Green Line Reconfiguration

[TheRatmeister]

But kind of a nitpick: If a quad-tracked Copley-Arlington subway turns out possible, I think that may avoid the need for Pike Hugger West altogether. Not saying this is more desirable -- but I have indeed been placing a lot of thoughts on a quad-track Boylston subway lately.

I highly, highly doubt it. Copley is right under the Public Library, Trinity Church, and the Old South Church. Expanding it in any way is very likely to be politically, financially, and/or technically infeasible. The only way to do quad-track would be to close the station, and I don't think it really needs to be stated why that's not a great idea. As long as Copley can't be quad-tracked, this proposal is a non-starter. This seems to have been a generally accepted truth for basically the subway's entire existence as I don't think it has ever been officially proposed, unlike Park-Gov't Center.

• The distance disadvantage -- the most significant drawback of Full Pike Hugger -- is by far the least notable for Huntington-Seaport than for any other service pattern. I measure 1.33 mi from Huntington/Exeter to Essex/Surface via the Pike. That becomes a "whooping" 1.29 mi via Exeter-Boylston-Essex, and 1.23 mi via Stuart-Kneeland-Surface. If you say Full Pike Hugger is longer, I'm not feeling it.

I'm focused on the distance that needs to be tunneled under narrow streets. In the case of the Essex St alignment, that's the whole distance. For Pike Hugger, it's the distance from the Pike to the Transitway hook on Essex St. Essex works out to about 3800ft, Pike Hugger is around 3400 ft if you tunnel under Hudson St, or 3700ft if you tunnel under Harrison St, plus the Pike-hugging section. Even if we assume the pike-hugging section is three times as easy to build, it's still 1400ft from Tremont St, so total under-street distance would need to be ≤3300ft. You would definitely see some saving from not including stations, but also, now you're not including stations. There's now no 2SR or better from Seaport to Kenmore and Huntington (and beyond) without either a quad-track Pike Hugger or a new bi-level station where the Huntington Ave Subway meets the Pike. The workarounds really start to pile up when you take an indirect route.

In contrast, no matter how you slice it, Essex is by far the hardest to build among the "eastern half" options: no other alignment runs into building mitigation risks even after accounting for forced TBM.

I'm not convinced this is an avoidable problem. For one, you still need to end on Essex St somehow to hook into the transitway, so any solution, as far as I can tell, needs to be under Essex at least as far as Lincoln St for the steep climb after passing under the sewer. And if you're not starting at the Post Office Sq incline, you need to get there somehow. So you either need to navigate the deep foundations and narrow streets of Downtown, or underpin Big Dig tunnels, or maybe both. I think this moves the problem, rather than solving it.

More broadly speaking, I think we should avoid "working backwards" to cross out theoretical alternatives because of being too married to particular alignments. Hope this doesn't come as an offense, but the overall impression I got from your comment was that you started by assuming Essex is the default option for the eastern half. Everything else builds upon this assumption, and other eastern alignments only come into play when absolutely necessary. Frankly, a key reason why I even wrote this post is to minimize the influence of engineering details from high-level decisions.

My interpretation of your alternatives was to compare them with the technical details and possibilities that exist outside of theory. For right now, that means that (in my opinion) since the Essex St subway is (likely to be) the most cost-effective connection between any part of GLRC and the Seaport Transitway, any alternative that doesn't make use of that better bring something else to the table. I think this has long been understood to be the hardest part of GLRC, for many of the same reasons that SL Phase III ran into. You have to navigate a complex underground landscape of subway, sewers, foundations, and highways. There are many ways to do that, but without evidence to the contrary, I'm inclined to choose the one that is generally shorter, has fewer incredinly sharp turns, has an existing portal location available, and has had a significant amount of its underground complexity already mapped.

Also, I really don't think trying to avoid the Charles St subway is the way to go here. In terms of engineering, that's probably the easiest part of any of this, a short, mostly flat cut and cover tunnel under a (relatively) wide street is probably not where the cost would really pile on.

On most days, LMA seems like a bigger destination than Kenmore is.

LMA is absolutely the bigger destination. Which is why it's great that it can be reached both from Huntington, or from Kenmore. The real differentiator is the section between MFA and Prudential, that alone needs to justify the added cost of connecting Seaport service to Huntington compared to Kenmore.

So if Kenmore can justify 32 TPH, why can't Huntington?

Kenmore alone doesn't justify 32 TPH. A trunk that feeds into Brookline, Allston, Brighton, LMA, and Harvard does.

