Green Line Reconfiguration

I forget where this has been discussed, but how much of a non-starter is it to quad-track the central subway (from kenmore to govt center?) The outside pair of tracks already exist from Boylston to Park St & at kenmore, and it would provide some capacity relief to the central core where all lines run through. Of course modernizing signals to moving blocks would probably also boost capacity significantly so trains can run closer together.. and this quad-track probably would be combined with some Pleasant St extension which won't be on the docket until at least 2060..

We'd need to get rid of Park St short turns (and bulldoze the loop), re-do the GL triangle platform at GC, fix the Boylston Curve, and build a duck-under junction for the westbound E-train towards Heath st. And it seems like Copley, Arlington, and Hynes would all be expensive as the tracks aren't deep enough under ground to build a mezanine.

I think the resources that would be required to do this would be better spent on other GL Reconfig infrastructure (D-E connector, Back Bay Subway) or adjacent projects like Blue Line to Kenmore.
 
I forget where this has been discussed, but how much of a non-starter is it to quad-track the central subway (from kenmore to govt center?) The outside pair of tracks already exist from Boylston to Park St & at kenmore, and it would provide some capacity relief to the central core where all lines run through. Of course modernizing signals to moving blocks would probably also boost capacity significantly so trains can run closer together.. and this quad-track probably would be combined with some Pleasant St extension which won't be on the docket until at least 2060..

We'd need to get rid of Park St short turns (and bulldoze the loop), re-do the GL triangle platform at GC, fix the Boylston Curve, and build a duck-under junction for the westbound E-train towards Heath st. And it seems like Copley, Arlington, and Hynes would all be expensive as the tracks aren't deep enough under ground to build a mezanine.
The BLX Red-Blue connector project would take a lot of pressure off of having to quad-track the GL between GC and Park. Quad tracking the GL between GC and Park was proposed in the early 1950s, but Tremont Street is too narrow to add two tracks to the existing GL tunnel. IMO, a separate two-track tunnel would have to be deep-bored on a route from Park station up Park Street and then heading east toward GC under the east side of Beacon Hill.
Note: edited to correct typos
 
Last edited:
I forget where this has been discussed, but how much of a non-starter is it to quad-track the central subway (from kenmore to govt center?) The outside pair of tracks already exist from Boylston to Park St & at kenmore, and it would provide some capacity relief to the central core where all lines run through. Of course modernizing signals to moving blocks would probably also boost capacity significantly so trains can run closer together.. and this quad-track probably would be combined with some Pleasant St extension which won't be on the docket until at least 2060..

We'd need to get rid of Park St short turns (and bulldoze the loop), re-do the GL triangle platform at GC, fix the Boylston Curve, and build a duck-under junction for the westbound E-train towards Heath st. And it seems like Copley, Arlington, and Hynes would all be expensive as the tracks aren't deep enough under ground to build a mezanine.

It's pretty much a total non-starter to quad-track the Boylston St subway in part (I believe) due to the abutting building foundations. (The T's work to build an elevator at Copley damaged a nearby historic building, an example of the perils of building on landfill.) The Granary Burying Ground poses similar problems between Park and Government Center -- it's just too narrow.

(Ironically, it may be surprisingly feasible to quad-track through GC itself -- as you say, cutting through the triangle platform -- and then a good chunk of GC <> Haymarket is already quad-tracked today, with the inner tracks leading to the Brattle Loop. And the old Haymarket station itself historically was also quad-tracked, and I believe the footprint is still sufficient to restore if needed.)

As @737900er alludes to, the way to quad-track the Boylston St Subway is to build the second set of tracks a quarter mile to the south:

1676076385914.png


By connecting the Huntington Subway to the Pleasant Street Portal, you can leverage the flying junction at Boylston instead of trying to build a new one at Copley. And by connecting the Huntington Subway to the Riverside Branch at Brookline Village, you can separate the branches evenly: B and C in the Boylston Subway, and D and E in the Huntington Subway. A full-length Huntington Subway would also mean that you can separate the "surface streetcar" legacy services on Commonwealth and Beacon (where you have bus stop spacing) from the faster and more reliable "modern light rail" services on the D and E, running in fully sealed ROWs.
 
(Ironically, it may be surprisingly feasible to quad-track through GC itself -- as you say, cutting through the triangle platform -- and then a good chunk of GC <> Haymarket is already quad-tracked today, with the inner tracks leading to the Brattle Loop. And the old Haymarket station itself historically was also quad-tracked, and I believe the footprint is still sufficient to restore if needed.)

I know this has been suggested before, though I still have a hard time envisioning how a quad-track GC would work.

Presumably the split to quad-tracking would need to be as close to the Park Street end of the station as possible (because otherwise what's the point if everything's going to back up in the station). Shiving an inbound/westbound track off the Brattle Loop (no relation) would seem to require removing/relocating both the staircase from the Green Line westbound platform and the Green-to-Blue elevator bank. It'd be a tight squeeze to fit both of them back in given how narrow the resulting platform would be (not to mention the possibility of ADA issues if you have to cross an active track to reach the Blue Line elevators, which might be unavoidable given how the Green and Blue levels interact).

