(I assume by "eastbound", you mean "northbound" right? The east-west model makes a lot less sense to me now that GLX is open.) I was a little less worried about redoing the stairs and headhouse etc on the northbound platform -- my main priority was simply to work within the existing footprint of the station. Any version of this rebuild will not be cheap, at all. If it becomes a requirement to completely leave the headhouse and existing stairs untouched, then yes, this whole idea is DOA (though, see caveat below).
Yeah, eastbound is interchangeable with northbound at GC. I think they still use eastbound/westbound as railroad directions operationally (which makes sense), even if the wayfinding signage at GC now says "North Station & North" on that side.
My vague recollection is that F-Line suggested some sort of in-station overpass to connect the two island LRT platforms, but I can't remember how, and can't see how that would work.
There isn't room for that without taking the roof off the 1898-built side, which may not be possible at all - that depends how deep under the street the station is.
Fascinating thought exercise, but firmly in Crazy Transit Pitches territory, I think. But still fun.
Nice, congrats. I like the green reconfig map a lot. I’d say a key idea that isnt mentioned as key ideas on that is the line to Harvard sq which would significantly alleviate downtown pressures; also, that the D-E connector + GL to Harvard Sq means there are light rail partial crosstown routes. Those seem “key” to me, since a lot of our collective GL visions are mere extensions, whereas the Harvard ext actually is more than that if you can go from Harvard to Kenmore and then transfer to get to Hyde Sq.
Glad you’re doing a site, will be cool to see where it goes.
Regarding the Harvard branch in my GLR map, the funny thing is, that's actually one of the ideas where I intentionally thought of as "not the final form, just as an intermediate stage". The same applies to northside F (Grand Junction) and G (Everett-Chelsea) branches. At the time of making the map, my opinion was that: In the long term, all three branches are better suited as part of a (roughly) segregated system that does not run into the Green Line's main trunk; but for the purpose of the map, and as initial builds, it's useful for both showing "an LRT branch can be built here", and "a benefit of GLR is it allows you to do that well".
For the Harvard branch in particular, I thought it was probably better off as a fully circumferential route that turns south around Kenmore, like my grey dashed lines indicated. As for the exact route:
@Riverside's typical idea has been turning south to the Highland Branch (today's D at Fenway and Longwood), and then using Fenway (the road) as a surface (or...) transitway to reach Ruggles.
@TheRatmeister's proposal is to turn from West Station to Harvard Ave (B), then to Coolidge Corner and LMA, likely as a TBM tunnel.
However, in the days since making the map, I'm actually beginning to think that a radial Harvard - West Station - Downtown route may still have value. Namely:
Connecting Harvard (thus northside Red Line and parts of Cambridge) to the Copley and Back Bay area, a major destination that's notoriously hard to get to from Cambridge given the distance
With some GLR designs, allows a one-seat ride from Seaport to Harvard, West Station and Kenmore (though really more about Kenmore-Seaport than Harvard-Seaport, since Kenmore-Seaport would otherwise be a three-seat ride)
With some GLR designs, allows a one-seat ride from Seaport to Harvard, West Station and Kenmore (though really more about Kenmore-Seaport than Harvard-Seaport, since Kenmore-Seaport would otherwise be a three-seat ride)
What about an pedestrian tunnel between Copley and Back Bay or Arlington and a hypothetical Bay Village Station? It would still be a two-seat ride rather than a one-seat ride, but it would still remove the extra trip+transfer at Park St, and would obviously be cheaper to operate because it's a pedestrian tunnel, not a rail line.
Connecting Harvard (thus northside Red Line and parts of Cambridge) to the Copley and Back Bay area, a major destination that's notoriously hard to get to from Cambridge given the distance
But would this be faster than the 1 is now? Harvard to Hynes CC is (At least timetabled as) 13 minutes on the 1, but even just Packards Corner-Hynes CC on the B is 10 minutes. Maybe a better solution for these journeys would be a new pedestrian bridge to parallel the Harvard Bridge.
It's probably worth clarifying that I have much more of a different philosophy on GLR, focused more on a modest expansion of the GL with a 2nd subway and restored/expanded western/southern services (A branch, Needham, Washington St) and a fast urban ring metro line to better connect the areas outside downtown, rather than a sprawling light rail network focused on maximizing one seat rides both to, from, and around downtown.
What about an pedestrian tunnel between Copley and Back Bay or Arlington and a hypothetical Bay Village Station? It would still be a two-seat ride rather than a one-seat ride, but it would still remove the extra trip+transfer at Park St, and would obviously be cheaper to operate because it's a pedestrian tunnel, not a rail line.
A tunnel connecting either is problematic because of the Pike -- you have to go really deep to connect to anything in Back Bay proper (north of the Pike) from Back Bay Station, on the south side if the Pike trench.
Copley has the added challenges of 1) having a station that is not even connected internally East/West and 2) is flanked by historic buildings on pilings. The last time the T tapped into Copley Station for a project (elevators) they did millions in damage to Old South Church.
