Shower thought: In a world where the "parallel" model is implemented and all Kenmore trains terminate at Park St, how feasible would it be to build a pair of crossovers south of Park St and convert it into a double stub-end terminal? Like this: View attachment 42681
While I believe the loop is still more efficient as a terminal, this adds flexibility to hold trains in the terminal for headway control. Someone with more knowledge in operations can correct me if I'm wrong.
First, my vague understanding is that these days the gold standard for terminals is a crossover leading into a pair of tracks (ideally with a center platform) followed by another crossover and then a set of tail tracks long enough for at least one trainset each. This provides storage space for sets that either need to be temporarily taken out of service, or for staging to prepare for peak periods. This also provides a high level of flexibility -- dispatchers can send terminating trains into either platform track and subsequently either back into service through a reverse move, or out of service to either of the tail tracks. (And the center platform means waiting passengers can easily access departing trains from either track.)
It is true that it takes a little longer to turn the train because the driver needs to switch ends, but having a brief break from revenue service isn't always a bad thing.
A loop, on the other hand, basically means that the train stays in revenue operation for the entirety of the turn, effectively doubling the length of the line. No opportunity to hold trains, no opportunity to let trains leapfrog each other... if three B's arrive into Park Street back to back, then you're gonna have three B's departing back to back as well.
So, yes, I would feel better about the inner tracks at Park Street if they had a crossover south of the station and could be more like a stub-end terminal. That would allow at least some flexibility in holding and reordering trains. The challenge I see is that the inner tracks lack a central platform; it'll be pretty obnoxious for a commuter to be on the western platform and discover that their train is actually departing from the eastern. I was looking at old plans the other night to get a sense of whether tracks could be rearranged to create a center platform, and I wasn't super optimistic about doing so unless the footprint of the station were expanded.
I actually really like the locations of Arlington and Copley stations, having used Arlington extensively to get to/from Public Garden myself. An argument can also be made that Copley station serves the Back Bay area better than the namesake station on OL (and the Huntington subway). For these reasons, I raise eyebrows at any proposal that reduces Copley and Arlington to a shuttle to Park St, but it may be just me.
Oh, I agree -- it's one of those "extraordinary claims require extraordinary evidence" situations, where you'd need to make a really convincing case that it would be worthwhile.
Another potential concern for a 4-track Back Bay subway is the width under Marginal St and the complexity of Bay Village. However, I think there's an alternative: A Stuart St subway, which was mentioned by Van in some of the discussions you linked earlier, in addition to a Marginal St subway. Not sure if it's a good idea for both subways to co-exist given complications with OL transfers and alignment near Pleasant St portal, but I suppose that in a real-world proposal, both subways will be studied.
I took a look at this earlier today. My estimate is that a 4-track ROW (assuming side-by-side tracks) would need to be 50 feet wide.
Huntington - Dartmouth: should be fine
Dartmouth - Clarendon: this is an area of concern since the width would need to increase significantly to fit platforms. Possibly this could be ameliorated by off-set platforms, such that the width is never greater than 4 tracks + 1 platform
Clarendon - Berkeley: would need to build under at least one building, but the 2-track alt needs to do the same
Berkeley - Arlington: looks like there is amply more than 50 feet between the edges of Cortes St and the Mass Pike; you'll need to fuss with the on-ramp, but again that's the same as the 2-track alt
Arlington - Tremont: here we have some constraint; however, we could probably stack our tunnels here, 2 over 2
As for Bay Village: if the Back Bay subway is quad-tracked and connected to Kenmore, I don't think we need to create a fancy junction here. The Huntington subway can curve up north under Tremont to reach the Pleasant Street Portal, while the Kenmore subway stays straightaway underneath Marginal, and we call it a day. (If we really felt a need to add more connections, I suspect we could build the necessary connections between Dartmouth and Tremont Streets.)
I think Stuart Street is a useful alternative to keep under consideration, but I suspect it will lose points when it comes time to design the interface with the old Tremont Street Subway; you'd be unable to leverage the flying interchange available at Pleasant Street, which then opens up a can of worms.
Yet another shower thought: Does there exist a design around Copley that enables three - or even four - of the routings below?
I think it depends on whether you are looking at flat junctions or flying junctions. If you're talking flat junctions, then it's easy-peasy of course. If we're talking about flying junctions, I think it'll be very hard. A flying junction requires both depth and width, neither of which are in large supply in that area.
