (Congress) BRT continue to be the fly in the ointment.

As you’ve noted, Seaport access is address by transferring to the Aqua Line from one of the branches. Copley and Arlington likewise have Aqua stations only a few blocks away; Kenmore-Seaport has an OSR via SL1, or a single transfer via one of the branch stations. That leaves Hynes as a gap; we might hope that improved service on the 1 bus would make it a trivial single transfer via Symphony, but it’s not perfect.

For the other areas served by Congress St BRT:
  • Allston/Brighton/Brookline/Highland > Seaport can be achieved via transferring Emerald > Aqua
  • Allston/Brighton/Brookline/Highland > Logan can be achieved via Emerald > SL1 [Kenmore-Logan], or via Emerald > Gold > Massport Shuttle
  • Allston/Brighton/Brookline/Highland > Southie can be achieved by bus transfer at Copley via the 9 (or "T9")
    • (Or by extending Seaport LRT down into Southie proper, which seems like significant cost for modest gain)
  • Allston/Brighton/Brookline/Highland > Chelsea/Everett can be achieved via Emerald > Gold and potentially a Last Mile bus/BRT transfer
    • This one bothers me the most -- a lot of the population here is not centered on the Eastern Route, so the radial BRT is more effective at reaching people where they are; having the transfer to the circumferential Gold definitely is helpful but probably still ends up being a double transfer
To your point about RAD’s up to Gov’t Center instead of turning all Kenmore trains at Park —certainly could be part of the equation. In my design I’m running 7.5 tph Nubian-Government Center; that could be swapped for one of the Kenmore branches to run all the way through, or take two of the Kenmore branches (A & C) and run every other train up to Government Center and have a short OOS transfer to State; the downside is that this reduces Nubian connections to Blue, although you’d still have the other half of Nubian trains (the “G”) available to provide that link.

~~~~

Two crazy ideas come to mind if we’re really looking to Connect Everything and create Two Seat Rides To Everywhere Via Downtown (could be used standalone or in tandem):

1) Quad track the Back Bay-Bay Village subway, build a connection between Hynes and Back Bay (with or without a Copley intermediate), reroute all Kenmore services via Back Bay, and then distribute all Kenmore, Huntington, and Nubian services evenly across the Tremont St Subway and the Seaport (or use my surface LMA transitway to connect into a quad track subway), as you and @Fred R discussed

2) Build a subway under Congress St and connect it to Back Bay LRT service (my Aqua), potentially with an Aldgate junction to also run Congress-Seaport service. Then build radial LRT to Chelsea and Everett and run it into Congress St

But both of these would be significantly more complex and complicated; they might be justifiable as "second-generation" projects after an initial set of builds, but I don't think they are necessary as fundamental pieces.

~~~~

The lack of direct Emerald-Navy (Kenmore-Congress) connection is emblematic of a larger problem: you can only layer in so many service patterns into the stretch between Park and Government Center. The maximum frequency on a service decreases exponentially as the number of services it has to share tracks with increases. (Obligatory blog post here.) And as we saw in my suggestion to short-turn my "F" at Park, and extend some of the Kenmore trains to take its place, you still have to turn some services at Park, and I'd argue that you end up with deteriorated service for everyone by trying.

Transfers are difficult, but are less so when they are between frequent reliable services. Turning Kenmore trains at Park enables 6-min headways and creates a "sealed" set of tracks that minimize disruption from delays elsewhere on the system. A Blue Line extension to Kenmore maintains the transfer to Blue Line while also providing a faster ride into downtown (and it'll be an easy transfer too -- just downstairs, and very easy to look at the departure boards and make a gametime decision which service will get you downtown faster). Finally, turning Kenmore trains at Park also enables maximum frequencies between Park and Government Center -- 30 tph, with a cross-platform transfer to either a waiting train or one that arrives in less than 2 minutes.

