Green Line Reconfiguration

[Riverside]

Happy Easter, happy Passover, happy Marathon Weekend. Here's some assorted noodling on LRT in Allston and Brighton.

With the redevelopment of Beacon Park has come proposals for a West Station. With those proposals have come suggestions to extend an LRT branch through West Station beyond to Harvard. See for example Vanshnookenraggen's post on the topic.

These ideas interact unsurprisingly with proposals for LRT on the Grand Junction, which has been tossed around for a couple of decades now. The current Grand Junction ROW points at its western end to the site of the proposed West Station, though many of us here have suggested adding a new leg to that junction to enable Grand Junction trains to run to Kenmore. (This dovetails with larger interest in better serving the Kendall-Longwood corridor.)

Finally, there is perennial interest in providing LRT service along the corridor of the 57 bus, the former route of the "A Line". In most proposals I've seen tossed around for this, the original ROW is reactivated, branching off Comm Ave at Packard's Corner.

However, what if this Brighton Branch were fed from West Station rather than Packard's Corner?

Some of this is necessarily handwavy because the exact location and alignment through Beacon Park is only vaguely defined. However, in general we're talking about ~2,100 feet of new street-level (hopefully reservation) track, as opposed to ~3,500 feet via Brighton Ave.

One advantage of accessing the 57 corridor via West Station is the creation of a one-seat ride from Brighton to employment centers in Kendall (and potentially beyond, depending where Grand Junction services travel to on the north end).

An additional advantage is that it enables Brighton trains to avoid the Central Subway, where capacity will be at a premium. (Assume that the B and C will retain their one-seat rides into downtown; further assume that a branch to Nubian will also take up a slot, and that we can reroute the D, E, and Needham branches... somewhere. That means you've got 1 or maybe 2 slots left in the Central Subway, if we assume similar capacity to today. Just in this post here, I've already mentioned proposals for three new branches -- Brighton, Harvard, and Grand Junction.)

One downside is that this Gold Line service would not provide access to Kenmore. This is remedied via upgrades to the 57 bus itself. Assuming an LRT line is built, we may hope that it will use dedicated lanes which could then be leveraged by the bus as well. Additionally, on Commonwealth near BU, the current Green Line reservation would be made available for conversion to bus lane if the Green Line subway is extended from Kenmore west. That would leave about 3,000-4,000 feet of Comm Ave between Packard's Corner and the new portal that would either require new bus lanes, or a rebuild of the reservation to serve both LRT and BRT. An "SL57" line could then maintain the current one-seat ride to Kenmore, while also increasing frequencies along the overlapping corridor.

If we want to move into Crazy Transit Pitches, there's one step further we could take. As you can see, Comm Ave and West Station are separated by a mere 1,500 feet, or about four city blocks. With careful design and planning, the LRT ROW in Beacon Park could be connected to Comm Ave, enabling the B Line to be rerouted north.

This reroute loses Green Line service to Amory St and Babcock St, but gains better connections to services to Harvard and Kendall. Amory and Babcock would retain SL57 service.

In addition to providing one-seat connectivity from West Station to a much larger swath of Allston, Brighton, and Brookline, this connection between Commonwealth and West Station would also enable Green Line service on Commonwealth to be supplemented by Gold Line service from Kendall (not shown on the map above).

That supplementary service in turn could enable a greater number of branches to feed into the Central Subway: pre-covid, the Central Subway handled 40 tph; historically each Green Line branch has seen 6-min peak headways, requiring 10 tph per branch, thus creating our "four branch" limit. It's possible (though not certain) that outer Comm Ave could "settle" for, say, 8-min peak headways to Kenmore if they were supplemented by 8-min peak headways to Kendall and beyond. (The frequency decrease's impact could be offset in part by the new availability of one-seat rides to Kendall and Sullivan -- avoiding the need for a transfer downtown.) 8-min headways can be achieved using only 7.5 tph, which would allow the Central Subway's 40 tph to serve five branches rather than four.

In theory, a full reroute could be unnecessary; Green Line trains could continue to run straight down Comm Ave, and be supplemented by Gold Line trains coming from West Station. This would allow for the same capacity increases described above, without losing Green Line service to Amory and Babcock. Connectivity to Kendall and Harvard would be slightly reduced relative to the scenario above, but still would be perfectly serviceable.