I tried sketching this out (reminded me of the headache it was to sketch out the Bay Village junction), but eventually gave up. Maybe there's something possible? (Like maybe you could do something with flat splits/crossovers between tracks going in the same direction, with grade separation between eastbound and westbound traffic?)
So, yes, I would feel better about the inner tracks at Park Street if they had a crossover south of the station and could be more like a stub-end terminal. That would allow at least some flexibility in holding and reordering trains. The challenge I see is that the inner tracks lack a central platform; it'll be pretty obnoxious for a commuter to be on the western platform and discover that their train is actually departing from the eastern. I was looking at old plans the other night to get a sense of whether tracks could be rearranged to create a center platform, and I wasn't super optimistic about doing so unless the footprint of the station were expanded.
A mitigation is to use different tracks for different branches. For example, all B trains leave from track 3, and all C trains leave from track 2. Even though passengers whose destinations are between Arlington and Kenmore still need to figure out which platform to go to, this at least reduces some degrees of randomness.
First, my vague understanding is that these days the gold standard for terminals is a crossover leading into a pair of tracks (ideally with a center platform) followed by another crossover and then a set of tail tracks long enough for at least one trainset each. This provides storage space for sets that either need to be temporarily taken out of service, or for staging to prepare for peak periods. This also provides a high level of flexibility -- dispatchers can send terminating trains into either platform track and subsequently either back into service through a reverse move, or out of service to either of the tail tracks. (And the center platform means waiting passengers can easily access departing trains from either track.)
Also, it's not just a terminal. There'd still be passengers from the Boylston/Kenmore branch (center-isolated) with destinations beyond Park Street. Right now, or even in a Boylston-center, Tremont-outer paradigm, it's a simple cross-platform transfer. You'd potentially (ahem*Red Sox games*ahem) be adding a lot of transfers through the stairs, with extra added confusion because that wouldn't consistently be the case, depending on which fence track you happened to come in on.
A mitigation is to use different tracks for different branches. For example, all B trains leave from track 3, and all C trains leave from track 2. Even though passengers whose destinations are between Arlington and Kenmore still need to figure out which platform to go to, this at least reduces some degrees of randomness.
This is true, though it creates tension with the benefit of dispatcher flexibility -- the more predictable you make it for passengers, the less flexibility you give the dispatcher.
Yeah, I agree with @Brattle Loop -- those underpasses seem more designed for "oops I went to the wrong track" rather than "this is the main thoroughfare for transfers".
Of course, I should note: despite all my doomsdaying, the loops at Park St and Government Center have done fine turning 1 or 2 branches for years. I'm not sure Park has ever done two (modern) services at once, but Government Center is turning two right now (albeit at reduced frequencies). If the Kenmore Division is just the B and C, then probably Park St will be fine as is -- maybe with a crossover added just to provide a little more flexibility for when it's needed. Adding a third branch is when questions get raised in my mind.
Of course, I should note: despite all my doomsdaying, the loops at Park St and Government Center have done fine turning 1 or 2 branches for years. I'm not sure Park has ever done two (modern) services at once, but Government Center is turning two right now (albeit at reduced frequencies). If the Kenmore Division is just the B and C, then probably Park St will be fine as is -- maybe with a crossover added just to provide a little more flexibility for when it's needed. Adding a third branch is when questions get raised in my mind.
GC loop handled B and D both for years with Park as a relief valve. One loop can handle the volume, the problem was the lack of sorting and inability to hold a car other than on the loop. Whereas with Park now there's tons of room to hold looping cars, because once they're past the T3-to-T4 switch north of Boylston they're no longer an impediment to traffic flow (unlike at GC where they hog the platform and block traffic).
FWIW, we've seen several GL weekend shutdowns north of Government Center in the last 1.5 years. Sometimes they short-turn all trains at GC, and typically the entire GL's headways get screwed up. Sometimes they short-turn some branches at Park (typically branches that use track 2) and some at GC, and it doesn't seem to be much better.
As a reminder, it's not so much about the number of branches and more about the TPH. Given that in most "complete" GL Reconfiguration proposals, the Kenmore division's capacity is completely filled up between Harvard*, Oak Square, Boston College and Cleveland Circle trains, this does raise concerns about reliability if all trains have Park as their only turning point.
* Harvard trains don't need to enter the Central Subway depending on design further west (and preferably shouldn't). But even if you take out Harvard, you'll probably want to use those Harvard slots on the other three branches to improve headways.