It looks like it might be possible to fork the outbound/eastbound track without hitting any stairs or escalators (though nuking the Dunkin' bunker), though it'd kind of box in the Green-to-Headhouse elevators (and mean that getting from those elevators to the inbound side and the Blue Line level could require crossing two active tracks with conflicting movements. It's all probably possible but that station's messed-up geometry likely makes it harder than one might imagine.

As for Haymarket, there's definitely not space to quad-track the current location, that platform having been built on top of the inner tracks. The original was quad-tracked as you said, and a good chunk of it still survives, but how bad shape it's in (and how much of it got blown to pieces by the 90s-era portal relocation and subsequent subway construction) is a bit unclear.

By connecting the Huntington Subway to the Pleasant Street Portal, you can leverage the flying junction at Boylston instead of trying to build a new one at Copley. And by connecting the Huntington Subway to the Riverside Branch at Brookline Village, you can separate the branches evenly: B and C in the Boylston Subway, and D and E in the Huntington Subway. A full-length Huntington Subway would also mean that you can separate the "surface streetcar" legacy services on Commonwealth and Beacon (where you have bus stop spacing) from the faster and more reliable "modern light rail" services on the D and E, running in fully sealed ROWs.

Definitely agree with this. On top of the difficulty digging in the Back Bay (maybe that's why Copley and Arlington were construction sites basically the whole time I was in high school...man, I can still smell the dust), there's no station with the geometry for double-island platforms, so you'd end up merging the tracks (probably between Boylston and Arlington) and then just having to wait in line through the Boylston curve and platform. Re-using the Boylston flying junction also gets you the island platforms through to Park for some redundancy in case of issues and better schedule management.
 
A couple months back I spent a lot of time finally digging in to one of the thorniest parts of the Green Line Reconfiguration: the Aldgate junction proposed for Brickbottom Junction. For some reason, I never actually got around to posting what I came up with. So, I'll look back at my notes, and see what we can do here...

There are 4 journeys Brickbottom Junction needs to support in order for all this to "work":
  1. Lechmere <> Medford
  2. Lechmere <> Union
  3. Cambridge (Grand Junction) <> Sullivan
  4. Cambridge (Grand Junction) <> Lechmere
(That last one is potentially optional, if you can find someplace for a Cambridge <> Sullivan LRT line to have a transfer to either/both of the Medford or Union Branches. I'm skeptical that's doable, and I think there is merit to a North Station <> Lechmere <> Kendall route.)

(And yes, I am excluding Lechmere <> Sullivan; as I've detailed elsewhere, I don't believe that this route is the best use of resources, and while the Lechmere -> Sullivan piece is very easy to build, the Sullivan -> Lechmere piece will be nasty, and either be inordinately inexpensive [tunneling or super-elevateds] or introduce bottlenecks [flat junctions].)

1. Lechmere <> Medford and 2. Lechmere <> Union

These are handled by the existing GLX infrastructure, which should require little modification.

4. Cambridge (Grand Junction) <> Lechmere

1676125683824.png


Going westbound, you reclaim some/all of the Somerville Avenue Extension road alongside the ROW as it crosses under McGrath Highway, branching off the Union Branch at the corner of the Brickbottom Artists' Association building, and start diving down below grade level (shown in dotted green here). I'm not sure of the exact depth an underpass would need to be in order to safely cross under the Union Branch and Fitchburg Line; that said, a 6% grade reaches a depth of 20 feet in 333 feet of running distance, which just about fits in under the McGrath overpass. Once underground, there should be enough space for a gentle curve (i.e. 80' radius or better) that avoids the foundation of the nearby building before tunneling under and then surfacing within the Grand Junction ROW.

If 20 feet isn't deep enough for an underpass, we can achieve 660 feet of running space to enable a 40 foot depth by cutting under the McGrath Highway embankment to swing wide into a question mark shape. Even with 85'-radius curves, it should be possible to avoid the building foundations. See the route in red (shown in solid color, but would be a tunnel for most/all of its length):

1676125641530.png


Going eastbound, we are basically going to clone the Union -> Viaduct incline and put it south of the mainline tracks. Now, this will require a little bit of rearrangement. There are currently three mainline tracks between McGrath and the Viaduct (let's call them North, Middle, South):

1676126390909.png


Vanshnookenraggen's track map is helpful for illustrating what's going on here; Middle and South are the through tracks, while North is a stub-end:

1676126447475.png


And a photo I took from an inbound train on the Viaduct also helps, somewhat:

1676126496633.png


As can somewhat be seen in the above photo and more clearly visible in the StreetView, there is available space next to the South track (possibly with some relocated support pillars). But, as the ROW continues east, it becomes constrained by the buildings at the eastern corner of the lot (as visible in the photo).

So I propose using the wiggle room available underneath the McGrath overpass to shift the mainline tracks "up" one "slot".

Before:

1676127501351.png


After:

1676127550060.png


This opens up the slot of the South track to create the new incline, and join the Viaduct just south of the current merge (new incline shown as a dashed line):

1676127649080.png


Note that you still will need to fit the new incline under the Community Path. I think this will be doable, though it may be necessary to relocate the Viaduct merge further south than I've shown it here (if you need to duck lower than I've estimated to clear under the Path, in which case you need additional running length to rise up to meet the Viaduct, which should still be doable).