Finally, we can come full circle: utilize our Comm Ave-West Station connection for all Gold Line service, including trains to Brighton:

I'm less keen on this version, in part because it ends up being a very tight curve for Brighton-West Station trains at Packard's Corner, and because it ends up creating a more circuitous route to West Station. However, it would have utility in an incremental build scenario, since each element (tracks on Brighton, Comm Ave-West Station connection, West Station-BU Bridge subway) will be used for multiple services and therefore would be usable even when the full interchange system isn't yet built.

~~~

Does any of this matter? Ehn.

This is a somewhat complicated solution to a somewhat unlikely scenario, in which all of the following happen:

• Rider expectations return to pre-covid levels, requiring 6-min headways on Green Line branches
• Kenmore-Harvard LRT branch built
• Push to provide LRT service to 57 corridor
• Park-Nubian Branch built
• Capacity constraints in Central Subway remain the same (or D, E & Needham unable to be rerouted out)

While not completely implausible, a lot of things would need to happen first before this idea becomes "urgent".

Is there merit beyond any urgent "need"? I do think there are benefits:

• Larger reach of one-seat rides to Kendall and Sullivan via Grand Junction -- if you're going to do the LRT conversion, why not do as much with it as you can? Try to build out as full a network as possible
• Improves connectivity to West Station, creating a stronger "transit center of gravity"
  • the current plans will see good(ish) ped access to the redeveloped Beacon Park, and an LRT branch to Harvard would provide reasonable transfers to areas north of the station; a Grand Junction line would provide access to Cambridgeport and Kendall; and Kenmore would be accessible via both said Green Line branch as well as continuing service to Lansdowne
  • however, transit access from the south and southwest will remain very poor, including to employment centers around St Elizabeth's as well as Boston College (see map from US Census of job density below); LRT service southwestward of the station would remedy that
  • in an NSRL-world, this connectivity becomes more urgent, as it will link Allston/Brighton residents to jobs located on the northside of the regional rail network
• Better transfer opportunities to Grand Junction services, and to services to Harvard
  • absent this proposal, Grand Junction services will "land" at West Station, Kenmore/BU, or both. If you're a B or 57 rider, and you wish to go either to Kendall or to Harvard, you'll need to travel all the way in to BU or Kenmore, transfer, and then double back
  • creating a proper transfer node at West Station will reduce crowding at, and west of, Kenmore

Job density in Allston/Brighton:

 

[737900er]

I concur that this could be done, but the value isn't quite there today. I think any reactivated A needs to go to Kenmore given the current situation.

Anecdotally, ridership on the 57 corridor is very Boston focused (BU, Back Bay, Downtown). People who want to live in the same general area but commute to Harvard or Kendall live on the 71/73 corridor. RER headways to Back Bay and South Station at West Station or Newton Corner will never reach Kenmore inbound levels, so I think this would require running the current 57 alongside the "new A." Brighton-Cambridge ridership is pretty peak-focused too, and the 64+86 do a good enough job for most of them.

 

[Riverside]

Yeah, I largely agree. I see this as a potential third-, fourth- or maybe even fifth-generation project, if it ever comes to pass. And very much would be envisioned to open new commuting possibilities, as opposed to serving existing demand. Like I said -- "Does any of this matter? Ehn."

so I think this would require running the current 57 alongside the "new A."

Just to be clear, I am indeed proposing that the 57 continue to run, exactly as it does today (except with dedicated bus lanes and other infrastructure improvements).

I think any reactivated A needs to go to Kenmore given the current situation.

I don't necessarily disagree, but on the other hand it becomes a prioritization question. Does A to Oak Sq outweigh a branch to Harvard, for example? That kind of thing is why I think something like this proposal here is useful to keep in the backpocket -- offers more flexibility, even if you can't reach the ideal state.

RER headways to Back Bay and South Station at West Station or Newton Corner will never reach Kenmore inbound levels

Actually, I'm not sure it would be that far off from rapid transit levels. From everything I've seen, the vision is for all trains to stop at West Station. (Which tbh surprises me, but whatever.) If you ended up having 15-min headways to Riverside, 30-min headways to Framingham, and 30-min headways to Worcester (if that's even possible -- I'd need to go back and look at the studies), then in theory you are at 7.5-minute headways. It's true that that still doesn't rise to Kenmore's level, but it could be offset by the improved travel time to Back Bay and South Station. (Though I still agree one way or another that the 57 remains important.)

 