But would this be faster than the 1 is now? Harvard to Hynes CC is (At least timetabled as) 13 minutes on the 1, but even just Packards Corner-Hynes CC on the B is 10 minutes.
That's a valid concern. Another route that's also worth consideration is a Red -- Orange/Green transfer via downtown.
But first, I'd advise against using today's B as a proxy. The 6 stops of street running can be eliminated rather easily in most proposals, especially since constructing a tunnel underneath the B reservation is cheap (and proposed on many LRT Urban Ring proposals). You can easily grade-separate West Station to Kenmore via Worcester ROW and BU Bridge, with one intermediate stop at BU Central/East.
@Riverside had done travel time estimates on this segment:
Now: one problem in the background of all this: the Red Line (normally/should be/can be) fast. Imagine a T1 <> Back Bay commuter: today, the fastest journey would be T1 > Red > Orange/Green. In theory, a Green Line branch that runs Park <> Kenmore <> Harvard would turn that into a 2SR: T1 > Green. But, because the Red Line is so fast, the 3SR might be just as fast:
Harvard <> DTX: 12 minutes + 2.5 min transfer + DTX <> BBY: 4 minutes = 18.5 min
Harvard <> Copley: 3.7 miles at 12 mph = 18.5 min
(Ha, I didn't actually intend to make those numbers equal -- some of those estimates are arbitrary, but that works out coincidentally well.)
I'm not saying that circumferential services to Red Line stations are useless, but the Red Line is too fast for its own good in some ways.
[Teban54's comments: The 2.5-min transfer at DTX may be optimistic given the expected headways on both lines, especially OL, but should still be close. There's also the option of transfering to Green at Park St which has better frequency, which is probably closer to the 2.5-min estimate. Meanwhile, I'm not sure how Riverside got the 3.7 miles for a GL Harvard branch: my own measurement on Google Maps has 4.1 miles.]
Here are my own estimates:
Like mentioned above, I measured 4.1 miles for Harvard-Copley, via F-Line's (open trench) alignment in Lower Allston. Assume full grade separation with 6 intermediate stops (as drawn on my GLR map, though there are arguments for two more stops at Western Ave and BU East).
Some analogs on today's Green Line:
D branch, Fenway to Chestnut Hill: 3.9 miles, 5 intermediate stops, 12.2 minutes. Extrapolating to 4.1 miles and adding 1 min for another stop gives 13 minutes.
Green Line trunk, Park St to Kenmore: 1.9 miles, 4 intermediate stops, 10 minutes. Extrapolating to 4.1 miles alone gives 21.6 minutes, but that has 8.6 intermediate stops. Cutting down on 2.6 stops gives 18-19 minutes.
This might be a sign of Green Line trunk being slower than average due to bunching and short headways.
Overall, I'd probably do a blend of the two that give 15-16 minutes.
The two estimations range between 13 to 18.5 mins, so let's do a blend and say 16 mins. In contrast, for a T1 rider from Harvard to Copley (not Hynes), the scheduled 13 minutes plus transfer may give an overall time close to 16 mins. Thus, for people with Copley as the destination, the Harvard GL branch seems competitive to T1 and the downtown transfer, though the data is messy.
Copley vs. Hynes: For people whose destination is Hynes, the T1 will probably be faster. But here's my unscientific anecdotal speculation: I suspect many T1 riders would actually prefer Copley over Hynes if given the choice, but use Hynes today because of lack of alternatives. Copley is right at the center of both the Newbury St shopping district and the Back Bay employment district, whereas Hynes is at the edge.
Of course, this needs more analysis to confirm. There's also the possibility that some T1 riders may walk west to the Fenway neighborhood, though I doubt it based on my experience.
Are you suggesting a pedestrian bridge to eliminate the sidewalks and/or bike lanes on Harvard Bridge, and thus free up room for transit priority on the old bridge? Yeah, that's exactly what I've been thinking for improving the Mass Ave corridor in recent times. (A pedestrian bridge is obviously not load-bearing for RL-Back Bay traffic.)
On the other hand, I'd say a Harvard-West Station-Copley rapid transit service offers several unique connections, such as Harvard-BU, basically anything involving West Station, and possibly Seaport. (They don't have to be mutually exclusive with another circumferential service further south.)
What about an pedestrian tunnel between Copley and Back Bay or Arlington and a hypothetical Bay Village Station? It would still be a two-seat ride rather than a one-seat ride, but it would still remove the extra trip+transfer at Park St, and would obviously be cheaper to operate because it's a pedestrian tunnel, not a rail line.