You'll also need to do a little bit of rearranging of the mainline tracks on the eastern side of the Viaduct, including weaving through some support beams, but it shouldn't be anything super aggressive. In the diagrams, light purple is used for the non-revenue mainline tracks, mid-purple is the typical revenue mainline tracks, and dark purple is new track that would need to be built to reconnect things (and the light purple fill is the yard, where I didn't draw out all the tracks).

Before:
1676128356516.png


After:

1676128476535.png


(Spoiler alert, this screenshot also includes new LRT tracks to support Grand Junction <> Sullivan, but I'll get to that.)

At this point, you now have a grade-separate flying junction that supports Lechmere <> Grand Junction service without creating crossing conflicts with other services.

3. Cambridge (Grand Junction) <> Sullivan

Half of this one is very easy, and half of it is very much not. Let's start with the easy one.

Going westbound this part of the junction is already built, in the form of the yard lead running from the GLMF to the Union Branch. With some modifications to the yard tracks, you have your revenue tracks connecting Sullivan to the Grand Junction (via the underpass discussed above) right there.

If you want to go all out (or include in a subsequent enhancement build), you could look at adding a parallel track north of the current track, connecting the current yard lead to the diverging track at McGrath, so that Sullivan -> Cambridge trains don't need to interfere with Lechmere -> Union trains. The problem here is that you would be right up against the back walls of the adjacent buildings, and I'm not sure how close you can actually put an active RoW against a building like that. The parallel track is valuable, as this stretch will otherwise see all:
  • Lechmere -> Union
  • Lechmere -> Cambridge
  • Sullivan -> Cambridge
trains, so this stretch could become a bottleneck. But I think it may still be workable.

1676152954497.png


Going eastbound is where things get hairy. Coming from Cambridge, this track needs to cross over or under:
  • McGrath Hwy
  • The Cambridge -> Lechmere incline
  • The Fitchburg Line
  • The Community Path
  • The BET yard leads
  • The Union -> Lechmere incline
  • The Viaduct
  • The Lechmere -> Union incline
  • A potential future Lechmere -> GLMF yard lead incline
Running flat at-grade will be a non-starter, unless you plan to sink the Fitchburg Line tracks, which will be both cost- and space-prohibitive. Running elevated looks more possible at first, but would need to weave under the Community Path, over the Viaduct, over the Lechmere -> Union incline, and then somehow interface with the Lechmere -> GLMF incline (probably needing to pass over it altogether). Alternatively, one could "go wide", remaining at surface level south of the Fitchburg Line from one side of the junction to another before rising into a separate viaduct over the BET. In both cases, however, I think that you can't make the necessary height changes within the short distances while still maintaining a 6% grade.

So, it's a tunnel. A short tunnel, but a tunnel, taking what I think is the most direct (shortest) route, but which also happens to cut right through the heart of Brickbottom Junction. Hopefully it could be dug in stages to reduce disruption to Fitchburg service, but it definitely will not be pretty. On the other side, it surfaces to join one of the tracks repurposed from the GLMF Yard to continue to Sullivan. (In the diagram below, the dashed blue line indicates a surface track that could connect to a Lechmere -> GLMF incline, if ever built.)

1676153951523.png


If memory serves from when I first sketched this out, the paths I've drawn should give enough running distance to drop 20 feet below surface by the time the tunnel hits the mainline tracks. On the eastern end, there is plenty of space to lengthen that if needed. It's a bit trickier on the western end, but I think it should be possible, if needed, to relocate the Cambridge -> Lechmere/Sullivan divergence point further west in order to provide additional distance to fit a drop to a greater depth than 20 feet.
 

Attachments

  • 1676152872517.png
    1676152872517.png
    1.8 MB · Views: 42
  • 1676155281868.png
    1676155281868.png
    1.6 MB · Views: 47
Summary and Conclusion

The good news is that this design can be built in stages: the Lechmere <> Cambridge leg can be built more or less independent of when the Sullivan <> Cambridge leg is built. The bad news is that I don't see a way to run Cambridge -> Sullivan without some tunneling.

Broadly speaking, if the T is willing to buy up some real-estate and/or use a TBM, it could be possible to build a flying interchange west of McGrath Highway, and then leverage the yard leads going from the Union Branch to the GLMF. This would be a simpler but likely costlier design, and would create bottlenecks going both eastbound and westbound. A rough sketch (don't look too hard at the turn radii, this is just the general idea):

QuestionMark.png


But, aside from that, my thinking comes down to:
  • A Lechmere -> Cambridge underpass at/west of McGrath (medium)
  • A Cambridge -> Lechmere incline south of the mainline tracks (hard)
  • A Sullivan -> Cambridge connection via the existing yard leads (easy)
  • A Cambridge -> Sullivan underpass in the center of the junction (very hard)
 
I recently did a “SWAGful” analysis of the estimated North South Rail Link costs. As it happens, I actually ran those calculations in order to have some points of comparison for this analysis: estimating the cost of the Green Line Reconfiguration.