I don't think using RADs to provide a nominal through-service from Kenmore to Government Center fully solves the problem -- you still don't actually get the direct connection to Congress St, and it makes the system more confusing and less legible. The suburb-to-suburb Brookline-Chelsea journey becomes a double-transfer depending how close to the Eastern Route ROW in Chelsea you're going for... but the reality is that many suburb-to-suburb journeys are still going to be double-transfers, due to the Last Mile Problem. So while this bothers my perfectionism as well, I think this design solves enough of the Big Problems that I'm willing to live with the imperfections.

~~~~

Also.

Walking between Park Street and State/Congress takes less than 10 minutes, 2-3 of which can be done within the Winter St Concourse. Technically, you could do almost all of it within MBTA subway stations, with literally only one block out in the open:

Park-State walking transfer.png


For crowding reasons, I probably would not encourage transferring passengers to enter State at Milk St, but even so, it's still a relatively short walk for able-bodied commuters. If the T can eventually manage to implement OOS transfers (out-of-system, sorry, probably should've defined that acronym earlier), then you can improve connectivity across the system (Chinatown-Boylston, Mass Ave-Symphony, I'm looking at you). So this could be another way to mitigate the lack of Emerald-Navy transfers to support journeys between Allston/Brighton/Brookline and Everett/Chelsea.

Also, I should note: the T already has OOS transfers; you just have to able to pay for a pass upfront. To put it another way, if you are middle-class with enough liquid cash to shell out 100 bucks at the end of each month, you enjoy significantly more flexibility to utilize the T's infrastructure most efficiently; if cash is tight for you, and you can't afford that or the weekly $22.50, then you are stuck waiting for the T to shuttle you through their tunnels so you can make the most of the $2.40 you spent to go through the faregate -- even if it means waiting 12 minutes for a delayed Red Line train just so you can go one stop.
 
I really like this. It shows just how distinct the legacy "Kenmore Division" is from the rest of the lines and how the Green Monster is essentially a transit system unto itself. I also think it would be fairly self-explanatory to people who don't read ArchBoston.

A: I know in the past we've discussed a SL57 proposal. This would seem to mostly negate the need for it, right? If you're running the A at 6-10tph, running a SL57 seems like overkill on the corridor. (I do think if you can find a cheap way of getting across Newton Corner to Watertown Yard the value of getting a yard site at the end of a line and killing a Key Bus starts to look appealing, especially because the Tremont St. section could probably survive without a dedicated RoW)

F/G: Is the F essentially just a short-turn version of the G or am I missing something here? I think Brickbottom would have to be the reporting location, so I wonder if it makes sense to extend it to Lechmere given the amount of deadhead?

J: You've convinced me that Chelsea-North Station LRT has limited value compared to an enhanced (T)111 as well at the T7 corridor! I do think that non-revenue trackage accommodating it and being able to run it during (frequent) Tobin work would add real value to the network though. (I also think Watertown falls in this category where [re🙃] enhancements to the 71 would be more valuable than LRTing the Watertown Branch)

Q: I think we've talked about LRT conversion of the transitway including conversion of the existing loop at South Station, but I do feel like the South Station-Seaport segment is worthy of additional peak short-turns if headways are relatively infrequent.

I do hope that the revised FAA rules would allow Massport to kick in a bit more than they do now with all this service to Airport (and that they finally build the damn Peoplemover).

Wow, this thing is probably gonna require a Riverside-esque facility north of North Station though.

Also, I should note: the T already has OOS transfers; you just have to able to pay for a pass upfront. To put it another way, if you are middle-class with enough liquid cash to shell out 100 bucks at the end of each month, you enjoy significantly more flexibility to utilize the T's infrastructure most efficiently; if cash is tight for you, and you can't afford that or the weekly $22.50, then you are stuck waiting for the T to shuttle you through their tunnels so you can make the most of the $2.40 you spent to go through the faregate -- even if it means waiting 12 minutes for a delayed Red Line train just so you can go one stop.

Oh I know all about this from my days of living in Quincy, working in Cambridge, and seeing someone who lived in East Boston :ROFLMAO:
 