In addition to what Jeff said about engineering, the pedestrian tunnel would also be quite long. At ~1,000 ft(not even including distance from the inbound GL platform to Dartmouth St), it will be the longest in Boston, and almost twice as long as OL-BL State (580' estimated). Google Maps estimates a 5-min walk. Granted, such long pedestrian walkways aren't non-existent worldwide, but still on the long side, and it probably gets to the point where placing automated walkways in the tunnel would make sense (which Singapore actually does at many stations). Additionally, it's not too much shorter than the distance between Eastern Route and the Blue Line at Wonderland (1300' minimum), and the length had been a reason why (this forum's consensus being that) a Wonderland CR station won't do well.
A tunnel connecting either is problematic because of the Pike -- you have to go really deep to connect to anything in Back Bay proper (north of the Pike) from Back Bay Station, on the south side if the Pike trench.
Copley has the added challenges of 1) having a station that is not even connected internally East/West and 2) is flanked by historic buildings on pilings. The last time the T tapped into Copley Station for a project (elevators) they did millions in damage to Old South Church.
The Pike itself isn't super deep there, at the same level as NEC and OL. Can't you just dig one level underneath (which you have to do anyway to build anything underground that's not at GL or OL platform level)?
While your second point is indeed a significant challenge, it can also be interpreted as an additional benefit of constructing such a tunnel. But overall, I do think the cost-benefit analysis is marginal (unless a brand new north-south line is built as a Crazy Transit Pitch).
It's probably worth clarifying that I have much more of a different philosophy on GLR, focused more on a modest expansion of the GL with a 2nd subway and restored/expanded western/southern services (A branch, Needham, Washington St) and a fast urban ring metro line to better connect the areas outside downtown, rather than a sprawling light rail network focused on maximizing one seat rides both to, from, and around downtown.
A quick comment on this. I do get your own philosophy, but I don't think there's a need to speak of "light rail" as an inherently bad thing, and perhaps more importantly, there's no inherent need to link GLR and many Urban Ring proposals here as the same thing.
The core concept of GLR is exactly what you described: A second subway that enables the branches you mentioned. In fact, these are exactly what I've drawn in my GLR map.
On the other hand, most LRT Urban Ring proposals don't necessarily need to interline with the (reconfigured) GL network at all. In fact, Riverside's proposal(which I think prompted this discussion) has zero interlining between his Urban Ring lines (Pink, Bronze) and GLR (Gold). This is an entirely new network, not a "sprawling light rail network" that's built into GLR, just that they (can) have possible track connections and share rolling stock.
What makes an Urban Ring fast or slow is its route character, not its rolling stock. The choice of LRT has its disadvantages (can't be fully automated, lower capacity than HRT), but just as many advantages (no need for 100% grade separation, revenue and non-revenue operational flexibility, no need for dedicated maintenance facilities and unique rolling stock if GL yards are enough, possibly sharing ROW for different services like discussed above for Harvard-WS).
Ultimately, I think the mode choice for UR just depends on your confidence in (1) Boston's willingness to build grade-separated ROWs that are often expensive, and (2) Boston's ability to build a huge project in one go or quick succession, rather than in phases that drag over decades. I'd say (2) offers even greater justification for LRT than (1), as intermediate pieces can be built and then tied into GL temporarily, offering better service in the mean time. But none of this must be inherently linked to GLR -- which is why I deliberately excluded any Urban Ring components from my own GLR map.
Are you suggesting a pedestrian bridge to eliminate the sidewalks and/or bike lanes on Harvard Bridge, and thus free up room for transit priority on the old bridge? Yeah, that's exactly what I've been thinking for improving the Mass Ave corridor in recent times.
Yes, exactly. The existing bridge can only fit 2 of the the following: Transit Lanes, Bike Lanes, or Sidewalks. People are generally the easiest (And cheapest) to design a bridge for, and pedestrian bridges are also just generally neat. This one would have an excellent view of the Pru/Back Bay, the Esplanade, and Downtown. And obviously with this done you'd be able to run bus lanes down Mass Ave all the way to Washington St without removing any street parking.
Travel time competitiveness of a Harvard-Copley branch
That's a valid concern. Another route that's also worth consideration is a Red -- Orange/Green transfer via downtown.
But first, I'd advise against using today's B as a proxy. The 6 stops of street running can be eliminated rather easily in most proposals, especially since constructing a tunnel underneath the B reservation is cheap (and proposed on many LRT Urban Ring proposals). You can easily grade-separate West Station to Kenmore via Worcester ROW and BU Bridge, with one intermediate stop at BU Central/East.
@Riverside had done travel time estimates on this segment:
Now: one problem in the background of all this: the Red Line (normally/should be/can be) fast. Imagine a T1 <> Back Bay commuter: today, the fastest journey would be T1 > Red > Orange/Green. In theory, a Green Line branch that runs Park <> Kenmore <> Harvard would turn that into a 2SR: T1 > Green. But, because the Red Line is so fast, the 3SR might be just as fast:
Harvard <> DTX: 12 minutes + 2.5 min transfer + DTX <> BBY: 4 minutes = 18.5 min
Harvard <> Copley: 3.7 miles at 12 mph = 18.5 min
(Ha, I didn't actually intend to make those numbers equal -- some of those estimates are arbitrary, but that works out coincidentally well.)