(Be forewarned, this post will also be highly “SWAGful”.)

One key difference between the NSRL and GLR is the latter’s ability to be decomposed down into smaller projects that are viable (or approach viability) on a standalone basis. So, first, I’m going to put forward such a decomposition, with some short-hand names:
  • “Magenta Line” (LRT between Seaport and Back Bay)
    • Back Bay - South Station
    • South Station - Transitway
    • Bay Village junction
  • Huntington Subway
    • Northeastern - Brookline Village
      • (numerous alternatives where the subway is shorter)
  • Nubian Line
    • Bay Village - Washington St portal
    • Surface branch on Washington (or subway via BUMC)
  • Grand Junction Line
    • McGrath Highway - BU Bridge
    • Brickbottom junction
    • BU Bridge junction
  • Allston/Harvard Branch
    • Harvard - BU Bridge
    • BU Bridge - Fenway
      • (optional depending on final network design)
  • Chelsea Branch
    • Green Line Maintenance Facility - Sullivan
    • Sullivan - Chelsea
    • Chelsea - Airport
  • Needham Branch
    • Newton Highlands - Needham Junction
  • Commonwealth Subway
    • Kenmore - St Paul portal
For the most part, all of these component projects can be done without dependencies on the others. Some will, however, have dependencies in order to achieve their full potential; for example, the Magenta Line will be much more successful if it can also take advantage of a Huntington Subway. But for minimum viability, the ordering of projects is largely fungible.

Build sequence aside, the objective here is to estimate cost. The advantage of the fundamental pieces of the GLR that have coalesced over the years is that few will require herculean feats of engineering. The majority of the ROWs are at surface level; the majority of the tunnels are cut-and-cover under wide streets, or streets where we believe there are few mysteries beneath the surface. (The Back Bay - South Station subway is probably the “riskiest” component in this respect.)

I argue that, as a result of this, we can craft meaningful albeit broad estimates based on route length and tunnel vs surface, with a handful of special projects to give extra cushion for unusually complicated pieces of infrastructure (e.g. the Bay Village junction).

Broken out into the pieces listed above, these figures come out to:

ProjectSegmentTypeLength (miles)
MagentaBack Bay-South StationTunnel1.4
South Station - HarborviewSurface1
Bay VillageSpecial
HuntingtonNortheastern - Brookline VillageTunnel1.7
WashingtonPleasant St - Washington StTunnel0.4
Nubian BranchSurface1.6
ChelseaGLMF - SullivanSurface0.5
Sullivan - ChelseaSurface2.75
Chelsea - AirportSurface2.2
Grand JunctionMcGrath-BU BridgeSurface2.3
BrickbottomSpecial
BU BridgeSpecial
AllstonHarvard - BU BridgeSurface2
BU Bridge-FenwaySurface0.6
NeedhamNeedham BranchSurface3.7
Comm AveKenmore - St PaulTunnel0.9

Now to discuss cost-per-mile coefficients.

The most readily available case study for us is GLX: at $2.3B for 4.3 miles, it was roughly $535M per mile. So we could just say that we should expect the cost to be about $0.5B per mile for surface tracks and call it a day.

But, I went and looked at as many as I could of the modern LRT projects recently opened or under construction in North America and, perhaps unsurprisingly, GLX is a modestly extreme outlier. The only project I found with a similar cost-per-mile was Los Angeles’ newly-opened K Line (sometimes called the Crenshaw/LAX Line), whose cost so far has come out to about $470M per mile. But, notably, this project included a 1.5 mile subway (about 67% deep bore and 33% cut-and-cover), which significantly increases the expense.

In decreasing order, I found that the following projects all had costs-per-mile between $100M and $275M, all (I believe) primarily surface running:
  • Seattle, East Link Extension
    • $271M
  • Los Angeles, Expo Line Phase 2
    • $227M
  • Washington DC, Purple Line
    • $210M
  • Ottawa, Confederation Line
    • $200M
  • San Diego, MidCoast Trolley Extension
    • $191M
  • Houston, Green & Purple Line
    • $152M
  • Montreal, REM
    • $122M
  • Los Angeles, Foothill Gold Line Extension Phase 2B
    • $119M
  • Seattle, Hilltop Tacoma Link Extension
    • $118M
I should note that there is a long tail of additional projects beyond this; many of those are really streetcar projects rather than modern LRT, but there are extensions in both LA and Sacramento (among other places) that saw per-mile costs comfortably under $100M. Still, I think the projects I’ve listed above – in particular the Los Angeles, San Diego, and DC projects – are pretty similar to what we are proposing with the Green Line Reconfiguration.

So I am going to suggest that $250M per mile is a reasonable cost-per-mile coefficient, understanding that some projects will be cheaper and others more expensive.