I really like this. It shows just how distinct the legacy "Kenmore Division" is from the rest of the lines and how the Green Monster is essentially a transit system unto itself. I also think it would be fairly self-explanatory to people who don't read ArchBoston.

Yup, exactly. (On all three counts!)

A: I know in the past we've discussed a SL57 proposal. This would seem to mostly negate the need for it, right? If you're running the A at 6-10tph, running a SL57 seems like overkill on the corridor. (I do think if you can find a cheap way of getting across Newton Corner to Watertown Yard the value of getting a yard site at the end of a line and killing a Key Bus starts to look appealing, especially because the Tremont St. section could probably survive without a dedicated RoW)

I think it's something that we can leave flexible in the plan. This iteration of the A Branch really is an extension + replacement of the 57A, meaning there'd still be potential use for a full length bus route between Kenmore and Watertown Yard. Whether it's at BRT frequencies/standards is a separate question, but definitely a possibility.

F/G: Is the F essentially just a short-turn version of the G or am I missing something here? I think Brickbottom would have to be the reporting location, so I wonder if it makes sense to extend it to Lechmere given the amount of deadhead?

You are not -- F is a short-turn of G to provide high frequencies on Nubian that don't pair-match well to anything on the northside. (The fact that it makes good on the promise of the "F Line to Nubian", plus that the Grand Junction gets served by the G and the J, is of course entirely coincidental.) My intention was that the F turns using the loop at Government Center. That's a fair point about access to the yard though. Potentially could be addressed by running the F and G as a single service where each car goes something like GC - Nubian - GC - Grand Junction - GC - Nubian - GC and then on back to Nubian -- basically two Nubian trips for every Grand Junction trip. But definitely would require some planning. But yes -- there is some slack capacity north of Government Center, so F trains going out of service could run-as-directed to Lechmere, for example.

J: You've convinced me that Chelsea-North Station LRT has limited value compared to an enhanced (T)111 as well at the T7 corridor! I do think that non-revenue trackage accommodating it and being able to run it during (frequent) Tobin work would add real value to the network though. (I also think Watertown falls in this category where [re🙃] enhancements to the 71 would be more valuable than LRTing the Watertown Branch)

The tricky part here is that, as @The EGE alluded to, it's somewhat tricky to thread (bidirectional) Lechmere-Sullivan tracks through the spaghetti at Brickbottom. There is a northbound yard track that branches off after Lechmere, so potentially that connection could be used on a limited basis when needed (particularly if run in peak direction only).

Q: I think we've talked about LRT conversion of the transitway including conversion of the existing loop at South Station, but I do feel like the South Station-Seaport segment is worthy of additional peak short-turns if headways are relatively infrequent.

I agree... if headways are relatively infrequent... which in my opinion they shouldn't be. :) Again, this service is going to hit practically every major employment center, and two of the largest transit centers in the region. (I exaggerate a little bit, but only a little bit: Harvard, BU, Longwood, Prudential, Back Bay, South Station/Financial District, Seaport.) In fact, IIRC, as the crow flies, the Q takes a very slightly more direct route from South Station to Harvard. If we do it right, it should be massively successful, and should have no need for short-turns at South Station.
 
You are not -- F is a short-turn of G to provide high frequencies on Nubian that don't pair-match well to anything on the northside. (The fact that it makes good on the promise of the "F Line to Nubian", plus that the Grand Junction gets served by the G and the J, is of course entirely coincidental.) My intention was that the F turns using the loop at Government Center. That's a fair point about access to the yard though. Potentially could be addressed by running the F and G as a single service where each car goes something like GC - Nubian - GC - Grand Junction - GC - Nubian - GC and then on back to Nubian -- basically two Nubian trips for every Grand Junction trip. But definitely would require some planning. But yes -- there is some slack capacity north of Government Center, so F trains going out of service could run-as-directed to Lechmere, for example.