I'm not saying that circumferential services to Red Line stations are useless, but the Red Line is too fast for its own good in some ways.
[Teban54's comments: The 2.5-min transfer at DTX may be optimistic given the expected headways on both lines, especially OL, but should still be close. There's also the option of transfering to Green at Park St which has better frequency, which is probably closer to the 2.5-min estimate. Meanwhile, I'm not sure how Riverside got the 3.7 miles for a GL Harvard branch: my own measurement on Google Maps has 4.1 miles.]
Here are my own estimates:
Like mentioned above, I measured 4.1 miles for Harvard-Copley, via F-Line's (open trench) alignment in Lower Allston. Assume full grade separation with 6 intermediate stops (as drawn on my GLR map, though there are arguments for two more stops at Western Ave and BU East).
Some analogs on today's Green Line:
D branch, Fenway to Chestnut Hill: 3.9 miles, 5 intermediate stops, 12.2 minutes. Extrapolating to 4.1 miles and adding 1 min for another stop gives 13 minutes.
Green Line trunk, Park St to Kenmore: 1.9 miles, 4 intermediate stops, 10 minutes. Extrapolating to 4.1 miles alone gives 21.6 minutes, but that has 8.6 intermediate stops. Cutting down on 2.6 stops gives 18-19 minutes.
This might be a sign of Green Line trunk being slower than average due to bunching and short headways.
Overall, I'd probably do a blend of the two that give 15-16 minutes.
The two estimations range between 13 to 18.5 mins, so let's do a blend and say 16 mins. In contrast, for a T1 rider from Harvard to Copley (not Hynes), the scheduled 13 minutes plus transfer may give an overall time close to 16 mins. Thus, for people with Copley as the destination, the Harvard GL branch seems competitive to T1 and the downtown transfer, though the data is messy.
Copley vs. Hynes: For people whose destination is Hynes, the T1 will probably be faster. But here's my unscientific anecdotal speculation: I suspect many T1 riders would actually prefer Copley over Hynes if given the choice, but use Hynes today because of lack of alternatives. Copley is right at the center of both the Newbury St shopping district and the Back Bay employment district, whereas Hynes is at the edge.
Of course, this needs more analysis to confirm. There's also the possibility that some T1 riders may walk west to the Fenway neighborhood, though I doubt it based on my experience.
Are you suggesting a pedestrian bridge to eliminate the sidewalks and/or bike lanes on Harvard Bridge, and thus free up room for transit priority on the old bridge? Yeah, that's exactly what I've been thinking for improving the Mass Ave corridor in recent times. (A pedestrian bridge is obviously not load-bearing for RL-Back Bay traffic.)
On the other hand, I'd say a Harvard-West Station-Copley rapid transit service offers several unique connections, such as Harvard-BU, basically anything involving West Station, and possibly Seaport. (They don't have to be mutually exclusive with another circumferential service further south.)
Copley-Back Bay connector
In addition to what Jeff said about engineering, the pedestrian tunnel would also be quite long. At ~1,000 ft(not even including distance from the inbound GL platform to Dartmouth St), it will be the longest in Boston, and almost twice as long as OL-BL State (580' estimated). Google Maps estimates a 5-min walk. Granted, such long pedestrian walkways aren't non-existent worldwide, but still on the long side, and it probably gets to the point where placing automated walkways in the tunnel would make sense (which Singapore actually does at many stations). Additionally, it's not too much shorter than the distance between Eastern Route and the Blue Line at Wonderland (1300' minimum), and the length had been a reason why (this forum's consensus being that) a Wonderland CR station won't do well.
The Pike itself isn't super deep there, at the same level as NEC and OL. Can't you just dig one level underneath (which you have to do anyway to build anything underground that's not at GL or OL platform level)?
While your second point is indeed a significant challenge, it can also be interpreted as an additional benefit of constructing such a tunnel. But overall, I do think the cost-benefit analysis is marginal (unless a brand new north-south line is built as a Crazy Transit Pitch).
One could certainly go deep for the connection. It just means needing a lot of space on both ends for vertical access (escalators and elevators). Likely, with the distance, you also want to provide moving sidewalks, which means a wide, deep tunnel. That feels like a lot of cost for the connection (although it does improve Copley station with the connection between sides).
The alignment needed, though, is really problematic, since the only place the two sides of the station align is on the BPL side of Dartmouth Street, so you need to tunnel right next to the McKim building. That sounds like a recipe for disaster.
My Green Line Reconfiguration map (ArchBoston version here) won the BPL Leventhal Map Center's Judge's Choice award for the Transportation Dreams challenge!
We asked for speculative maps of Boston's transit system, and you answered... here are the results!
www.leventhalmap.org
Needless to say, I am very honored to receive this award among at least 11 other great submissions, each of which being extremely valuable in its own right (two of which are from @Riverside, and one from@TheRatmeister).