GC loop handled B and D both for years with Park as a relief valve. One loop can handle the volume, the problem was the lack of sorting and inability to hold a car other than on the loop.
Yeah, and to be clear I agree that each loop can handle the 20 tph historically brought by two branches. Three branches would, I think, tip us a bit above 20 tph — unclear how much, but that’s where I think there’s some question.
The lack of this, is what I don't like about the crossover idea. Why not have C-trains cross over to Backbay, while B-trains stay on the Boylston St. subway? And likewise, we could have E-trains cross-over to Boylston St., while D-trains continue on the Seaport branch routing. If you are coming from Allston, you can do a cross platform transfer at Kenmore in order to get to the Seaport, and somebody coming from Coolidge Corner can do a cross platform transfer at Kenmore for downtown service. That is a far easier transfer than having to do the State Street style multi-tunnel and mezzanine marathon to transfer from new Back Bay to new Copley.
The lack of this, is what I don't like about the crossover idea. Why not have C-trains cross over to Backbay, while B-trains stay on the Boylston St. subway? And likewise, we could have E-trains cross-over to Boylston St., while D-trains continue on the Seaport branch routing. If you are coming from Allston, you can do a cross platform transfer at Kenmore in order to get to the Seaport, and somebody coming from Coolidge Corner can do a cross platform transfer at Kenmore for downtown service. That is a far easier transfer than having to do the State Street style multi-tunnel and mezzanine marathon to transfer from new Back Bay to be Copley.
To be fair, the traditional parallel model also lacks any means to send Kenmore trains to Seaport (although it can have Huntington trains to Boylston and Park St via the Pleasant St incline).
To me, this thought exercise highlights the potential need for a Kenmore-Bay Village connection so that you can send some trains from Kenmore to Seaport, even in a parallel model. Aside from connectivity as you mentioned, another big reason is to avoid capacity constraints at the Park St loop, which will likely become the bottleneck for the Kenmore division more so than the Central Subway itself.
Two challenges in doing that:
Engineering: How to enable both Kenmore-Bay Village and Kenmore-Arlington, preferably via a flying junction?
Capacity of the Back Bay subway: Ideally, we would like as much capacity as possible on the Huntington lines through Prudential and LMA, in a way that's not hindered by any Kenmore trains. This is where quad-tracking Back Bay-Bay Village may be a sensible idea, even in a parallel model.
Ideally, the branches could look something like this:
A: Oak Square - Kenmore - Bay Village - Seaport
B/C: Boston College / Cleveland Circle - Kenmore - Arlington - Park St
D: Riverside - Prudential - Bay Village - Government Center and north
N/E: Needham Junction / Heath St - Prudential - Bay Village - Seaport
F: Nubian - Bay Village - Government Center and north
Comparing to the parallel model, any additional trains into Back Bay subway fed by the Kenmore division are absorbed by the two additional tracks from the west. Once you get past Bay Village, the two reverse branches (GC and Seaport) won't be bottlenecks, especially not Seaport (GC still has Nubian trains). This also allows higher frequency to Seaport than you could otherwise get, albeit at the cost of interlining with street-running branches.
To be fair, the traditional parallel model also lacks any means to send Kenmore trains to Seaport (although it can have Huntington trains to Boylston and Park St via the Pleasant St incline).
To me, this thought exercise highlights the potential need for a Kenmore-Bay Village connection so that you can send some trains from Kenmore to Seaport, even in a parallel model. Aside from connectivity as you mentioned, another big reason is to avoid capacity constraints at the Park St loop, which will likely become the bottleneck for the Kenmore division more so than the Central Subway itself.
Two challenges in doing that:
Engineering: How to enable both Kenmore-Bay Village and Kenmore-Arlington, preferably via a flying junction?
Capacity of the Back Bay subway: Ideally, we would like as much capacity as possible on the Huntington lines through Prudential and LMA, in a way that's not hindered by any Kenmore trains. This is where quad-tracking Back Bay-Bay Village may be a sensible idea, even in a parallel model.
Ideally, the branches could look something like this:
A: Oak Square - Kenmore - Bay Village - Seaport
B/C: Boston College / Cleveland Circle - Kenmore - Arlington - Park St
D: Riverside - Prudential - Bay Village - Government Center and north
N/E: Needham Junction / Heath St - Prudential - Bay Village - Seaport
F: Nubian - Bay Village - Government Center and north
Comparing to the parallel model, any additional trains into Back Bay subway fed by the Kenmore division are absorbed by the two additional tracks from the west. Once you get past Bay Village, the two reverse branches (GC and Seaport) won't be bottlenecks, especially not Seaport (GC still has Nubian trains). This also allows higher frequency to Seaport than you could otherwise get, albeit at the cost of interlining with street-running branches.