Operationally, does it really matter where each car is turned versus where it's signed? Could you just sign some basically-F cars as G-Nubian heading west/south from Brickbottom/Lechmere entering service and extend them north to Lechmere from GC when they need to come out of service, or just interchange them to the other northern branches as needed? (Mostly I'm wondering based on a time when I found out that Chicago's CTA interlines cars bound for their Brown Line out of the Midway yard, running them as Orange to the Loop then as a completely different service after that, and vice-versa. On the subject of that "Brownage" line, it's a pleasant surprise when it's 7 degrees and you're dreading the change at the Loop to get to Midway.)

The tricky part here is that, as @The EGE alluded to, it's somewhat tricky to thread (bidirectional) Lechmere-Sullivan tracks through the spaghetti at Brickbottom. There is a northbound yard track that branches off after Lechmere, so potentially that connection could be used on a limited basis when needed (particularly if run in peak direction only).

I still think finding a way to handle the Sullivan-Lechmere leg of an LRT Urban Ring would be extremely useful for operational flexibility. Being able to feed the northeast quadrant of the Ring (slash Airport side of the Gold Line) from the Central Subway probably wouldn't be a necessary full-time service pattern (though potentially for either operational or political purposes, Everett/Chelsea might rate some limited Brattle Loop short-turns. It would absolutely come in handy in the event of other-line service disruptions; having the ability to, for instance, maintain the one-seat ride (even via a more circuitous route) between GC and the Airport even if the Blue Line was shut for work would be useful, and it would also be good to be able to not have to short-turn as much on Gold in the event of work on the GJ section. Might not make it on cost grounds, but depending on how the Sullivan-GJ link had to go, would be at least worth exploring.

I agree... if headways are relatively infrequent... which in my opinion they shouldn't be. :) Again, this service is going to hit practically every major employment center, and two of the largest transit centers in the region. (I exaggerate a little bit, but only a little bit: Harvard, BU, Longwood, Prudential, Back Bay, South Station/Financial District, Seaport.) In fact, IIRC, as the crow flies, the Q takes a very slightly more direct route from South Station to Harvard. If we do it right, it should be massively successful, and should have no need for short-turns at South Station.

Agree on the "shoulds". That said, we've witnessed instances where lack of, inadequate, or improperly-located turnbacks have caused problems for services in case of disruptions. I think you're correct that, done properly, the Q shouldn't need short-turns at South. If that's an argument that we shouldn't need the loop in regular service, I agree, but to the extent that it's being interpreted as an argument we don't need to convert the loop at all, I think we absolutely do.
 
Riverside, this is awesome work. I definitely like your idea that services that don't run through Park Street get different colors. Some comments, in no particular order:
  • In practice, "green" and "emerald" are too close to work on a map. I arbitrarily chose magenta for that reason, but it's probably getting close to the point where colors are no longer supportable and you have to switch to numbers.
  • A key tenet of my system is that the Kenmore Division and Bay Village division are isolated in revenue service, so that delays on one line don't cascade. Mixing the Highland Branch services is probably fine, especially since the one turns at Kenmore, but I don't love mixing the A with the Q and R. You're probably not getting dedicated lanes for the A past Union Square - Cambridge street probably isn't wide enough for 2 general lanes + 2 transit lanes + bike lanes - so if it exists it shouldn't be sharing tracks with heavy trunk lines.
  • In general I'm very nervous about the Q and J. Those are very long lines with a lot of grade crossings. Something I've found doing transit projects in my day job is that as your frequencies go up, your ability to get priority at grade crossings (intersections or grade crossings on private ROW) goes down. On the Grand Junction and Chelsea lines you can mitigate that with some targeted grade separations, especially since most of those Everett/Chelsea crossings need to get eliminated for Regional Rail anyway, but the Fenway, Lower Allston, Ruggles-Nubian, and Huntington Avenue segments are going to be brutal. (Same with trying to get high frequencies on Washington Street.) It's why I went aggressive on grade separation for all high-frequency segments.
  • The junction between the Union Square/Porter and Grand Junction lines is going to be nasty, since you have to separate the Grand Junction tracks entirely from the Fitchburg Line. No way you're getting a transfer station in there. Inbound, it will be a steep continuous grade from the duck-under of the Fitchburg Line until it meets the Medford Branch tracks.