More importantly, I would like to extend my thanks to all of you on this forum. As mentioned in my technical statement (quoted in the announcement), “I would like to thank many ArchBoston members who made invaluable contributions in more than 10 years. While I participated in many discussions myself, my map would not have been possible without the creativity, analyses and insights of other members, such as Riverside, F-Line to Dudley, vanshnookenraggen, davem, The EGE, Brattle Loop, 737900er, TheRatmeister, and many more.” (Please don't be mad at me if I left you out, lol.)
The announcement linked above also provides a very detailed description of the GLR concept, largely based on descriptions I've included on my map.
My Green Line Reconfiguration map (ArchBoston version here) won the BPL Leventhal Map Center's Judge's Choice award for the Transportation Dreams challenge!
We asked for speculative maps of Boston's transit system, and you answered... here are the results!
www.leventhalmap.org
Needless to say, I am very honored to receive this award among at least 11 other great submissions, each of which being extremely valuable in its own right (two of which are from @Riverside, and one from@TheRatmeister).
More importantly, I would like to extend my thanks to all of you on this forum. As mentioned in my technical statement (quoted in the announcement), “I would like to thank many ArchBoston members who made invaluable contributions in more than 10 years. While I participated in many discussions myself, my map would not have been possible without the creativity, analyses and insights of other members, such as Riverside, F-Line to Dudley, vanshnookenraggen, davem, The EGE, Brattle Loop, 737900er, TheRatmeister, and many more.” (Please don't be mad at me if I left you out, lol.)
The announcement linked above also provides a very detailed description of the GLR concept, largely based on descriptions I've included on my map.
The alignment needed, though, is really problematic, since the only place the two sides of the station align is on the BPL side of Dartmouth Street, so you need to tunnel right next to the McKim building. That sounds like a recipe for disaster.
This gets you directly under Dartmouth St rather than under the BPL building. You can even swing further east under part of Copley Square if needed, provided you stay clear of Trinity Church.
Trying to access that when it's busy (on the inbound side) would be pretty awkward. Copley's platform isn't super wide. (Though T riders are likely familiar with the awkwardness of single access points at platform ends.)
Because of the transfer most riders need to make at Nubian, the number of transfers remains the same or even fewer with the Fairmount Line. BL is 3 transfers once the Red-Blue Connector gets built, down from 4 via SL4/5, GL is 3 transfers, the same as SL/4/5 currently, and OL is still only one 1 transfer, since it basically never makes sense to switch to the SL for transferring to the OL, just stay on the bus you were already on to Ruggles or Jackson Square, or take the 30/31 to Forest Hills. If we assume GL-Reconfig, which is what this about, GL D/E is only 1 transfer with Fairmount, another improvement.
The "10 minutes from Nubian to Boylston" claim is where I remain skeptical. Yes, the D Line and LA's A Line can clear 2 miles in 10 minutes, by making 1 or 2 stops, running in a dedicated ROW, where they never have to yield. That seems vastly different from any conditions possible on Washington at ground level.
Boylston to Nubian is about 2 miles. To travel that distance in 10 minutes, the average speed needs to be 12 miles per hour. About .25 miles of that distance is in the subway, so let's assume 20 mph there, about the same as between Blandford St and Kenmore. That comes out to about 45 seconds between Eliot Norton and Boylston. So that leaves 9:15 for the remaining 1.75 miles, or about 11.3 miles per hour required on the Washington St section. This is really not that much of a stretch with signal priority and intersection elimination. This is not just a time I made up, it's well within the realm of possiblity on this route.
We have seen countless proposals bringing rail transit, and even specifically radial transit, to Everett and Chelsea. So why shouldn't the core part of Roxbury get the same?
Fair point, but it would certainly seem like the better solution is to extend the Washington St Line down to ~Warren @ Quincy then to get all of Roxbury within walking distance of a direct rail service, not build a second line from Nubian to downtown. A Ceylon Park infill would also help with bringing more of Grove Hall into the Fairmount Line walkshed.
If the Fairmount Line offers a faster, more convient way to get where they're going then they will switch. And I think it's very clear that the Fairmount Line will always be faster than a bus journey if the number of transfers is the same, which it usually is.
Boylston to Nubian is about 2 miles. To travel that distance in 10 minutes, the average speed needs to be 12 miles per hour. About .25 miles of that distance is in the subway, so let's assume 20 mph there, about the same as between Blandford St and Kenmore. That comes out to about 45 seconds between Eliot Norton and Boylston. So that leaves 9:15 for the remaining 1.75 miles, or about 11.3 miles per hour required on the Washington St section. This is really not that much of a stretch with signal priority and intersection elimination. This is not just a time I made up, it's well within the realm of possiblity on this route.
Thanks for figuring the math there on the impact of the short stretch of subway. It did indeed annoy me that I glossed over it, so appreciate you hashing it out.