Rule of thumb in any sort of major infrastructure network redesign -- design for maximum flexibility in the network connections.
Only a fool thinks they really know what demand patterns on the network will look like in 30, 40, 50 years from now, but the infrastructure can only be reconfigured within the planned flexibility. So always build in reconfiguration options.
Quick sketch and not too much thought into it, but what about using SL Phase 3 planned tunnel under Charles st? You have the room to make a flying junction on the Central Subway with the PO sq provisions. Then you avoid the Back-Bay - Bay Village subway
The lack of a Kenmore <> Seaport OSR (and the kludge of a 2SR via a transfer at Copley/Back Bay) has irritated me for years. The ability to reverse-branch the Huntington Division to serve both Downtown and the Seaport is really quite useful, but is much harder to replicate from the Kenmore Division.
The lack of this, is what I don't like about the crossover idea. Why not have C-trains cross over to Backbay, while B-trains stay on the Boylston St. subway? And likewise, we could have E-trains cross-over to Boylston St., while D-trains continue on the Seaport branch routing. If you are coming from Allston, you can do a cross platform transfer at Kenmore in order to get to the Seaport, and somebody coming from Coolidge Corner can do a cross platform transfer at Kenmore for downtown service. That is a far easier transfer than having to do the State Street style multi-tunnel and mezzanine marathon to transfer from new Back Bay to new Copley.
I like the elegance of this, but I think this level of NYC-style reverse-branching + interlining would create a dispatching and reliability nightmare. The reverse-branch from Huntington works because of the quad-track running into Park St. To achieve the same level of isolation on the Seaport, you'd need to quad-track a much, much longer stretch of track.
Ideally, the branches could look something like this:
A: Oak Square - Kenmore - Bay Village - Seaport
B/C: Boston College / Cleveland Circle - Kenmore - Arlington - Park St
D: Riverside - Prudential - Bay Village - Government Center and north
N/E: Needham Junction / Heath St - Prudential - Bay Village - Seaport
F: Nubian - Bay Village - Government Center and north
Comparing to the parallel model, any additional trains into Back Bay subway fed by the Kenmore division are absorbed by the two additional tracks from the west. Once you get past Bay Village, the two reverse branches (GC and Seaport) won't be bottlenecks, especially not Seaport (GC still has Nubian trains). This also allows higher frequency to Seaport than you could otherwise get, albeit at the cost of interlining with street-running branches.
From a crayoning perspective, I really love the idea of the alternating reverse branches. But I just don't think it's a good idea to interline the B or C with new modern LRT builds. Maybe with aggressive stop consolidation and elimination of grade crossings it could be okay? But they're just always going to be providing a fundamentally different service than the rest of the network.
Quick sketch and not too much thought into it, but what about using SL Phase 3 planned tunnel under Charles st? You have the room to make a flying junction on the Central Subway with the PO sq provisions. Then you avoid the Back-Bay - Bay Village subway View attachment 42726
S/O Van for the map, one of my favorites
What a trip the Silver Line Phase III 2005 EIS is. A lot of the discussion on the Charles Street alignment focuses on the portal location, and on how it would interface with the double-stacked Silver Line tunnels underneath the Boylston Street Subway. To me the piece that's most immediately interesting are the geological profiles. (I don't really know how to interpret them, but they're interesting.) Some excerpts:
The alignments compared against the original shoreline of the Shawmut Peninsula (see also here):
The soil profile under Boylston (and Essex) Streets:
The soil profile under Charles St:
The EIS called for double-stacked tunnels under Charles Street, mined as opposed to cut-and-cover.
Key differences between all that and using Charles Street for a Kenmore <> Seaport Green Line tunnel:
Silver Line tunnel needed to rise to a portal around Tufts Medical Center; a Green Line tunnel would likely continue south to connect to a tunnel under Marginal Road
Silver Line tunnel needed to duck down a full two levels beneath the Green Line; a Green Line flying junction would not need to be as deep, particularly if the eastbound track branched at grade to the outside rather than doing a duckunder on the inside (see diagram below)
I don't really know enough about geology or tunneling to assess all of the above, so my initial thought is, maybe?