    1667856641532.png
  • Nitpick: you're missing the current Union Square (Somerville) station on your map
  • No Mystic Valley Parkway extension???
 
  • In practice, "green" and "emerald" are too close to work on a map. I arbitrarily chose magenta for that reason, but it's probably getting close to the point where colors are no longer supportable and you have to switch to numbers.

Would this be where we'd be looking at something of a more NYC-style nomenclature, with lettered and/or numbered services (we're already sort-of used to letters on the Green branches), with or without different colors based on trunk/lines configurations?
 
Operationally, does it really matter where each car is turned versus where it's signed? Could you just sign some basically-F cars as G-Nubian heading west/south from Brickbottom/Lechmere entering service and extend them north to Lechmere from GC when they need to come out of service, or just interchange them to the other northern branches as needed? (Mostly I'm wondering based on a time when I found out that Chicago's CTA interlines cars bound for their Brown Line out of the Midway yard, running them as Orange to the Loop then as a completely different service after that, and vice-versa. On the subject of that "Brownage" line, it's a pleasant surprise when it's 7 degrees and you're dreading the change at the Loop to get to Midway.)
Dispatching definitely is not an area of expertise for me -- @The EGE probably has good comments here about the logistics of running a short-turn service that doesn't have a (substantial) yard at either end. I'm pretty confident that there's a viable solution one way or another, though; the "high water" mark of my proposal is 30 tph between Park and GC (15 tph each to Huntington and Nubian), and short-turning the F means that you only have about 23 tph from GC to Lechmere, so there would be "slots" for F trains going out of service to be extended as-needed to Lechmere.

Mostly the point of short-turning the F is:
  • keep service tight on the Nubian branch
  • avoid unnecessarily high frequencies on one of the northern branches
  • avoid a weird 2:3 mapping between northside and southside branches (this one's sorta hard to explain, so I'll leave it summarized here, but am happy to elaborate if anyone wants)
And then just in general, the idea is to demonstrate a workable model, leaving room for variability in the future but ensuring viability on Day 1. So, yes, absolutely there are a number of potential "practical" implementations of a short-turned F:
  • extend runs as needed to reach the yard
  • integrate F and G as single service
  • short-turn F at GC on paper only; in practice, destination sign changes at dispatcher's discretion and train continues north
  • in a second or third generation build, F gets extended north to take over a fourth branch (e.g. Watertown or Everett)
  • various combinations of the above
I still think finding a way to handle the Sullivan-Lechmere leg of an LRT Urban Ring would be extremely useful for operational flexibility. Being able to feed the northeast quadrant of the Ring (slash Airport side of the Gold Line) from the Central Subway probably wouldn't be a necessary full-time service pattern (though potentially for either operational or political purposes, Everett/Chelsea might rate some limited Brattle Loop short-turns. It would absolutely come in handy in the event of other-line service disruptions; having the ability to, for instance, maintain the one-seat ride (even via a more circuitous route) between GC and the Airport even if the Blue Line was shut for work would be useful, and it would also be good to be able to not have to short-turn as much on Gold in the event of work on the GJ section. Might not make it on cost grounds, but depending on how the Sullivan-GJ link had to go, would be at least worth exploring.
I agree with the value of the connection to promote operational flexibility. I'll get into this more below, but I think it matters what the target is; if it's full revenue service, then you really need to have full flying junctions, but if it's only under certain conditions then it may be possible to "settle" for a flat junction in certain places. That will have a big impact on cost.

@ceo and I went back and forth a little bit about how to expand Brickbottom Junction into a full Aldgate junction back in June (and we didn't even try to tackle the western part of the junction, at the McGrath Hwy divergence). I haven't been able to lay my hands on the GLX build specs to see what may already be future-proofed, but looking at the satellite view, it's gnarly.

This is what it looks like today, plus hypothetical LRT tracks coming down from Sullivan:

1667847956088.png


Blue indicates the yard leads. Connecting Sullivan-McGrath would probably just be a matter of hooking into the yard leads coming from the current Union branch, and then doing some rearrangement of the yard and/or mainline tracks, so I went ahead and drew in that connection here. You can see the northbound spur, currently not connected, but well positioned for a northbound Lechmere -> Sullivan connection, like this:

1667849217506.png


That's the easy part. The hard part is getting from the Sullivan southbound track to the Lechmere southbound track.

In principle, you could go waaaaay up high and swing out and around to join the Lechmere southbound track on the outside:

1667849331781.png


But I have no idea how much running space you'd need in order to climb the "double high" height at a reasonable grade, and then swinging "outside" requires expanding the footprint even more. I have no idea if this approach is feasible, and even if it is, I'm sure it's among the most expensive options.

If you shift all Lechmere northbound traffic on to the diverging track and add a connection between that curve and the northbound Medford track on the far side, you can shorten your "double high" viaduct and join the Lechmere southbound track on the "inside":

1667859808929.png


This also seems pretty daunting.

What seems more achievable -- at the expense of the flying junction, hence my open question about revenue vs special circumstances service -- is building on to the existing Lechmere northbound yard lead, adding a second track with a level crossing. (It's a little hard to tell on the diagram, but I've sketched this as lefthand running so that southbound Sullivan-Lechmere trains don't block northbound Lechmere-Union trains while waiting to merge. I am by no means married to this concept.)

1667860267241.png


That said, if someone can come up with a reasonable design that incorporates a flying junction, I'm all for it. I just haven't been able to figure it out myself yet.

Agree on the "shoulds". That said, we've witnessed instances where lack of, inadequate, or improperly-located turnbacks have caused problems for services in case of disruptions. I think you're correct that, done properly, the Q shouldn't need short-turns at South. If that's an argument that we shouldn't need the loop in regular service, I agree, but to the extent that it's being interpreted as an argument we don't need to convert the loop at all, I think we absolutely do.

Oh yeah, I definitely would want to keep & convert the loop to have it available for turnbacks.

@The EGE, this post has been sitting open in a tab all day, so I'm gonna go ahead and post, but I did see your reply -- great points and appreciate the response! More thoughts later.
 
Riverside, this is awesome work. I definitely like your idea that services that don't run through Park Street get different colors. Some comments, in no particular order:

In practice, "green" and "emerald" are too close to work on a map. I arbitrarily chose magenta for that reason, but it's probably getting close to the point where colors are no longer supportable and you have to switch to numbers.
Yeah, I admit I'm biased on this point -- the colors are just straight up fun and aesthetic to me. To be fair, there are systems that distinguish between a "light green" and a "dark green" line (which is really what I'm going for here, "emerald" just being another word for "green" that also has local connections thanks to the Emerald Necklace). I think the colors vs numbers issue is moot here -- we already are at a system design that calls for letters.

From a placemaking perspective, we could still call the A, B, C, and R, the "Green Line", and just have a different shade of green on the map. To @737900er's point, a full-build LRT network quickly turns into "the Green Monster", which is why -- as you note -- I give different colors for the routes that bypass Park St, and likewise think a different color is justifiable for the "Kenmore Division".
A key tenet of my system is that the Kenmore Division and Bay Village division are isolated in revenue service, so that delays on one line don't cascade. Mixing the Highland Branch services is probably fine, especially since the one turns at Kenmore, but I don't love mixing the A with the Q and R. You're probably not getting dedicated lanes for the A past Union Square - Cambridge street probably isn't wide enough for 2 general lanes + 2 transit lanes + bike lanes - so if it exists it shouldn't be sharing tracks with heavy trunk lines.
Yes, this was something I had mixed feelings about. The Q also becomes somewhat problematic in that it intermingles with the "heavy metro" D/E as well as the "light metro" J, which also is less ideal (though I also gave up the ghost on perfect separation there by having the G and J mix on the Grand Junction).