[Teban54]

I concur that this could be done, but the value isn't quite there today. I think any reactivated A needs to go to Kenmore given the current situation.

Anecdotally, ridership on the 57 corridor is very Boston focused (BU, Back Bay, Downtown). People who want to live in the same general area but commute to Harvard or Kendall live on the 71/73 corridor. RER headways to Back Bay and South Station at West Station or Newton Corner will never reach Kenmore inbound levels, so I think this would require running the current 57 alongside the "new A." Brighton-Cambridge ridership is pretty peak-focused too, and the 64+86 do a good enough job for most of them.

Here's another idea: How about running half of Oak Square/Watertown trains and half of Boston College trains to Central Subway via Amory St, and the other half on both branches to Kendall via West Station? This way both branches (past Packards Corner) still get one-seat rides to downtown albeit at lower frequencies, the Commonwealth Ave stations (where the B has the greatest ridership) keep the same frequencies as today, and you can get rid of the 57.

 

[737900er]

Here's another idea: How about running half of Oak Square/Watertown trains and half of Boston College trains to Central Subway via Amory St, and the other half on both branches to Kendall via West Station? This way both branches (past Packards Corner) still get one-seat rides to downtown albeit at lower frequencies, the Commonwealth Ave stations (where the B has the greatest ridership) keep the same frequencies as today, and you can get rid of the 57.

The problem I see is that the headways are really bad. You're looking at 12-16 minutes on each of these routings:
Watertown to Kenmore via Oak Square and Comm Ave
Oak Square to Kendall via GJ
Boston College to Kenmore via Comm Ave
Boston College to Kendall via GJ

Watertown to Kendall via Oak Square wouldn't see that much ridership on the Watertown to Oak Square segment. Watertown riders would continue to prefer 71 to Red to Kendall. Newton Corner riders would just take the RER to West and transfer to Green. Tremont St in Newton isn't very dense from Newton Corner to Oak Square. A to Kendall would basically have to be Oak Square to Kendall only.

Everyone's headways got absolutely destroyed just to shave ten minutes off trip times for the few who want to go to Cambridge and avoid having to transfer at Park Street. For those who did benefit, a lot of their trip time gains were lost to longer headways. Even for Watertown to Kenmore riders the 57 bus is better than 12-16 minute LRT headways. The BC to Kenmore trains would be absolutely packed -- the B west of Packards Corner has higher ridership than the entire C.

Oak Sq/BC to Kendall via GJ routings in addition to the existing trips might relieve some stress on Park Street, but having them instead of the existing trips creates way more problems than it solves.

 

[Teban54]

The problem I see is that the headways are really bad. You're looking at 12-16 minutes on each of these routings:
Watertown to Kenmore via Oak Square and Comm Ave
Oak Square to Kendall via GJ
Boston College to Kenmore via Comm Ave
Boston College to Kendall via GJ

Watertown to Kendall via Oak Square wouldn't see that much ridership on the Watertown to Oak Square segment. Watertown riders would continue to prefer 71 to Red to Kendall. Newton Corner riders would just take the RER to West and transfer to Green. Tremont St in Newton isn't very dense from Newton Corner to Oak Square. A to Kendall would basically have to be Oak Square to Kendall only.

Everyone's headways got absolutely destroyed just to shave ten minutes off trip times for the few who want to go to Cambridge and avoid having to transfer at Park Street. For those who did benefit, a lot of their trip time gains were lost to longer headways. Even for Watertown to Kenmore riders the 57 bus is better than 12-16 minute LRT headways. The BC to Kenmore trains would be absolutely packed -- the B west of Packards Corner has higher ridership than the entire C.

Oak Sq/BC to Kendall via GJ routings in addition to the existing trips might relieve some stress on Park Street, but having them instead of the existing trips creates way more problems than it solves.

Thanks, I had no idea the B west of Packards Corner still has that much ridership.

In this case, I think there's no choice but to fit a resurrected A (or at least a majority of A trains) to Kenmore and Central Subway. Though it might still make sense to make West Station a regional transfer hub, with the A (mayyybe also B?) diverted there before heading to Kenmore.

 

[737900er]

The issue with running the A into the Central Subway is figuring out the slots, retaining the possibility of Needham/Nubian/Seaport routings into the Central Subway, and being politically feasible.

Short turning C at Kenmore is doable mechanically, but politically impossible.
Short turning A at Kenmore is impossible mechanically, but politically doable.