The second question then becomes how do you cross Bay Village to get to a Marginal Road subway to the Seaport? There's already gonna be an awful lot going on there. Very very rough diagram based on one of my designs for the Bay Village junction:
(Unmarked on this diagram are provisions for a Park <> Seaport leg of the wye, which would travel in stacked tunnels between the green and cyan tracks in this diagram.)
Hmm. This may be somewhat doable. Would need to sketch further...
So, my initial thoughts are that the Charles option has the appeal of an (easier) flying junction, and of maintaining full service to Copley and Arlington. Not that those are trivial, but aside from them I don't immediately see a huge benefit of it over a quad-tracked Back Bay subway.
And then comes the question of how to get it across Bay Village. But that actually raises an interesting question: Is our objective to create a Kenmore <> Seaport link, or to provide relief to Park St? Because a stub-end terminal under Charles, abutting a Bay Village station would give you someplace other than Park St to turn trains. And with transfers available to northbound Green Line, eastbound Seaport, and (with a bit of a walk, though not really more than at Park-DTX) northbound Orange Line, it wouldn't, you know, be the worst place to terminate something like a resurrected A Line.
The other thing to consider along these lines is whether the Public Garden incline could be used as a jump-off point for a tunnel going someplace else. Such as, for example, cutting under the Common (I know, I know, I know) to a new station at Park Street Under.
A little messy. Platforms at Bay Village are not marked but they're there (see original post). The criss-cross in the wye is also pretty rough, but the measurements should be the same as in my original post, so they should fit. There are two paths to Nubian included here -- one with a portal at the original Pleasant Street location, and the other with a duckunder the Mass Pike; this aspect I'm a little less sure about, because IIRC there was something dicey about the spacing/curvature of the duckunder option -- in my previous design, the Nubian subway utilized the west bellmouth at Pleasant, and I definitely had a reason for doing that. But the surface portal alt should be fine (and if I really wanted a separate subway under Albany, I could branch off of the Marginal Road subway somewhere east of here regardless).
And yeah -- moving the double-stack platforms (used for Park <> BBY service) to the western end of the station opens them up for a flying junction to a stacked pair of tunnels under Charles. So that actually works okay.
In theory, capacity shouldn't be a problem. Coming from Huntington, you'll have, let's say, 15 tph going BBY <> Park, and 15 tph going BBY <> BOS. That means that the stacked platforms at Bay Village will only be seeing 15 tph, which should provide ample capacity for at least one line diverted from Arlington, maybe even two. And then the Marginal Road subway will just be receiving the 15 tph from Huntington, meaning it also will have spare capacity to absorb a line from Arlington, although I think merging schedules will be complicated.
So... there we are. A concept for a Kenmore <> Seaport OSR -- and a completion of the partial Aldgate junction around Boylston to boot. I continue to maintain that the interlining here will ultimately create more problems than it solves, but it's still a useful piece of the puzzle. Props to you @BosMaineiac for having the lightbulb moment!
And @Teban54, while this design doesn't technically provide for a OSR between Huntington Ave and Arlington station, I believe this design would allow for all combinations of Kenmore & Huntington to Downtown & Seaport.
Sketching up the diagram on Friday evening made me realize I hadn't actually posted my current (and hopefully semi-definitive) iteration of the proposal for Bay Village. So here it is:
I started my blog so I would have someplace other than ArchBoston posts to put excessive amounts of detail. So, if you are interested in excessive amounts of detail, take a gander here.
A few quick bullets:
Tremont station is a little further north than it should be -- imagine it being scooched down to the southwest by ~15 feet
(Despite my effort at detail, the diagram is still mostly illustrative, and it took forever to draw the flying junction right, so I didn't want to redraw)
As outlined at the link, there is enough room at both ends to fit the necessary grades
I'm not entirely sure why I think/thought the Surface Line needs to have a question-mark alignment going through the park -- looking at it again today, I think it should be able to run up directly parallel to Shawmut Ave? But again, it's illustrative, not maximally precise
I am not at all certain about the spacing of a portal around Hudson Street
Yes, this design does not prioritize Park <> Seaport service, as I've written about previously.
It can still be accommodated through the Surface Line
And I believe the flying junction at Marginal & Tremont could actually be built such that a third leg of the wye could be built at a later date if desired
Concourse and headhouse placement are particularly illustrative and imprecise
The key to this whole shebang is ultimately the dual level Tremont station: splitting Park <> Back Bay service vertically is necessary for the flying junction with Seaport <> Back Bay services, and the Nubian services on the lower level need to be well-positioned in order to dive deep under the Mass Pike and Orange Line.