Where would you carry out intersection elimination?
I know 11.3 mph does not seem like a stretch to you, but it still seems like a tall order to me. As @Teban54 illustrated, that would mean being as fast, if not faster, than the current subway. That seems like a strong claim.
Are there real-world examples of a surface LRT (or even BRT) service operating at the speeds you are suggesting here, in an environment similar to Washington St in the South End? The closest analogue in Boston is (I would argue) the C Line, which is nowhere close to sustaining the average 11.3 mph targeted here. But are there other examples elsewhere in the world?
Are there real-world examples of a surface LRT (or even BRT) service operating at the speeds you are suggesting here, in an environment similar to Washington St in the South End? The closest analogue in Boston is (I would argue) the C Line, which is nowhere close to sustaining the average 11.3 mph targeted here. But are there other examples elsewhere in the world?
Here in Berlin or here in Salt Lake City. Both are about 2 mile segments with ~10 minute travel times. Both feature 7-8 stops, close to the 8 or 9 that would make sense on the Washington St Line.
Depends on how brutal you want to be. The general principle is that alleys never have left turns, nor do quiet residential streets if it doesn't severely limit accessibility. Main streets like Melnea Cass and Mass Ave obviously keep their intersections. Here's the changes I would make from a surface level glance:
Tremont St @ Oak St West: Change the intersection to only allow for right turns. This would mean EB Tremont->SB Shawmut and SB Tremont->EB Oak St only, no straight or left turning traffic.
Washington St between Oak St and Marginal Rd: 4 lanes, transit runs on the 2 western lanes and vehicle traffic runs on the two eastern lanes. Crossover into the Median happens at the Marginal Rd intersection. This makes the transition between the Tremont St subway and the Washington St surface route seamless with no intersections or street running.
Herald St + Marginal Rd: Traffic lights at both to act as one mega intersection despite being split by I-90
William E Mullins Way and Traveller St: No left turns on Washington
Perry, Savoy, Rollins, Waltham, and Union Park: No left turns on Washington
Mystic St, Brookline St: No left turns on Washington
Concord St, Worcester St, Springfield St: No left turns on Washington
Northampton, Lenox, Newcombe, Thorndike, Ball Streets: No left turns on Washington
Eustis/Williams and Dade Streets: No left turns on Washington
Ruggles St: Extend through the parking lot to meet Harrison Ave, add traffic light
Make Washington transit only between Ruggles/Eustis, widen to 4 lanes between Ruggles/Warren.
So with all these changes made, this would bring the total number of intersections the line would need to cross between Nubian and Eliot Norton Park down from ~20 (Depending on how you count it) to 7. Even if you're less aggressive than this it's still possible to reduce the number of car crossings substantially.
To be clear: as shown on the map I included, I advocate for two branches to Nubian, for exactly the two groups you describe here: an I-93 el for Dorchester/Mattapan transfers, and a modern streetcar line for Washington St locals. Yes, SL4/SL5 are bursting at the seems and yes, they would benefit from the greater capacity of light rail (low floor or high floor -- thank you for finding those stats, by the way!) And yes, I agree that center-running bus lanes would be a big improvement even with the current rolling stock.
Fair point, but it would certainly seem like the better solution is to extend the Washington St Line down to ~Warren @ Quincy then to get all of Roxbury within walking distance of a direct rail service, not build a second line from Nubian to downtown. A Ceylon Park infill would also help with bringing more of Grove Hall into the Fairmount Line walkshed.
My response to @TheRatmeister's (very valid) point is in line with @Riverside's comment above: Why not both?
All three of us seem to agree on the usefulness of some rail service down Warren St (preferably to Grove Hall or even Franklin Park, but can stop at MLK Blvd to avoid the narrower sections of Warren St further south -- though the latter still leaves an 18-min walk to Grove Hall, longer than Four Corners/Geneva does). But in a world with two Nubian branches, one via a Washington St streetcar and one via an I-93 El, there's nothing stopping you from extending one of them to Warren St. (I myself now prefer extending the I-93 El, detailed below.)
Regardless of what you do south of Nubian, though, I think there are good arguments for two branches' worth of services to Nubian:
Capacity relief. As @TheRatmeister said in the earlier discussions, today's Silver Line -- which sees at least 50% of its ridership from South End -- already warrants capacity upgrades, and South End's density justifies it. A single branch is unlikely to be able to absorb more demand originating south of Nubian. (SL4/5 already runs twice as frequently as any Green Line branch, so a single branch with 2x capacity per vehicle doesn't change the crowding leve. Even 4x capacity is probably only enough for induced demand within South End.) Running 2 branches' worth of trains will likely be required if you want to attract riders from Nubian at all, even if they're all Washington streetcars.