Given the wins that Newton would get with a "New A" + Needham Branch + RER to Riverside (and maybe even a cleaned up Newton Corner), it might be possible to negotiate them into allowing Highland trains to be short-turned at Kenmore, but that would be hampering the whole operation because Highland is the best performing branch.

The most that has been run in recent years is about 44 trains per hour, so it's almost certainly impossible to exceed this.

You could maybe run something like the below, but it's very kludgy (although that's nothing new for the Green Line), would require most of the branches to be operated with 2-car trains of Type 10s, and probably wouldn't be compatible with a new signal system, would piss off a lot of people,.

 

[Riverside]

@737900er spot on. You are going down the same rabbit hole I went down about a year ago, starting with this post. My focus at the time was a bit different (Essex St vs Bay Village -- since writing, I've changed my thinking a bit and largely embraced the Bay Village option as the least-imperfect alternative), but much of the discussion is exactly what you are describing now. (Down to a spreadsheet I still have that's basically identical to yours.) The conversation went back and forth for about four pages, and then I wrote a follow-up about a month later.

There are two problems. First is that particular branches have to feed in to the Central Subway. 40tph divides reasonably into four branches, serviceably into five branches, okay-ish into six branches, and declines rapidly once you get to seven branches or more.

Nubian needs to run in to the Central Subway. B and C both have over a century of one-seat rides to Park St, to say nothing of the political clout to keep them. So, you have one or two slots left, maybe three if you convince everyone to take significant frequency reduction (unreasonably, in my opinion).

So you need to choose two of the following to run into Park St:

• Seaport
• Heath St/Hyde Sq/Arborway via Huntington
• Riverside via Huntington or Riverside via Kenmore
• Needham via Huntington or Needham via Kenmore
• Oak Sq/Newton Corner/Watertown
• Harvard
• Grand Junction

And that's where all this gets dicey. To run all of those services into Park St at current frequencies, Park would need to be able to handle 100 trains per hour in each direction. Even if we dropped every branch down to 10-minute headways (unacceptably low, unless supplemented by other services), Park would still need to handle 60 trains per hour in each direction -- not impossible, but as mentioned would be insufficient. Trying to provide a one-seat ride to Park from all potential branches will result in worse service for everyone.

As you say, this is where the prioritization question comes in.

Now, @Teban54 is right to think about interlining possibilities (in my opinion). This is essentially how we currently meet the ridership demands of the Central Subway -- combine several branches along a single stretch.

And I do think there is potential for the kind of 2x2 matrix that you describe (Kenmore-Oak, Kenmore-BC, Kendall-Oak, Kendall-BC) -- but the key is in the underlying frequencies of each service pattern, as @737900er describes. In that kind of topology, you can't combine two low-freq services to make a mid-freq combined service: you need to start with each service having a usable frequency, and then combine two mid-freq services to make a high-freq combined service.

This was what I was trying to describe here:

That supplementary service in turn could enable a greater number of branches to feed into the Central Subway: pre-covid, the Central Subway handled 40 tph; historically each Green Line branch has seen 6-min peak headways, requiring 10 tph per branch, thus creating our "four branch" limit. It's possible (though not certain) that outer Comm Ave could "settle" for, say, 8-min peak headways to Kenmore if they were supplemented by 8-min peak headways to Kendall and beyond. (The frequency decrease's impact could be offset in part by the new availability of one-seat rides to Kendall and Sullivan -- avoiding the need for a transfer downtown.) 8-min headways can be achieved using only 7.5 tph, which would allow the Central Subway's 40 tph to serve five branches rather than four.

Which would look like:

• BC-Packards Corner: total 15 tph (4 min)
  • BC-Kenmore: 7.5 tph (8 min)
  • BC-Kendall: 7.5 tph (8 min)
• Oak Sq-Packards Corner: total 15 tph (4 min)
  • Oak Sq-Kenmore: 7.5 tph (8 min)
  • Oak Sq-Kendall: 7.5 tph (8 min)
• Packards Corner-Kenmore: total 15 tph (4 min)
  • from BC: 7.5 tph (8 min)
  • from Oak: 7.5 tph (8 min)
• Packards Corner-Kendall: total 15 tph (4 min)
  • from BC: 7.5 tph (8 min)
  • from Oak: 7.5 tph (8 min)

From a network design perspective, it could be done, but requires careful balance. (And of course also becomes an operational headache quite easily -- becomes all the more important to keep each piece of the overall network in good repair.)

Going full-tilt to one-seat rides everywhere creates big headaches. But, it's worth remembering that going full-tilt in the other direction -- transfers transfers transfers -- brings its own downsides, and understandable political opposition. Any final design will doubtless use a combination of both strategies, and the question is which branch should get which strategy.