I like the elegance of this, but I think this level of NYC-style reverse-branching + interlining would create a dispatching and reliability nightmare. The reverse-branch from Huntington works because of the quad-track running into Park St. To achieve the same level of isolation on the Seaport, you'd need to quad-track a much, much longer stretch of track.
From a crayoning perspective, I really love the idea of the alternating reverse branches. But I just don't think it's a good idea to interline the B or C with new modern LRT builds. Maybe with aggressive stop consolidation and elimination of grade crossings it could be okay? But they're just always going to be providing a fundamentally different service than the rest of the network.
Speaking for myself, the main reason that now made me support a Kenmore-Seaport connection is for capacity of both Seaport and the entire Kenmore division (A/B/C).
Let's make a handwaving assumption that the Central Subway can support 30 tph, but the Park St inner loop can only support 20. Also suppose you need 10 tph for each of the three Kenmore branches (6 min headways). For reference, today's GL branches have peak frequencies of 6-8 mins.
Actually, I'm not sure.
In theory, I would have expected loops to be fairly efficient terminals due to no need for reversal. But the tightness and resulting speed limits of the Park St loop may make things worse.
The best guesses I can make are from what the T is already doing during the Government Center garage closures. Sometimes, all 4 branches turn at GC. Sometimes, two branches turn at Park St and two at GC. It does appear that whenever all trains run to GC, there are long lines of passengers waiting at GC and irregular headways, which would suggest low reliability, even if not necessarily low capacity. The fact that short-turning two branches at Park even needs to be considered, to me, suggests that 30 tph into Park loop may not be a good idea as a full-time service pattern.
Under these assumptions, it becomes obvious that the extra 10 tph from Kenmore have to go somewhere other than Park. At least the need will be there when a resurrected A branch enters service - Park St may be able to handle the B and C.
Another place that could benefit from higher frequency is - guess what - Seaport. Ideally, in a full-build GLR world, we would like maximum service at all of these three places: Back Bay-Huntington, Government Center (Park St outer) and Seaport, all of which are important destinations. The first two are easy - Back Bay is part of the main trunk, and GC has supplements from Nubian. But Seaport will never see full service as long as the Huntington reverse branching exists.
Running short-turn trains between Seaport and South Station (mayyybe Bay Village) is one option. The other option is connecting two places that both need additional capacity: Kenmore and Seaport.
To be clear, this option has some drawbacks, as you have mentioned. Running streetcar services into Seaport will really compromise the latter's reliability, and as seen from various, frequently made proposals to deinterline the NYC subway, this DeKalb-like service pattern does have potential to really screw things up. The benefits, on the other hand, include:
Killing two capacity issues in one stone
OSR from Kenmore to Seaport
Much better transfer from all Kenmore branches to the Huntington Subway (cross-platform) and Orange Line (at Tufts Medical Center)
At the end of the day, it's a tradeoff between reliability and (capacity + connectivity). I suppose this becomes more of a design choice at that point, one that will be studied more extensively should the proposals ever materialize. The good thing is that it may not be a necessary decision until the A branch becomes a thing.
On an engineering level, i.e. how to connect Kenmore and Seaport:
That's a very smart and elegant design! Swapping BBY->Park to the westernmost track does open up interesting opportunities.
A nice by-product of the design is that what used to be BBY-Park platforms now also have services to Seaport and Kenmore. This enables a few nice transfers:
Same-platform transfers between Park and Seaport, which may be relevant for two-seat rides from, say, Medford/Tufts to Seaport.
Transfers from Nubian (Albany subway) to both Kenmore and Seaport also become much easier, within the same station and at most going up/down one level.
If the Kenmore-Seaport service is run frequently enough, this may completely eliminate the need for a Shawmut Ave platform on the Marginal St subway.
Edit: On a second thought, such a platform may still be necessary to enable transfers from every single Seaport train to northside Orange Line, without backtracking to Back Bay or waiting for more limited Kenmore-Seaport trains.
The only concerns and drawbacks that I could find are: (Here, "Albany" = Nubian subway, "Washington" = Nubian surface)
The elegance of two Park-bound tracks on a single platform can no longer be maintained, as the BBY->Park track is on the westernmost track and Albany->Park track on the easternmost track.