Boosting capacity in the Tremont St subway, up to Gov't Center and North Station. In a GLR world, the segment from Park St to North Station suffers from the loss of Kenmore routes and reverse branching on the Huntington routes. Only two branches (D/E) come from the west. If you want to run the maximum capacity of 4 branches to GC and NS -- especially given ongoing discussions about this exact segment -- then 2 branches have to come from Nubian.
Technically, you can continue running the B and/or C branches to GC and NS, but that creates merges between Boylston Inner and Boylston Outer, defying the whole purpose of GLR.
Subsequently, if you have your mind set on running 2 branches' worth of trains, the argument for an I-93 El increases quite significantly, beyond just "faster service for Nubian":
Increasing reliability and reducing bunching, especially on Washington St. If 2 branches run on the surface, the segment will see trains every 3-4 mins. This seems too frequent for an urban corridor to handle, given the number of stops and crossings. (Reddit users were quite strongly against someone else's proposal to run such frequencies through a surface transitway.) Things become much more manageable with just one branch running every 6-8 mins on the corridor, focusing on local demand.
Clearer separation of "express" and "local" demand. Assuming my preferred alternative of "local" (Washington streetcar ending at Nubian) and "express" (I-93 El ending at Grove Hall via Nubian) routes, most people at or south of Nubian will take the express. If you're running 2 branches anyway, this frees up capacity on the local trains for South End. In addition to providing more capacity given a fixed frequency, it also reduces dwell times.
Comparing red to blue cells, there seems to be a curious disparity in that inbound trips see greater demand from South End (16% higher), while outbound trips see greater demand from Nubian (28% higher). But regardless, they're close enough. And given that a notable crowd do end their trip at intermediate stops without crossing the Pike, I suspect they're more likely to come from Nubian than other intermediate stops, which boost the share of South End riders even further.
Using the same methodology as the previous analysis, the following region in South End along SL4/5 has 10,248 job takers with a density of 16,344 jobs/mi^2. It has virtually the same population as Warren St, but compressed in half the space, making it twice as dense -- probably one of the densest in Boston. (These residents are also wealthier and whiter than Warren St, demonstrating the ongoing gentrification in South End.)
I know 11.3 mph does not seem like a stretch to you, but it still seems like a tall order to me. As @Teban54 illustrated, that would mean being as fast, if not faster, than the current subway. That seems like a strong claim.
Are there real-world examples of a surface LRT (or even BRT) service operating at the speeds you are suggesting here, in an environment similar to Washington St in the South End? The closest analogue in Boston is (I would argue) the C Line, which is nowhere close to sustaining the average 11.3 mph targeted here. But are there other examples elsewhere in the world?
Here in Berlin or here in Salt Lake City. Both are about 2 mile segments with ~10 minute travel times. Both feature 7-8 stops, close to the 8 or 9 that would make sense on the Washington St Line.
While I very much appreciate the insights of real-world LRT/BRT routes that can run fast through urban neighborhoods, I think it's worth being cautious about ideal vs. reality. Unless I'm mistaken, these seem like best-case scenarios, and just because they're achievable somewhere else doesn't mean we should expect the same in Boston (just like how extrapolating a best-performing short segment of the C branch to any branch in its entirety sounds equally unrealistic). When our existing streetcars already run 8.3 mph at best, I find it hard to believe a new extension can run faster; moreover, any efforts to make the Washington St branch reach 11.3 mph would probably be better spent speeding up the B/C/E branches, and the complete lack of incentives to do that makes me even more doubtful. (Especially given the MBTA appears to have a speed restriction of 25 mph along all street-running sections and 10 mph in mixed traffic, but that's an organization issue.) I'm willing to assume around 9 mph with reasonable improvements, but it would take a lot of work from the MBTA, even today, to make me believe in 11.3 mph.
Another factor is that a grade-separated I-93 El will probably run faster than you give credit for. As I've shown earlier, grade-separated LRT branches such as the D and GLX routinely achieve 20 mph, and the same is true for heavy rail as well. (As for why the Green Line trunk only runs 10-11 mph, I think it's due to both congestion and stop spacing, neither of which are issues on I-93 El.)
Using speeds of 9 mph and 20 mph respectively, with the common starting point of Washington/I-90, I arrive at time of 10.7 min on Washington and 6.3 min on I-93. That's a bigger difference than the "10 vs 8" you mentioned, and the 4.4 min difference is like moving the Ruggles bus hub to Stony Brook. (Even with 10 mph on the surface, it's still 9.7 min vs. 6.3 min for a difference of 3.4. Also, this does not consider possible differences in grade separation in the vicinity of Bay Village.)
Depends on how brutal you want to be. The general principle is that alleys never have left turns, nor do quiet residential streets if it doesn't severely limit accessibility. Main streets like Melnea Cass and Mass Ave obviously keep their intersections. Here's the changes I would make from a surface level glance: [...]
Some interesting thoughts here! I do think keeping Herald, E Berkeley, W Dedham/Reynolds, Newton and Mass Ave make sense. At least one of Brookline and Concord likely needs to be kept though, as they're one-way streets paired with Newton as well as offering access to parts of BMC. I also imagine most of these intersections will still see heavy pedestrian traffic, meaning you'll still need pedestrian crossings, but they're at least easier to manage than cars.