I believe there are very plausible builds for Alternatives 3/4/5 that allow Huntington to connect to Park/GC, nor do I think any of them requires flat junctions that absolutely cannot be replaced with flying junctions. Hopefully this will be made clearer once I get to elaborate in future posts.

If you think there are I'd be more than happy to see them. I just can't think of any way to do 3/5 without using the junction at Copley, which for the reasons I outlined earlier, I really do not think can be replaced with a flying junction at any reasonable cost. 4 is definitely possible, I just don't really see it being worth it between the full Pike Hugger subway and the required flying junction to merge Kenmore/Huntington trains into a single trunk.

 

[Tallguy]

OL to Nubian is the solution. Ridership on the southern half of the line drops off hard. Even with OLX to WRox, there should be sufficient space. A flying junction is nessary, but I think feasible

 

[Teban54]

@TheRatmeister, I'll start by saying that your thoughts have opened up quite a few new perspectives and considerations for me, which I truly appreciate:

• The practical considerations of modifying (existing) Copley, especially w.r.t. building preservation
  • This seems to push any Kenmore-BBY-BV proposal to Exeter more, compared to @Riverside's original idea; but I do think a K-S split west of Exeter is feasible
• The mindset of treating Charles St as a leverageable asset
  • In particular, you're right about its easy of construction (70' at its narrowest)
  • Though I'd point out that the Boylston junction still complicates things in various ways
• The fact that I had overlooked challenges of feeding into Park Inner

There are a few things that I want to push back on, however.


#1: Practicality of avoiding "Essex"

I'm not convinced this is an avoidable problem. For one, you still need to end on Essex St somehow to hook into the transitway, so any solution, as far as I can tell, needs to be under Essex at least as far as Lincoln St for the steep climb after passing under the sewer. And if you're not starting at the Post Office Sq incline, you need to get there somehow. So you either need to navigate the deep foundations and narrow streets of Downtown, or underpin Big Dig tunnels, or maybe both. I think this moves the problem, rather than solving it.

We should separate the following two concepts: "Essex East" east of the Big Dig, and "Essex West" west of it.

Essex East is unavoidable. So is the dive from South Station down under the Big Dig (namely the sewer).

But I'd argue that Essex West is entirely avoidable, via various Kneeland or Pike Hugger East proposals.

Also note that the Post Office Sq Incline merge -- or more accurately, the need to merge at some point X on the Central Subway -- does not apply for Alternative 3 (which doesn't have K-SP trains at all), as I'll elaborate below.


#2: "Cost effectiveness" of Essex West

(in my opinion) since the Essex St subway is (likely to be) the most cost-effective connection between any part of GLRC and the Seaport Transitway, any alternative that doesn't make use of that better bring something else to the table.

I really, really, really have to disagree with this. Is Essex West the shortest? Absolutely. Most cost-effective? At least, a lot more analyses need to be done.

Essex West has various proven engineering challenges, many of which are unique among (almost) all GLR infrastructures:

• Long sections under streets that are only 40' wide
• One or two stations that not only need to go down to Basement 3 level, but are also constrained to 40' wide streets
• Known cost blowouts in the past

Even though the recent, valuable contributions from you (and many others) on this topic have demystified some limitations of Essex, that's far from all. I believe that most factors above can't be eliminated (and that's before we even start discussing other drawbacks of Essex West that were not listed).

The only other street with a hard 40' limit in the GLR context is Hudson St south of Chinatown Gate. But:

1. For building Pike Hugger East alone, alternatives exist. In particular, I've been favoring the Lincoln St alignment more and more recently. If you can resolve the turn from Essex, everything else is easy - Lincoln St is almost as wide as Charles St.
2. This is also a good argument for Stuart-Kneeland, whether it's between Charles St-Chinatown or the full-length Copley-Chinatown. It's not just an "Essex wannabe", especially when street width is considered.
3. Even if you're forced to use Hudson, it doesn't seem as bad as Essex West is. You don't need to add platforms nor go down to Basement 3.

To me, these limitations of Essex -- and the inapplicability of most of them to other eastern alignments -- are more than enough to put alternatives on the table.


#3: Alternative 3 clarifications

This is more of a general theme: I noticed your understandings of Alternative 3 is way different from mine.

To me, Alternative 3 (H-SP, K-PI+GC, N-GC) is actually the easiest of all:

• Build the Full Pike Hugger, and operationally, have all Huntington trains run through it
• Nubian trains meet Huntington at the Bay Village interchange, then simply continue north via Pleasant St portal reactivation
  • Track connections between Huntington and Tremont can be easily added, allowing non-revenue connections
• Kenmore trains simply use the Central Subway to Boylston Inner
  • 16 TPH of Kenmore trains switch from Boylston Inner to Boylston Outer
  • The rest stay on inner tracks to Park St Inner

All relevant infrastructure had already been included in Tripod, which had been thoroughly discussed on this forum for 10 years. (In fact, even the Bay Village interchange is now, strictly speaking, optional.) This doesn't need Copley modifications, doesn't need a Charles St junction... It needs nothing more than Full Pike Hugger -- the only requirement.