I haven't had a chance to play around with the junction layout at Boston University, but, assuming a Commonwealth Subway, I did consider that it might be advantageous simply to have two sets of tracks -- one set "downstairs" for the A (headed to the subway) and one set upstairs for the G, J, & Q (headed to the surface transitway), but that's likely overkill.

The A doesn't need to go to West Station, though. It could just follow the 57's route. West Station will have connections to Back Bay via the Q, and downtown via the G (plus via Regional Rail). And... ultimately what I want is to extend the Blue Line to West Station, which would eliminate the need for the A to West Station altogether. That's a much larger capital expense, though, and while I want the Commonwealth Subway future-proofed to support HRT tracks as well (hence the dashed line), I don't want to make it a prerequisite.
In general I'm very nervous about the Q and J. Those are very long lines with a lot of grade crossings. Something I've found doing transit projects in my day job is that as your frequencies go up, your ability to get priority at grade crossings (intersections or grade crossings on private ROW) goes down. On the Grand Junction and Chelsea lines you can mitigate that with some targeted grade separations, especially since most of those Everett/Chelsea crossings need to get eliminated for Regional Rail anyway, but the Fenway, Lower Allston, Ruggles-Nubian, and Huntington Avenue segments are going to be brutal. (Same with trying to get high frequencies on Washington Street.) It's why I went aggressive on grade separation for all high-frequency segments.
This is really useful feedback. (And provides an occasion to remark that I try always to be mindful in discussion with you and others that, like, this is your frickin' job, whereas it is decidedly not something I have any training or professional experience in.)

The J -- for what it's worth -- I think could be broken up into shorter segments to mitigate some of your concerns. I also don't know that the J needs to run all the way to Nubian -- although I think Nubian-Ruggles-Longwood is going to be inundated with buses such that there will need to be dedicated transit infrastructure anyway.

I should also point out that I left the Q's frequency somewhat undefined -- anywhere from 6 tph to 15 tph. Some of that was dependent on how high we could push the frequency through the Longwood Transitway.

I don't love it, but we could reimagine this corner of the map more simply:

[insert map: J to West Station, Q runs on its own through Longwood Transitway, and maybe adds a branch to Brigham Circle?, A runs via Packards Corner -- I need to turn in soon, so I'll post a map of this later!]

This could reduce the frequency requirements through Longwood while still maintaining increased connectivity across the system.

The junction between the Union Square/Porter and Grand Junction lines is going to be nasty, since you have to separate the Grand Junction tracks entirely from the Fitchburg Line. No way you're getting a transfer station in there. Inbound, it will be a steep continuous grade from the duck-under of the Fitchburg Line until it meets the Medford Branch tracks.
Yeah, as you can see above I'm terrified of that divergence! I hadn't considered swinging wide to make room for a double flyover, but yeah, that could work. A stop at McGrath Hwy is nice, insofar as, boy howdy, would it be nice to find a transfer location between the erstwhile Urban Ring and the radial Green Line branches, but that's in part why I've run the G down Grand Junction -- to provide a transfer point at that station at Cambridge St.

(On a similar note: I do like the idea of a Porter-Union-Sullivan LRT line, especially if it extended out to Watertown. If we needed to break up the J, we could do a Grand Junction-Sullivan service and a Chelsea-Porter service. That would allow Union to serve as an Urban Ring/Green Line transfer point as well. I see that more as a second or third generation proposal, however.)
Nitpick: you're missing the current Union Square (Somerville) station on your map
Not so much "missing" I believe, as "carelessly misplaced" well to the northwest 😂 Will be fixed in a subsequent edition.
No Mystic Valley Parkway extension???
Not necessarily, though that's another error to correct -- it should be a dashed line like I have at Porter. Despite the level of detail here, I am technically still trying to be agnostic on as much of the system as possible -- you'll notice that I don't commit to a particular Heath-South extension either. Whether ending at Medford/Tufts, Mystic Valley Parkway, West Medford, Winchester, or Woburn, that branch is mostly going to be the same either way: 7.5 tph into the core, with potential additional short-turns at Brattle Loop depending on demand.