Due to this track reordering, the Washington tracks can no longer use the eastern bellmouth of the Pleasant St portal by itself, as all 4 Tremont tracks need to be fed into the portal in this exact order. Worse still, there doesn't seem to be a good solution to feed the Washington tracks into the portal at all.
The intersection of Seaport->Park and Albany tunnels just south of Tremont station looks dicey. You have Seaport->Park going from (-2) to (-2.5) on a curve, and Albany<->Park from (-3) to (-2.5), while crossing each other within a short distance from the station.
In your blogpost, two analogous tracks do cross each other, but much further apart due to the much gentler curve of the BBY->Park track.
Looks like it might be okay, but probably need to check the grades.
The Shawmut Ave station platform on the Marginal St subway, if we still want one, likely needs to be moved further east to accommodate for elevation changes of the westbound track.
I came up with a solution to address the first two points:
(Any changes are in red. Arrows not in red are added for clarity.)
The key change is making the Albany subway left-running, at least for this part of the route. The tracks can be sorted out south of Mass Pike if right-running is desired for the rest of the subway to Nubian.
Turns out, this solves both problems perfectly, better than I imagined:
Northbound Marginal->Park and Albany->Park tracks are now the western and middle tracks of Tremont Under. This naturally allows for an island platform with all Park-bound services (plus Arlington). Both tracks merge after the station, and feed into Track 2 of the Pleasant St portal (numbered left to right).
Southbound Park->Albany and Park->Washington tracks both come from Track 3 of the portal, and split as soon as possible. The Albany-bound track goes down to (-2.5) and becomes the easternmost track of Tremont Under, forming another island platform with the opposite track (or a side platform). The Washington-bound track rises to street level together with its opposite track.
Northbound Washington->Park track feeds into Track 4.
Basically, Marginal services take Tracks 1 & 2, Albany takes 2 & 3, and Washington takes 3 & 4.
The only thing that irks me is the asymmetry of timing of merges in each direction, but that's very minor in the grand scheme of things.
I'm not entirely sure why I think/thought the Surface Line needs to have a question-mark alignment going through the park -- looking at it again today, I think it should be able to run up directly parallel to Shawmut Ave? But again, it's illustrative, not maximally precise
I found some historical photos of tracks when they were still in use. Looks like the eastern branch (towards City Point) was already very close to Shawmut Ave, so I think that's definitely doable.
But that actually raises an interesting question: Is our objective to create a Kenmore <> Seaport link, or to provide relief to Park St? Because a stub-end terminal under Charles, abutting a Bay Village station would give you someplace other than Park St to turn trains. And with transfers available to northbound Green Line, eastbound Seaport, and (with a bit of a walk, though not really more than at Park-DTX) northbound Orange Line, it wouldn't, you know, be the worst place to terminate something like a resurrected A Line.
The other thing to consider along these lines is whether the Public Garden incline could be used as a jump-off point for a tunnel going someplace else. Such as, for example, cutting under the Common (I know, I know, I know) to a new station at Park Street Under.
This is actually a very innovative idea! It's true that the most immediate need for another terminal out of Kenmore is to provide relief to the Park St loop. So in theory, you could extend the line somewhere else that's neither Park nor Seaport, perhaps not even Bay Village.
A 1-minute Crazy Transit Pitch I got out of it:
The basic idea is a partial Stuart St subway, which had been seen as a "backup option" in place of a Marginal St subway. The street is wide enough, and further enough from the messes of Boylston and Chinatown, to avoid SL Phase 3's pitfalls. A station at the Washington St intersection allows transfers with the Orange Line's Tufts Medical Center station.
The line meets the Marginal St subway again at Chinatown Gate (with or without non-revenue track connections). After that, a few options exist:
Terminate around South Station, if there's space.
Open up to a portal around or north of South Station, becoming @Riverside 's Rose Line running at-grade along the Greenway.
Or turning towards Post Office Square, either at grade or underground, essentially creating a resurrected "Post Office Sq branch" - which was ironically the intended use of the Arlington-Boylston bellmouth. Either terminate there, or go up one more stop to meet the Orange Line at State.
If we're really dreaming... Go up even further north to become a second downtown LRV trunk with streetcars to the north, absorbing the likes of T7/93 and T111!
This is probably too much to expect from a line whose other end consists of streetcars, one that's meant to be Park St relief to begin with (although the streetcar nature was also why I was more open to street-running). But still, it's possible, and may even be one of the most reasonable pitches for an alternative route from the Arlington-Boylston bellmouth.