Alternative 3 is far from perfect, but its issues are mostly with route design, not engineering. As long as you're not completely against Pike Hugger East, then Alternative 3 has little to no unresolved engineering challenges.

So yeah, I'm not sure what misconceptions you had that greatly complicated Alternative 3...? Was it a false assumption that Kenmore trains are forced to feed into Boylston Outer (as opposed to the red highlight above)?

 

[TheRatmeister]

On Essex:

Long sections under streets that are only 40' wide

I could be wrong, but for just the tunneled sections, this shouldn't be a problem. Especially if they're stacked, you have more than enough horizontal clearance to avoid any building basements or foundations, as long as they aren't under the street itself. It's when you get to stations that this becomes a problem.

This is also a good argument for Stuart-Kneeland, whether it's between Charles St-Chinatown or the full-length Copley-Chinatown. It's not just an "Essex wannabe", especially when street width is considered.

Stuart/Kneeland is better here, but it has the problem of being terrible for transfers. Any station at Tremont St would need a 2+ minute walk to connect at Boylston. (OL interchange would probably be fine, occurring at Back Bay however.) I think it's also quite unlikely that you'd be able to make a flying junction at Tremont & Stuart work, so you'd need to build both a subway under Columbus Ave to Stuart St, and an additional 1200ft Pike Hugger to meet the Tremont St Subway. (Unless you want to send all Huntington trains to Seaport but I will continue to argue that this is an extremely terrible idea.) You also likely need more stations, possibly as many as 4 (Bay Village, Statler Park, Emerson, Chinatown Gate). All except Bay Village also still need to be very deep to get under the Tremont St and Washington St subways and the Big Dig tunnels. It seems like the OL at Stuart St could be as deep as 40 feet rather than its shallow depth at Chinatown. So while you avoid some complexity from station width, depth remains challenging and is still likely to bloat costs.

If you can resolve the turn from Essex, everything else is easy - Lincoln St is almost as wide as Charles St.

It's a big if though. That's a lot of digging that needs to happen under the CAT. While obviously not as sharp as the street corner, the 'extended' curve could still be too sharp to be done with a TBM, and if that's the case that's probably enough to kill the idea right there given you'd essentially just need to mine out a massive area underneath the CAT. The other issue is that you have to be ready to climb like hell as soon as you get to South Street. It's already a steep gradient, shortening the available distance any more could necessitate grades over 6%.

So yeah, I'm not sure what misconceptions you had that greatly complicated Alternative 3...? Was it a false assumption that Kenmore trains are forced to feed into Boylston Outer (as opposed to the red highlight above)?

Maybe it was that, I'm not entirely sure. I was definitely tired though which did not help. Obviously it still has the issue of Huntington being cut off from Gov't Center and the GLX which I'd still consider to be a deal-breaker, but it's at least technically possible.

 

[stick n move]

We need some more drawings in these threads for when the concepts really start getting in the weeds. Trying to glance by here every once in a while and see what is being talked about can get pretty rough.

 

[Charlie_mta]

Long sections under streets that are only 40' wide

That didn't deter BERY from planning the Post Office Square tunnel along Essex St. If they felt they could do it, I have confidence it can be done:

 

[Stlin]

That didn't deter BERY from planning the Post Office Square tunnel along Essex St. If they felt they could do it, I have confidence it can be done:

View attachment 62103

Given that Post Office Square was proposed as the easier alternative to the now-unthinkable option of underpinning Park Street with a 3rd layer (basically creating a Red Line sandwich) I don't put a huge amount of stock in the constructability of BERY era engineering - their willingness to brute force a solution requires an elixir of courage and manpower that has been lost to time.

 

[The EGE]

There was also a century less of buried utilities and high-rises to complicate construction. And while the BERy's subway/surface transfer stations tended to be spacious, the downtown stations were often rather cramped, never mind modern needs like elevators and escalators and their machinery.

A lot of the downtown subways would not be considered feasible to construct today because design standards, construction safety standards, rolling stock size, and the built environment have all changed. That's why almost all the halfway-reasonable pitches on this board are combinations of existing/known alignments (Grand Junction, NSRL, disused Tremont Street Subway branch), urban-renewaled streets (Marginal, Cambridge), and wide streets suitable for cut-and-cover (Commonwealth, Huntington, Storrow).