The other piece of this are the tunnels. American subways are infamously costly to build. Looking at recent projects (with Crossrail thrown in for comparison), cost per mile:
  • New York, East Side Access
    • $3.7B
  • New York, Second Ave Subway
    • $2.5B
  • London, Crossrail
    • $1.5B
  • Los Angeles, Regional Core Connector
    • $0.9B
  • San Francisco, Central Subway
    • $0.9B
  • Los Angeles, Purple Line Extension Phase 1
    • $0.9B
  • Seattle, University Link
    • $0.6B
  • Seattle, Northgate Link
    • $0.4B
That is quite a spread. It’s worth noting that most of these comparison projects are deep bored whereas most of the proposed tunneling for GLR would be cut-and-cover, which Alon Levy suggests may reduce costs by up to half.

I’m somewhat arbitrary going to choose a tunneling coefficient of $1.5B per mile, and try not to worry about it too much. While I’m modestly optimistic about our ability to control costs on surface projects, I think our particular combination of landfill, colonial-era narrow streets, and large number of historic buidings makes it more likely that tunneling costs will be higher than we might hope.

(In theory, we could look at the SF Central Subway, which was deep bored, and say, “Well, we’re mostly going to be cut-and-cover, so perhaps we can get that down to $0.5B per mile!” But equally we could look at either NYC project, apply the same logic, and get a cost of ~$1.4B per mile.)

Either way, tunneling accounts for only 21% of the total route miles of the projects listed above; their cost obviously will have an impact, but not necessarily a narrative-changing one.

[continued below]
 
[continued from above]

When we add the coefficients to the table above, we get the following:

ProjectSegmentTypeLength (miles)Cost ($B)Cost per mile ($B)Subtotal ($B)
MagentaBack Bay-South StationTunnel1.4$2.1$1.5
South Station - HarborviewSurface1$0.25$0.25
Bay VillageSpecial$0.25
$2.65
HuntingtonNortheastern - Brookline VillageTunnel1.7$2.6$1.5
$2.6
WashingtonPleasant St - Washington StTunnel0.4$0.6$1.5
Nubian BranchSurface1.6$0.4$0.25
$1.0
ChelseaGLMF - SullivanSurface0.5$0.15$0.25
Sullivan - ChelseaSurface2.75$0.7$0.25
Chelsea - AirportSurface2.2$5.5$0.25
$1.4
Grand JunctionMcGrath-BU BridgeSurface2.3$0.6$0.25
BrickbottomSpecial$0.25
BU BridgeSpecial$0.25
$1.1
AllstonHarvard - BU BridgeSurface2$0.5$0.25
BU Bridge-FenwaySurface0.6$0.15$0.25
$0.65
NeedhamNeedham BranchSurface3.7$0.8$0.25
$0.8
Comm AveKenmore - St PaulTunnel0.9$1.4$1.5
$1.4
Surface Subtotal16.65$4.2$0.25
Tunnel Subtotal4.4$6.6$1.5
Special Subtotal$.75
TOTAL21.05$11.55

To more easily summarize it per project, I’ll list out the subproject subtotals here (though, again, bear in mind that these are highly imprecise estimates):
  • “Magenta Line”:
    • $2.7B
  • Huntington Subway:
    • $2.6B
  • Nubian Line:
    • $1.0B
  • Grand Junction Line:
    • $1.1B
  • Allston/Harvard Branch:
    • $0.7B
  • Chelsea Branch:
    • $1.4B
  • Needham Branch:
    • $0.8B
  • Commonwealth Subway:
    • $1.4B
It’s also worth noting that some of these projects are more “required” than others. For example, the Commonwealth Subway would be highly valuable for increasing speed and reliability, but isn’t critical to any particular project. Likewise, the Chelsea Branch is the most expensive “radial” extension, but isn’t necessarily required for the overall network. (And probably will be seen in different light over the next 10 years as we see the effects of a Silver Line Extension.)

Likewise, some of these projects might have smaller builds – a Chelsea Branch that terminates in Chelsea rather than Eastie, or a Huntington subway that paritally utilizes a surface connection to Brookline Village. Dropping Comm Ave and the Chelsea Branch lowers the estimate to $8.75B. I’ve also tossed in $250M per “special” project (i.e. the Aldgate junctions) to vaguely account for their complexity, but those could be overestimates.

Some of you can probably see where I’m going, based on having heard similar numbers recently.

For comparison: a proper NSRL (4 tracks, 2 southside portals, 2 stations) would be 6.6 miles of tunnel construction:
  • Core (tracks 1 & 2): 2.3 miles
  • Core (tracks 3 & 4): 2.3 miles
  • Back Bay lead: 0.5 miles
  • South Bay lead: 0.7 miles
  • Fitchburg lead: 0.4 miles
  • Sullivan lead: 0.4 miles
At 6.6 miles, using the same estimation method used above for the GLR projects with a tunnel cost-per-mile coefficient of $1.5B, that version of the NSRL comes out to $9.8B.

Which is to say: for the rough cost of a proper and useful NSRL, we could likely do a full-build Green Line Reconfiguration. These are megaprojects of broadly similar cost.

Now, it’s true that we could build a leaner NSRL with two tracks rather than four, which would drop the estimated cost to $6.35B. But we could also further drop the Needham Branch ($0.8B) and shorten the Huntington Subway to Brigham Circle, with a surface connection to Brookline Village (shaving that sub-project down by $0.9B), which would drop the GLR cost down to $7B.