On a second thought, you can even make this a legitimate alternative pitch to connect Kenmore and Seaport, without interfering with the Bay Village interchange. It's more expensive due to tunneling under Stuart St, but not much longer than a full Charles St subway (only about 700 ft more), and has the advantage of a much faster and less roundabout route than Bay Village.
Shower thought: In a world where the "parallel" model is implemented and all Kenmore trains terminate at Park St, how feasible would it be to build a pair of crossovers south of Park St and convert it into a double stub-end terminal? Like this: View attachment 42681
While I believe the loop is still more efficient as a terminal, this adds flexibility to hold trains in the terminal for headway control. Someone with more knowledge in operations can correct me if I'm wrong.
Some updates on efficiency of loop terminals. I read up a bit more on the NYC subway, which has several loops on the IRT that are still in use today, such as City Hall and South Ferry.
Turns out, the outer loop at South Ferry was a major bottleneck for the 1 train, as explained here, and was much less efficient than modern stub-end terminals there like Hudson Yards and 96 St on the Second Avenue Subway. Its replacement with a new stub-end terminal allowed the 1 to increase capacity. (More posts on this topic include this and this.)
With the Park St loop being even tighter than South Ferry, I think it will likely become a bottleneck for the Kenmore division (though luckily it doesn't have platforms on the curve itself unlike South Ferry). I'm not sure if the conversion to a stub-end terminal as proposed above will be feasible, and if doing so will allow maximum capacity to support three Kenmore branches. So this does highlight the potential need for a reverse branch from the Central Subway.
Edit:
Upon further reading, looks like the 6 and <6> trains can achieve a combined 24-30 tph using the City Hall loop. So maybe keeping Park St as a loop is okay?
Tangentially related to this thread, but the following piece from a Reddit user is useful for explaining bottlenecks on the GLX and to keep in mind when designing northside branches.
Science Park to North Station is a bottleneck in both directions because of the steep hill and the way the signals are set up. If there are two trains in a row leaving North Station heading towards Science Park, the second train has to wait at the turnback, before the uphill, until the first train completely clears the Science Park platform. That easily adds a few minutes of waiting. A delay in the subway can often result in a queue of trains waiting to go up.
On the way down to North Station, a train at Science Park has a red signal and waits on the platform until the train ahead of it either clears the turnback, or reaches the North Station platform (not exactly sure which).
Medford and Union Square are only two stub end tracks each, which is nowhere near enough capacity to serve two lines, at least with how efficiently things are run now. Even with only one line there are times where you'll be waiting outside the station because trains are occupying both platforms.
Edit: Before the Medford line opened, there was a brief time where both the E trains and half of the D trains were running out of Union Square, and that was busy enough. Crews that started at Riverside looped Government Center, while crews that started at Union did the whole trip to Riverside. Obviously that resulted in weird headways but I think the only reason it was done was because sending all the D trains to Union would've resulted in way too much traffic.
TL;DR: The Science Park-North Station bottleneck, as well as both terminals, limit capacity on GLX to not much more than what we're already seeing today.
Perhaps that means my idea of an interlined Urban Ring that goes through Lechmere isn't as crazy as once thought? The route can probably run as much TPH as the current B+C branches without compromising GLX. On the other hand, this means Green Line branches to Grand Junction and Chelsea become major question marks. An ultra-long extension to Waltham may also become a problem if you have to run those trains all the way to, say, Riverside.
Tangentially related to this thread, but the following piece from a Reddit user is useful for explaining bottlenecks on the GLX and to keep in mind when designing northside branches.
TL;DR: The Science Park-North Station bottleneck, as well as both terminals, limit capacity on GLX to not much more than what we're already seeing today.
Perhaps that means my idea of an interlined Urban Ring that goes through Lechmere isn't as crazy as once thought? The route can probably run as much TPH as the current B+C branches without compromising GLX. On the other hand, this means Green Line branches to Grand Junction and Chelsea become major question marks. An ultra-long extension to Waltham may also become a problem if you have to run those trains all the way to, say, Riverside.
On a second thought a day later, the Science Park bottleneck seems like purely a signal problem that should be fixable with a better signaling system, allowing multiple trains to be on the Science Park-North Station stretch at a time. Just climbing up the ramp itself shouldn't cause much troubles.
While this won't let you easily increase service on the Union Sq and Medford/Tufts branches until the stub ends get resolved (i.e. lines extended), it should still allow new branches.