~~~

FWIW, what I've landed on for the moment for a Brighton Branch would be:

1) Extend from West Station, not Packard's Corner
2) Keep the 57 in place, with BRT enhancements, running at ~10 min headways from Watertown to Kenmore
3) Run Green Line trains from Park St via West Station to Oak Sq, at "half branch" frequencies -- 10-12 min headways
4) In essence what we're doing is converting the 57A to LRT and extending it to downtown -- it's still a service enhancement, but builds on existing rider behaviors that expect a transfer at Kenmore anyway
5) If necessary, Gold Line trains can later be layered on top to increase frequencies to Oak Sq
6) Ideally, this would combined with a B reroute to West Station, so that Gold Line riders headed for downtown will have a higher frequency of Green Line trains to transfer to (Harvard, Oak, and BC), but even without the reroute, Gold passengers would still be able to transfer to a cumulative 15 tph service, mirroring a similar current experience at Kenmore for 57 passengers

And extend the Blue Line to Kenmore to offer a faster transfer downtown as well as immediate transfer to Blue to avoid traveling all the way to Government Center.

~~~

Finally: @737900er, something for you to add to your spreadsheet is sending some fraction of Riverside, Needham, and Huntington trains (assuming a D-E connector is built, and the Huntington subway extended) to the Seaport rather than Park St. In your chart, you could reroute those Kenmore short-turns to the Seaport instead.

 

[737900er]

Finally: @737900er, something for you to add to your spreadsheet is sending some fraction of Riverside, Needham, and Huntington trains (assuming a D-E connector is built, and the Huntington subway extended) to the Seaport rather than Park St. In your chart, you could reroute those Kenmore short-turns to the Seaport instead.

This discussion got me thinking:

If we built the Commonwealth Ave Subway and with connections to Harvard and GJ by @F-Line to Dudley , would it be worth the trouble of connecting the Commonwealth Ave Subway to the Kenmore loop? I realize it's a big project, but so is the proposal.

This would enable service patterns like Highland to GJ via Kenmore Loop to keep trains out of the Central Subway and take pressure off Park Street. If the Huntington Ave Subway was ever built, it would also allow Highland trains to retain their connection to Kenmore without having to run Kenmore short turns, a Longwood-Kenmore dinky, or split service patterns.

Alternatively, it would allow restored A trains to run to Kenmore without taking Central Subway slots.

 

[Teban54]

This discussion got me thinking:

If we built the Commonwealth Ave Subway and with connections to Harvard and GJ by @F-Line to Dudley , would it be worth the trouble of connecting the Commonwealth Ave Subway to the Kenmore loop? I realize it's a big project, but so is the proposal.

This would enable service patterns like Highland to GJ via Kenmore Loop to keep trains out of the Central Subway and take pressure off Park Street. If the Huntington Ave Subway was ever built, it would also allow Highland trains to retain their connection to Kenmore without having to run Kenmore short turns, a Longwood-Kenmore dinky, or split service patterns.

Alternatively, it would allow restored A trains to run to Kenmore without taking Central Subway slots.

The idea of a Kenmore modification to allow a B-B loop and/or a B-D connection has been in discussion for ages way before both of us joined. The general consensus is that using the existing turn is too difficult, as F-Line explained in 2019:

Kenmore Loop only goes C/D to C/D between the outermost tracks of the station. The B takes the two center tracks. Loop pulls out too close to the platform for there to be new crossovers accessing the loop from the inner B tracks, and the cross-cutting traffic would induce delays.

Structurally, it would be heinously disruptive to try to blow up the loop and reconfigure it for a B-to-D/D-to-B direction, or spread stuff out to try to create a new 'lower' loop underpinning the tunnel just east of the station. Also, a B-to-D loop would be much tighter and slower than the current one, while probably needing to overcompensate for the changed geometry by bulbing-out more under the south side of Kenmore Square (at increased building mitigation risk) rather than staying centered under the Square like the current loop.