(And note that TransitMatters optimistically estimates that the entire commuter rail network can be electrified for $800M-$1.5B, meaning we can probably add about $1B to the NSRL costs, which, once again puts NSRL and GLR neck-and-neck.)

For comparison: the pre-covid systemwide ridership on the commuter rail was a bit more than 75K riders per day, while the subway segment alone of the Green Line was comfortably above 85K riders per day, if not higher.

So, where does this leave us? A few concluding points:
  • The “full build” of the Green Line Reconfiguration would probably be in the neighborhood of $10B
  • A “minimum foundation build” of the GLR, consisting of the following projects, could be roughly $6B
    • Magenta Line
    • Huntington Subway to Brigham Circle with surface connection to Brookline Village
    • Grand Junction Line
    • Allston Branch
  • While there are a range of possible costs for either megaproject, in general a GLR will be comparable to an NSRL in cost
There is a larger conversation to have here, but I think it’s worth considering: we may only have the bandwidth to successfully advocate for one megaproject in this generation. I submit that the Green Line Reconfiguration may be more tractable, more feasible, more successful, and ultimately help more people.
 
On the question of deep bore versus cut-and-cover, I think that needs a more thorough economic evaluation.

I am convinced we opt for cut-and-cover here because our local trades cannot execute deep bore. So the profit (and skim) goes elsewhere.

Your excellent description of the tunneling conditions here:
"I think our particular combination of landfill, colonial-era narrow streets, and large number of historic buildings makes it more likely that tunneling costs will be higher than we might hope."
would seem to argue for more deep bore -- because we simply run into a lot of obstacles near the surface.

Also cut-and-cover is highly disruptive of surface operations. So for the duration of the project, surface transit is massively displaced. That has a real economic cost that is not captured in the project budget, but hurts the region economically during construction.
 
Summary and Conclusion

The good news is that this design can be built in stages: the Lechmere <> Cambridge leg can be built more or less independent of when the Sullivan <> Cambridge leg is built. The bad news is that I don't see a way to run Cambridge -> Sullivan without some tunneling.

Broadly speaking, if the T is willing to buy up some real-estate and/or use a TBM, it could be possible to build a flying interchange west of McGrath Highway, and then leverage the yard leads going from the Union Branch to the GLMF. This would be a simpler but likely costlier design, and would create bottlenecks going both eastbound and westbound. A rough sketch (don't look too hard at the turn radii, this is just the general idea):

View attachment 34248

But, aside from that, my thinking comes down to:
  • A Lechmere -> Cambridge underpass at/west of McGrath (medium)
  • A Cambridge -> Lechmere incline south of the mainline tracks (hard)
  • A Sullivan -> Cambridge connection via the existing yard leads (easy)
  • A Cambridge -> Sullivan underpass in the center of the junction (very hard)
Riverside,

Looking at your great idea for a junction of the Gold line and Green Line, I realigned it to avoid some large buildings and to enlarge the curve radius. The tunnel (shown in black) would curve out under McGrath Hwy, but avoid the bridge footings for the existing McGrath Hwy bridge over the GL and Fitchburg Div RR.

52699194974_5653ee69b0_o.jpg
 
Riverside,

Looking at your great idea for a junction of the Gold line and Green Line, I realigned it to avoid some large buildings and to enlarge the curve radius. The tunnel (shown in black) would curve out under McGrath Hwy, but avoid the bridge footings for the existing McGrath Hwy bridge over the GL and Fitchburg Div RR.

52699194974_5653ee69b0_o.jpg
Thanks @Charlie_mta! Yeah, I think that could be one possibility as well. It feels "big" or "roundabout" on paper, but probably the curve radii are gentle enough that trains would be able to move through speedily. The remaining downside is that it still creates a pinchpoint with all of those services going through the section between the viaduct and McGrath Hwy at once, but that may just be something to live with. Also, one advantage of your alignment here is that it would be relatively easy to later convert this into a fully separated crossover with an el like I was talking about in Crazy Transit Pitches, so there's some modest future-proofing here.
On the question of deep bore versus cut-and-cover, I think that needs a more thorough economic evaluation.

I am convinced we opt for cut-and-cover here because our local trades cannot execute deep bore. So the profit (and skim) goes elsewhere.

Your excellent description of the tunneling conditions here:
"I think our particular combination of landfill, colonial-era narrow streets, and large number of historic buildings makes it more likely that tunneling costs will be higher than we might hope."
would seem to argue for more deep bore -- because we simply run into a lot of obstacles near the surface.

Also cut-and-cover is highly disruptive of surface operations. So for the duration of the project, surface transit is massively displaced. That has a real economic cost that is not captured in the project budget, but hurts the region economically during construction.
No doubt there needs to be a more thorough economic evaluation (though I'll point out that the numbers I've presented here likely are more in line with bore rather than C&C, given that, I believe, most of the comparable projects I cite were also bored -- the numbers "work" even in the more expensive scenario).