Beyond @Teban54's analysis (which I agree with), one significant shortcoming of the Fairmount Line is its lack of connectivity to the rest of the system. No connection to Green, Orange, or Blue. And South Station is good for access to downtown, but it doesn't cover everything. I do think that (near-)rapid transit frequencies on the Fairmount Line will be transformative in their own right, but I don't think they fill the gap being discussed.
If the Fairmount Line offers a faster, more convient way to get where they're going then they will switch. And I think it's very clear that the Fairmount Line will always be faster than a bus journey if the number of transfers is the same, which it usually is.
Because of the transfer most riders need to make at Nubian, the number of transfers remains the same or even fewer with the Fairmount Line. BL is 3 transfers once the Red-Blue Connector gets built, down from 4 via SL4/5, GL is 3 transfers, the same as SL/4/5 currently, and OL is still only one 1 transfer, since it basically never makes sense to switch to the SL for transferring to the OL, just stay on the bus you were already on to Ruggles or Jackson Square, or take the 30/31 to Forest Hills. If we assume GL-Reconfig, which is what this about, GL D/E is only 1 transfer with Fairmount, another improvement.
I think there are two different considerations here: (a) From residents' homes to rapid transit (including a Nubian "subway"); and (b) From the said rapid transit station to their destination.
For (a), I think one of @TheRatmeister's implicit assumptions is that people will always walk to a Fairmount Line station, compared to taking a bus to a non-mainline rapid transit station. I don't think that's a guarantee:
For riders along Warren St north of Grove Hall, a bus transfer to Nubian is highly like to take less time and be more convenient, mostly because of how far the region is from Fairmount. (This is true even if you ignore a possible GL extension to Grove Hall.)
Even for those along the 15, 23 (Washington) and 28 (BHA) corridors, I suspect a sizable number will still need to do "bus-Fairmount" as opposed to "walk-Fairmount" anyway. Because of differences in stop spacing, the average walk to a Fairmount Line station will naturally be longer than to the nearest bus stop. For riders west of Fairmount -- most of 28 and part of 15 -- they're likely agonistic.
Even for bus riders east of Fairmount, once they're on a bus, taking it all the way to Nubian (or a Grove Hall/MLK station) has less impact on convenience than getting off at a Fairmount station. While that indeed takes more time, it can be made up if (b) offers improvement, as we'll see below.
For (b), if we think of the rapid transit station as the starting point, a Nubian branch has a pretty notable advantage overall:
Blue Line: 1 transfer via Nubian, 2 (circuitous) transfers via Fairmount
Green Line: 1 transfer via Nubian, 2 transfers via Fairmount
Reconfigured Huntington trunk has 1 transfer for both, but I'd argue it's less important for similar reasons as OL South
Red Line: Both are 1 transfer, but these transfers are heavily geared towards RL North, where Nubian is more convenient
Transfers from SL4/5 to rapid transit at DTX far outweighs South Station (see below)
Route 23 riders have the option of going to Ashmont
Seaport: Both are 1 transfer in a GLR world, though Fairmount has an advantage without GLR
Commuter Rail: Clearly in favor of Fairmount
Downtown destinations: Generally, it appears that the Park/DTX area is more popular among SL4/5 riders today than South Station (see below). While it may be a result of imbalanced frequencies on SL4 vs. SL5, it may also be a chicken-and-egg problem. Plus, for areas north of Park/DTX, it's a clear win for Nubian.
(I'm omitting Orange Line here, because as you rightfully pointed out, most riders can stay on the bus to Ruggles as long as the bus routes are not truncated. Plus, for OL South, crosstown buses or future circumferential routes will be more convenient anyway.)
This is not to mention that virtually all Fairmount transfers will have to go through the most crowded section of the Red Line, paralyzing the busiest part of the system even further.
From the 2015-17 passenger survey. The last 3 columns are added for comparison. Stations who are listed in the same column were combined in the dataset, with no way to separate them.
Source
SL4 South Station
SL5 DTX and Boylston
SL4+5 Chinatown and TMC
SL4+5 Nubian
SL4+5 all remaining stops
Walk
311
1058
815
740
4787
Rapid Transit
128
982
1616
0
0
Bus or SL (Waterfront)
229
60
0
2406
164
Commuter Rail
155
0
0
0
0
Total (including driving etc)
868
2130
2445
3314
4963
In conclusion, I think in a world with a Nubian "subway", people who will choose Fairmount are mainly: those who are close to a Fairmount station, or at least substantially closer than a bus stop; and those who need a bus to either option, but are aiming at South Station or the Red Line. While I do think this would apply to a significant number of people (hence the need for Fairmount improvements), that's likely still lower than a simple "population within walkshed" approach would suggest.