Now, UR thru-running loopage might've be a desireable thing if the south half of the Ring were also light rail branching off Longwood or Brookline Village. But the lack of available ROW's south vs. near-perfect string of ROW's north means you're probably just dividing the thing in half as northern LRT and southern BRT out of necessity and changing upstairs/downstairs at Kenmore to get between the NW and SW Ring quadrants. It's the only way to ideally deploy modes, and since a radial line is intrinsically a quick-on/quick-off transfer centric affair there's extremely few people who would be riding it for more than a quadrant at a time...much less riding end-to-end like it's Alewife-Braintree warped into an oval.

For that reason, a hop-across at Kenmore to a Longwood-fetching D satisfies the 66 audience plenty well. In a case like that, existing Kenmore Loop serves up an opportunity to strengthen the transfers. We have our E-to-D surface connection, and its usefulness is primarily going to be peak-period augmentation of Huntington Ave. service like the old days pre-1985 when Heath-Lechmere turns interlined simultaneous with Arborway-Park turns. Only with the connection at Brookline Village you can--instead of short-turning or running west to Reservoir--opt to turn east off of a side platform on Pearl St. onto the inbound D, and loop at Kenmore. Then reverse back to Brookline Village, back onto the connector, and back down Huntington as a regular E. It adds zero new congestion to Kenmore because the loop splits before B/UR and C/D merge onto the 2-track Central Subway mainline. But it does mean that--however much you throttle up/down service levels on that little wraparound--you can guarantee those 66'ers a waiting train to Longwood on nearly every slot.

^Very useful indeed^. This is but one of many things having more interconnects enables.

F-Line proposed a boomerang that would allow a bidirectional B-D connection in 2015, but the images are dead:

Here's what I'm thinking for a Kenmore reconfig that doesn't blow too much shit up:

Current config:

Future config:

Red arrows are flow of traffic in the D-to-B direction, blue arrows the flow of traffic in the B-to-D direction. New Lower loop, which is out by the Raleigh St. side, would be double-track but unidirectional. And the inner track on a much shorter length of loop tunnel so the outer one would encase it.

Then decommission the upper loop to free up a little space on the wall side and shave some wall space elsewhere to build an express track bypassing the C/D platforms. UR trains making the BU Central<-->Kenmore<-->Longwood boomerang would stop on the B platforms, skip the D platforms. This, collectively, is what keeps the boomerang moves segregated from fouling Central Subway capacity to/from Hynes.

I know the "boomerang" keeps getting brought up by several members here, so if anyone has recollections of how the proposal was like, I would be happy to read them.

Unrelated to the above, I personally think the best shot for a B-B loop would be to bury a Commonwealth Ave subway below the current platforms. This "Kenmore Lower" platform would connect to the Commonwealth Ave subway to the west, have its own loop or crossover below the current C/D loop to the east, and join the Central Subway between Kenmore and Hynes further east. You can also extend it eastward via the Turnpike as a Back Bay-Seaport subway as @Riverside described here.

If there's a way to connect the C/D tunnel to Kenmore Lower, it would allow a B-D connection, but I'm not sure how doable it is. Even if not, this would already allow a B-B loop for Harvard, GJ and possibly Watertown trains to short-turn at Kenmore Lower (edit: and Boston College if really needed), which would be incredibly useful especially for the first two.

 

[Tallguy]

The idea of a Kenmore modification to allow a B-B loop and/or a B-D connection has been in discussion for ages way before both of us joined. The general consensus is that using the existing turn is too difficult, as F-Line explained in 2019:


F-Line proposed a boomerang that would allow a bidirectional B-D connection in 2015, but the images are dead:

I know the "boomerang" keeps getting brought up by several members here, so if anyone has recollections of how the proposal was like, I would be happy to read them.

Unrelated to the above, I personally think the best shot for a B-B loop would be to bury a Commonwealth Ave subway below the current platforms. This "Kenmore Lower" platform would connect to the Commonwealth Ave subway to the west, have its own loop or crossover below the current C/D loop to the east, and join the Central Subway between Kenmore and Hynes further east. You can also extend it eastward via the Turnpike as a Back Bay-Seaport subway as @Riverside described here.