Realistically, we are only talking about 2 tunnels that would be built for GLR: an extension of the Huntington Ave Subway, and a subway between Back Bay and South Station. I'll break these into 6 sections, going west to east:
  1. South Station <> Harrison Ave (ish)
  2. Harrison Ave <> Berkeley St under Marginal Rd
  3. Berkeley St <> current Huntington Subway via Back Bay Station
  4. Extending Huntington Subway under current reservation
  5. Brigham Circle <> S. Huntington under narrower Huntington Ave
  6. D-E Connector
Sections 1, 3, and likely 6 are going to be constrained by surrounding infrastructure/environment such that I'm guessing the choice of tunneling method will be dictated simply by what's possible. Section 1 worries me for how much it will need to thread through the spaghetti of highway tunnels, but should all be shallow enough and under open streets enough to use C&C. Section 3 will probably be ugly (unless we demolish the Garage, which, you know, I wouldn't be opposed to), but, again, I think engineering will outweigh economics on that one. And I think the big question on Section 6 is what the impact of the Muddy River will be.

For Section 2 under Marginal Road, I'm not particularly worried about the surface impact of cut-and-cover. There are almost no businesses directly abutting the street, and I think all of the housing has access from other streets. Reconfigurations to Herald St and Oak St could make up for the temporary loss of eastbound traffic lanes. And part of the value of that alignment is because we believe the fill under the street is "clean", which makes it an ideal candidate for cut-and-cover.

So the big questions remaining are Huntington Ave. Section 4 is slightly less acute because digging under the reservation can reduce the impact on the rest of the street, but it is true that there would still be significant impact.

In general, I am skeptical of deep bores, especially along alignments that are directly under city streets. One downside of this approach has become very clear with the opening of Grand Central Madison and SF's Central Subway: deep bore stations are deep. Like, "adds most of five minutes to your walk" deep. It's one thing to have a random station here and there which is that deep (e.g. Porter or WMATA's Wheaton station), but deep boring an entire corridor and burying the stations 7 or 8 stories underground -- that seems like a pretty negative drawback: significantly more expense for a system that is significantly less friendly to riders.

Of all the above, Sections 5 and 6 are the one place where I could probably be convinced to do a deep bore: between Brookline Village and Brigham Circle, you're likely to only have one subway station, so that reduces the impact.

But doing a deep bore under the reservation on Huntington seems penny-wise-pound-foolish to me. I don't deny the impact of surface disruption, but there also is impact from the current status quo, and I argue you dilute the impact of the new subway anyway by burying it so deep.
 
On the question of deep bore versus cut-and-cover, I think that needs a more thorough economic evaluation.

I am convinced we opt for cut-and-cover here because our local trades cannot execute deep bore. So the profit (and skim) goes elsewhere.

Your excellent description of the tunneling conditions here:
"I think our particular combination of landfill, colonial-era narrow streets, and large number of historic buildings makes it more likely that tunneling costs will be higher than we might hope."
would seem to argue for more deep bore -- because we simply run into a lot of obstacles near the surface.

Also cut-and-cover is highly disruptive of surface operations. So for the duration of the project, surface transit is massively displaced. That has a real economic cost that is not captured in the project budget, but hurts the region economically during construction.
Clarendon St Garage demo is already planned, is it not?
 
The tunnel (shown in black) would curve out under McGrath Hwy, but avoid the bridge footings for the existing McGrath Hwy bridge over the GL and Fitchburg Div RR.

52699194974_5653ee69b0_o.jpg

Given that Somerville seems poised to give the initial OK a lab project on that triangular parcel north of the main line, this could actually be a really clever way to shoehorn in this connection in on the site's northern tip, given that there's no reasonable way an Urban Ring project comes together fast enough to do anything more than get the lab developers to leave space in between their foundations.
 
Given that Somerville seems poised to give the initial OK a lab project on that triangular parcel north of the main line, this could actually be a really clever way to shoehorn in this connection in on the site's northern tip, given that there's no reasonable way an Urban Ring project comes together fast enough to do anything more than get the lab developers to leave space in between their foundations.
Yeah, I figured there had to be a development proposal for that lot. That's where planning and futuring comes in.
 
Yeah, I figured there had to be a development proposal for that lot. That's where planning and futuring comes in.
Actually I was going off this development plan, which shows that parcel as a parking lot:

I, too, am disheartened by the 588 parking spaces included in this project. However, it appears as though this is one of the locations for "consolidated" parking for the whole Brickbottom neighborhood.

Looks as though 100 Chestnut's parking count is 229. This brings the running total for Brickbottom to 817. Not a great start for TOD imoh. Hopefully because these two sites are included in the consolidated parking plan for Brickbottom, there will be significantly less going forward (save the Washington St. @ McGrath corner parking site).

I've been really happy to see the inclusion of an E. Somerville CR station in the 'small area plan' images! Though, it's even more reason to limit parking.

View attachment 34409
 
I think I recognize that image from @LexSEDotVille's post is from the city's Brickbottom planning exercise, which AFAIK hasn't been codified into zoning yet. And, in any case, we know how much planning studies are worth once a developer gets a Big Idea that city officials take a shine to...
 