~~~

Overall: Like I said, one of the things I really like about your design is how much of my system can map into the infrastructure you've outlined. A radial Seaport-Back Bay-Longwood-Harvard service would fit right in with what you've proposed, for example. In a lot of ways, your design is largely what I would propose if I believed we could get away with the larger capital projects you call for. And maybe we can! I certainly would love it. Failing that, though, I think we still can come tantalizingly close even with much more modest investments; if we can articulate both a "minimum viable build" and a "full buildout" for a Reconfigured Green Line, I think that could become very compelling.
 
Yeah, I admit I'm biased on this point -- the colors are just straight up fun and aesthetic to me. To be fair, there are systems that distinguish between a "light green" and a "dark green" line (which is really what I'm going for here, "emerald" just being another word for "green" that also has local connections thanks to the Emerald Necklace). I think the colors vs numbers issue is moot here -- we already are at a system design that calls for letters.

From a placemaking perspective, we could still call the A, B, C, and R, the "Green Line", and just have a different shade of green on the map. To @737900er's point, a full-build LRT network quickly turns into "the Green Monster", which is why -- as you note -- I give different colors for the routes that bypass Park St, and likewise think a different color is justifiable for the "Kenmore Division".

I think there's benefits to having color differentiation between the services that hit different trunks/nodes, but I also do like the idea of the colors potentially being at least somewhat linked (particularly to distinguish them from the HRT and BRT/bus services). The existing Green Line, as you noted, already has branch lettering, which is finally becoming actually relevant on the north/east end of the system given the GLX branches, it makes much more sense to extend that system. (If necessary, numbered branches might be more appropriate for future branched HRT lines, because the branching ability of those lines means there'll be fewer of them.) I do wonder to what extent it would be necessary to dump the green color altogether (at least for anything other than the Kenmore Division) given the ingrained understanding of the current Green Line moniker (I've witnessed more than a few Bostonians, like other non-New Yorkers, utterly lost by means of misunderstanding of the NYC subway's use of colors versus other places', including Boston's), or if it'd be sufficient to heavily emphasize the service designations rather than the colors, at least for LRT.

On the topic of colors, since playing with them is indeed fun, I'd probably chuck the emerald in the bin (or maybe keep it but not use it as an official reference) and go whole-hog on referring to services rather than lines, and anything LRT hitting Park (or, well, the Park-GC area, such as Brattle Loop turns, 'cause I like my avatar there being green) would use green as its trunk color. The Aqua line is fine (though I'd call it Turquoise or Cyan if it were operationalized, because otherwise a bunch of tourists are going to get lost trying to find the Aquarium), but only so long as that BRT service isn't that teal color because it's way too similar. I'd personally turn that into the gold line (give the shiny-metal colors to the BRT stuff, plus at least it wouldn't be making a real transit line a similar color to the buses, and pick something else for that J service. (Brown and pink are readily available, while one of magenta or violet, as well as indigo, could be doable if the CR was kept to a single standard - the equipment and some of the signage has a tendency to look a lot pinker than the digital versions.)
 
This is what it looks like today, plus hypothetical LRT tracks coming down from Sullivan:

View attachment 30415
That said, if someone can come up with a reasonable design that incorporates a flying junction, I'm all for it. I just haven't been able to figure it out myself yet.
Here's my idea for a flyover connection. I'm guessing it's doable, but obviously a detailed design would be needed to verify the grades, etc.:
52137553898_a266a36b36_b.jpg


Also here's my idea for the intersection of GLX (to Union Sq) and a Grand Junction LRV line, an effort to squeeze everything into pretty much the existing ROW.:
52136925773_b8d699aa25_h.jpg
 
Also here's my idea for the intersection of GLX (to Union Sq) and a Grand Junction LRV line, an effort to squeeze everything into pretty much the existing ROW.:

How high is the existing GLX viaduct there? It appears to still be ascending, so I'd be concerned about whether it's possible to cross the Fitchburg ROW there with sufficient clearance for the CR trains (especially without compromising potential future electrification), and the north side of the Fitchburg ROW is so close to the GLX incline that there's no room to get back down to the existing ramp's height from a CR-clearance height.
 
Why are the Fitchburg tracks to the west of the GL? I assume to avoid a LRT/ heavy rail grade crossing with the GJ? Seems wicked short sighted since even if its possible to to tie into the GJ here (not convinced), best case its looking like Boylston 2.0 except on one heck of a grade. Was time separation like the NJ river line off the table?
 
Why are the Fitchburg tracks to the west of the GL? I assume to avoid a LRT/ heavy rail grade crossing with the GJ? Seems wicked short sighted since even if its possible to to tie into the GJ here (not convinced), best case its looking like Boylston 2.0 except on one heck of a grade. Was time separation like the NJ river line off the table?

Unless there's a plan on the table for GJ, you don't add tens of millions to the cost of your current project to accommodate it. LRT on the GJ exists only as a fever dream on this message board - they're having a hard enough time getting a bike path built there and it requires major reconfiguration on the Boston end to even link to anything.
 
The NJ River Line is otherwise freight-only, so freight runs at night and light rail during the day. This is absolutely not an option with a heavily-used commuter rail line.
 
As @Riverside mentioned, I did some work on this earlier this year, in this Google map. I basically threw up my hands at getting from GJ southbound towards Lechmere, but I did figure out how to do the GJ/GLX Union Square Branch split without modifying the McGrath Highway bridge.
 
Can someone smarter than me explain the pitfalls of putting a single track surface line up Linwood Street and turning onto the SB track to Lechmere using Poplar street? Could eminent domain make the curve less tight?
 

Back
Top