If there's a way to connect the C/D tunnel to Kenmore Lower, it would allow a B-D connection, but I'm not sure how doable it is. Even if not, this would already allow a B-B loop for Harvard, GJ and possibly Watertown trains to short-turn at Kenmore, which would be incredibly useful especially for the first two.

F-Line, can you repost?

 

[F-Line to Dudley]

F-Line, can you repost?

Don't have the images, and honestly don't remember it all that well 7 years later. Pretty sure the point of the exercise was just proving physical/geometric feasibility more than whether it made sense. So I MS Paint'ed the Kenmore Station structures underneath a Google Maps overlay, then traced a lower-level B-to-C/D loop just east of the station that would cram inside the street grid. It had a pronounced bulb out to the Raleigh/Beacon St. intersection, but was acceptable curve radius to work.

To be honest, I'm not real keen on the idea today. Maybe I was in 2015, but definitely not now or recently. While feasible in the absolute, it would be very expensive and construction-disruptive to the station and the Square to try to shiv it in, be relatively tight on radius (though looser than the current loop) and thus not real fast, scrape about as close as possible to building foundations, might require structural nuking of the current C/D-to-C/D loop for creating room, and would almost certainly preclude ever building a BLX-Kenmore level on the station. So I have a hard time seeing where all the pain and suffering is going to end up providing a functional-enough "boomerang" alignment to achieve net benefit. Especially when we're tallying up far-future benefits of that BLX level, which might clear out enough thru-to-Downtown loading that you could terminate some C's or D's to loop @ Kenmore instead of running thru to create more bandwidth for Ring+B patterns.

 

[ceo]

Google Maps now has up-to-date imagery of Brickbottom Junction, so I took a stab at figuring out how the Urban Ring could be built through there, and out to the Mystic River. Here's the result, zoom in to see the details. Some notes:

• We can't route the Urban Ring northbound up the Union Square inbound ramp and out the current stub track, because there are piers for the McGrath Highway bridge right where the turnout would have to be (you can see where they are on this map). But, assuming it doesn't completely hose CRMF operations, we can instead cannibalize the "Paint Shop" lead that ends right there. This has the advantage of the two routes not having to share track, but the rather severe disadvantage that there's no northbound-to-inbound connection between the Urban Ring and the GLX.
• This also requires some rearrangement of the CRMF "back side" tracks so they can connect to the Fitchburg Line.
• Urban Ring southbound does have to share track with Union Square outbound, and also has to cut off a corner of the layover yard. It has to split from the Union Square line to the north and duck under all four tracks, again because of where the piers are.
• Urban Ring uses two of the three "Valley Track" alignments around the back of the CRMF. The other one gets a new connection to the CMRF trackage to get to the Fitchburg Line.
• The south wye of the Yard 10 lead gets realigned to connect to the remaining Valley Track, and the Urban Ring ducks under it. The north wye gets cut off since there's nothing for it to connect to.
• The Sullivan Square platform sits on the alignment of the east current storage track, and Urban Ring northbound uses the alignment of the current Orange Line southbound track. The Orange Line shifts over one so that the current "express track" becomes the northbound track. Alternatively, switch the UR platform and southbound track, so the southbound platform is on the west side of the ROW. (The track alignments under the I-93 viaduct cannot be altered, because of the piers.)
• Urban Ring ducks under both commuter rail lines and crosses the Mystic on the alignment of the former bridge, thus staying on the east/south side of the Newburyport/Rockport Line the rest of the way.

 

[Riverside]

I like it! But to confirm, I don't think I see trackage that could support Grand Junction -> Lechmere or Sullivan -> Lechmere, correct? I could be convinced that Grand Junction -> Lechmere isn't worth the trouble, but not being able to hook Sullivan (and Everett and Chelsea) into the Central Subway seems like a pretty big drawback.

 

[Brattle Loop]

I like it! But to confirm, I don't think I see trackage that could support Grand Junction -> Lechmere or Sullivan -> Lechmere, correct? I could be convinced that Grand Junction -> Lechmere isn't worth the trouble, but not being able to hook Sullivan (and Everett and Chelsea) into the Central Subway seems like a pretty big drawback.