The BLX Red-Blue connector project would take a lot of pressure off of having to quad-track the GL between GC and Park. Quad tracking the GL between GC and Park was proposed in the early 1950s, but Tremont Street is too narrow to add two tracks to the existing GL tunnel. IMO, a separate two-track tunnel would have to be deep-bored on a route from Park station up Park Street and then heading east toward GC under the east side of Beacon Hill.
Note: edited to correct typos
The space beneath Tremont is so tight that you could hear the green line during the shows at the Beacon Hill Cinema! I caught a show there in the early 90s.
Untitled.png
 
A few comments on this wonderful analysis:

For example, the Commonwealth Subway would be highly valuable for increasing speed and reliability, but isn’t critical to any particular project.
I actually think that in a full-build GLR, the Commonwealth Subway will be more important than shown here. Depending on build configuration, up to 4 branches may traverse the stretch at what's now BU Central: Boston College, Oak Square, Harvard, and Grand Junction. That's the same number of branches as present-day Copley-GC (which is already a bottleneck on today's Green Line), but even worse, running at street level for 0.7 miles with no grade separation and possibly no signal priority.

To make things worse, each of the 4 branches will likely involve some street-running, with the possible exception of Harvard. Especially BC/Oak Sq, and especially if Grand Junction service originates all the way from Chelsea or Airport. That's a huge reliability concern that may be even worse than today's Green Line (since the D has full grade separation).

(An implicit assumption here is that Urban Ring uses Commonwealth Ave to reach Kenmore and/or turn south somewhere. This may be challenged - for example, a surface alignment via Mountfort St and Park Dr is possible. However, that will further increase the complexity of BU Bridge junction, as mentioned below.)

I've been brainstorming this section of the Urban Ring/GLR for a bit, but I think even a 4-track Green Line subway isn't too crazy, especially if we want to insulate Urban Ring (Gold Line) patterns from Green/"Emerald"/Kenmore lines. A 2-track subway is workable, but a 2-track surface route may lack enough reliability to add any branches.

Another consideration here is the BU Bridge junction. You need the junction to enable services to Packards Corner, West Station and Grand Junction. A surface intersection at the current BU Bridge junction doesn't make the cut to me given the existing traffic patterns, and a viaduct may have difficulty descending onto the BU Bridge ROW. So you need a tunnel for the junction even in a smaller build - and if that's the case, it may be an additional argument for just building the full Commonwealth Subway once, instead of in two stages (and disrupting traffic and B branch operations twice).

Likewise, the Chelsea Branch is the most expensive “radial” extension, but isn’t necessarily required for the overall network. (And probably will be seen in different light over the next 10 years as we see the effects of a Silver Line Extension.)
We'll see how ridership evolves after SL3 extension and SL6 are implemented, but if they're huge successes, it may be an argument in favor of LRT conversion. LRT does miss out on Glendale and Upper Broadway in Everett (and has less convenient pedestrian access from Sweetser Circle), but the Silver Line alternatives analysis mentioned that the Glendale-Kendall alternative and Chelsea-Kendall alternative have similar projected ridership, so there's good potential for LRT and BRT to complement each other on this corridor.

  • Huntington Subway
    • Northeastern - Brookline Village
      • (numerous alternatives where the subway is shorter)
I know you probably didn't include any southside Urban Ring infrastructure towards the GLR cost estimations (e.g. Nubian to Huntington), but I think it might be a good time to revisit The EGE's idea of a D-E connector via Longwood instead of Brookline Village.

One thing you illustrated well here is how "expensive" a subway D-E connector is - about 0.9 miles from LMA station to Brookline Village. Tunneling via Longwood Ave (or Francis St but I'm doubtful on that) is even shorter on paper at 0.7 miles. Of course, a Longwood subway will be much more expensive due to building mitigations, but I think this is where a full cost-benefit analysis is needed.

And as the exchange between you and JeffDowntown shows, if deep bore turns out to be the choice for a subway D-E connector anyway (due to environmental constraints), it further reduces the disadvantage of a Longwood subway. You can do C&C from Northeastern to LMA or Brigham Circle (plus a portal further south for the E), and deep bore under one of the Longwood streets to Highland Branch. You'll likely have only one station within LMA proper anyway (maybe 2).

You did make a good point about time to exit a deep-bored station. This may end up with a situation where the time savings from having to walk from current D/E stations to the LMA hospitals may be outweighed by time to reach the surface from a station at the heart of LMA.

(Edit: I also think that an excessive amount of street-running to Riverside/Newton Highlands at any time is likely politically infeasible, since the D today is fully grade-separated. In fact, you might get complaints even with a short street-running section, say from Brookline Village to Brigham Circle. )

  • Allston/Harvard Branch
    • BU Bridge - Fenway
      • (optional depending on final network design)
I'm assuming you're using a street-running route via Mountfort St and Park Dr. It can work in a minimum build, but I'm concerned about reliability, given traffic, several signalized intersections, and the low likelihood of getting any dedicated ROW. This is in addition to adding even more complexity to the BU Bridge junction - now you have two possible directions on the southeast side.
 
Last edited:
Java King found quite a detailed document on proposals for a Green Line to Nubian (and Mattapan), in his post here.
 

Back
Top