Grand Junction -> Sullivan looks like it'd be possible with modifications (completions, really) to the Union/Medford junction and some reconfiguration of the maintenance facility tracks. One of the unbuilt spurs that has a stub there would have run Union Sq-branch -> carhouse, so that alignment could send cars headed towards Sullivan from GJ and the existing carhouse lead could handle the reverse direction. The other unbuilt spur stub is in the right direction to throw traffic towards Sullivan from Lechmere, but I agree that there doesn't appear to be any existing alignment for handling cars Sullivan -> Lechmere (though Sullivan <--> GJ looks possible).

 

[ceo]

I like it! But to confirm, I don't think I see trackage that could support Grand Junction -> Lechmere or Sullivan -> Lechmere, correct? I could be convinced that Grand Junction -> Lechmere isn't worth the trouble, but not being able to hook Sullivan (and Everett and Chelsea) into the Central Subway seems like a pretty big drawback.

Yes, and I can't see a way to do that without building a honking big flyover of the entire Brickbottom Junction complex, and the bike path overpass might be in the way. If nothing else, people coming in from Everett and Chelsea can transfer to the Orange Line at Sullivan. In an ideal universe that would be a cross-platform transfer, which would require rebuilding the north end of the Orange Line viaduct.

 

[Riverside]

I was playing around with rerouting the northbound Medford track along the curve of the northbound Union spur (probably as a second track, since the grades of the viaduct would start to diverge), under the theory that you could then run something up on the inside of the junction, and join the southbound Medford viaduct track not far from where the southbound Union curve joins.

The problem then becomes that you'd still need to have the Sullivan-GJ southbound track cross over the Sullivan-GJ northbound track, and I don't have a great answer for that. I fudged it in this diagram as a underpass -- that area has been railyard since, like, the dawn of time, so there shouldn't be too many mysteries under there. But I haven't done the math to check whether there's enough running length for the grade.

Of course, I guess you could take the same idea I'm proposing here but run the Sullivan -> Lechmere track on a double-high viaduct over the northbound Medford + Union viaduct(s)... sounds more complicated but maybe wouldn't be.

Terrible sketch of the general idea:

Yes, and I can't see a way to do that without building a honking big flyover of the entire Brickbottom Junction complex, and the bike path overpass might be in the way.

Actually, this might not be as crazy as it sounds. [EDIT: more clearly, there might be an easier way to do this.] You only need to take care of southbound trackage -- northbound Lechmere -> Sullivan can still be handled exactly the way you've described above.

A single track along the current track alignment parallel to New Washington St, rising to a viaduct and descending in between the tracks just south of East Somerville, could do the trick. The GLX ROW has extra width on the eastern side, so there's room to relocate the northbound Medford track over one "slot" to make room for the junction without touching the western southbound Medford track or the Community Path at all.

Vague sketch:

It's a little roundabout, but not that much. This alignment is a smidge over a mile, while the eastern alignment (i.e. your current Lechmere -> Sullivan northbound alignment) is .75 miles.

It might also be possible to do it with an underpass cut under the Commuter Rail embankment -- not sure. It wouldn't be a small build, but it would require fewer changes to the existing GLX junction.

 

[ceo]

The problem with the North Washington St alignment is that's the freight connection for the Lowell Line.

 

[Riverside]

The problem with the North Washington St alignment is that's the freight connection for the Lowell Line.

True, but the buildings there are pretty far apart — may be possible to slot a dedicated track in there, even if the pavement need to be rearranged. Or a viaduct or a tunnel.

 

[Arlington]

Does a Grand Junction GL brach really need to be fed from both ends?

and if both ends, does it need a Wye on the west? Does it need a flying junction ?

Sure, make sure these connections are possible, but are they a prerequisite for cost effective service?

would having/not a West Station change your answer?