 

[AndrewOnTheMBTA]

I think this is the best thread for this. I was walking to Silver Line Way thru the part that's built under the Congress St building, and I realized that almost all caternary is gone, and I hadn't picked up on it. So, battery BRT is here to stay?

Also, I can never not see this as a ripe spot for light rail conversion, at least for SL2. The toughest part would be going by the new South Boston Innovation and then Harpoon, if all of SL2 were to be LRT. But if it just went to SLW, are there any specific LRT conversion benefits, even if it wasn't as green line for the beginning (just a SL2 LRT)? Would GL type 10 trainsets fit fine?

 

[JeffDowntown]

I think this is the best thread for this. I was walking to Silver Line Way thru the part that's built under the Congress St building, and I realized that almost all caternary is gone, and I hadn't picked up on it. So, battery BRT is here to stay?

Also, I can never not see this as a ripe spot for light rail conversion, at least for SL2. The toughest part would be going by the new South Boston Innovation and then Harpoon, if all of SL2 were to be LRT. But if it just went to SLW, are there any specific LRT conversion benefits, even if it wasn't as green line for the beginning (just a SL2 LRT)? Would GL type 10 trainsets fit fine?

Big challenge, unless you connect the Silver Line LRT into the Green Line, is you would need a yard and maintenance facility. It would be an orphan line like the Mattapan Trolley. Which is really costly to support relative to its transit value.

 

[TheRatmeister]

Big challenge, unless you connect the Silver Line LRT into the Green Line, is you would need a yard and maintenance facility. It would be an orphan line like the Mattapan Trolley. Which is really costly to support relative to its transit value.

I mean, you could maybe connect it up to Cabot Yard with Track 61. Obviously Cabot would need new facilities for the new vehicles though, so I don't think it improves the value proposition very much.

 

[Teban54]

Big challenge, unless you connect the Silver Line LRT into the Green Line, is you would need a yard and maintenance facility. It would be an orphan line like the Mattapan Trolley. Which is really costly to support relative to its transit value.

This reminds me: What would be good locations for (new and expanded) yards for Green Line Reconfiguration?

This question still applies even without a connection to the Seaport Transitway (even though obviously most GLR proposals include one). Any GLR proposals come with service increases, so more trains will be operated and stored.

Very quick thoughts off the top of my head, that haven't been evaluated whatsoever:

• Riverside Yard expansion
• New yard near Needham Industrial Park
• New yard near Alewife Industrial Park, if the GLX Union Square branch is extended beyond Porter
• New yard in east Seaport with integrated development?
• Reactivation of Watertown Yard for rail, if a resurrected A branch goes all the way there
• Eastern Route ROW: Integration of MBTA Everett Shops and/or new yard in industrial Everett and Chelsea
  • This assumed any new Eastern Route / Grand Junction service has track connections to GLR, which is sometimes debatable

 

[The EGE]

Those all seem generally logical. My very quick add-ons:

• Riverside expansion is tricky because of the proposed development. There is a plausible case for double-decking the storage tracks, though.
• The inside of the Needham Junction wye was proposed in the 1945-47 Coolidge Commission plan. End-of-line yards are somewhat preferable, but the industrial park wouldn't be bad either. There's currently a tree trimming business located inside the wye, which would be on the easier side to relocate.
• I agree that an Alewife-area yard is only worthwhile for an extension past Porter. The Fitchburg ROW is pretty narrow west of Porter, so a tunnel as far as Sherman Street is likely needed - too much cost just for a yard. Porter is only a mile and a half beyond Union Square; the differential in needed LRVs is small enough that a new yard wouldn't be needed.

 

[737900er]

This reminds me: What would be good locations for (new and expanded) yards for Green Line Reconfiguration?

This question still applies even without a connection to the Seaport Transitway (even though obviously most GLR proposals include one). Any GLR proposals come with service increases, so more trains will be operated and stored.

Very quick thoughts off the top of my head, that haven't been evaluated whatsoever:

• Riverside Yard expansion
• New yard near Needham Industrial Park
• New yard near Alewife Industrial Park, if the GLX Union Square branch is extended beyond Porter
• New yard in east Seaport with integrated development?
• Reactivation of Watertown Yard for rail, if a resurrected A branch goes all the way there
• Eastern Route ROW: Integration of MBTA Everett Shops and/or new yard in industrial Everett and Chelsea
  • This assumed any new Eastern Route / Grand Junction service has track connections to GLR, which is sometimes debatable

Green eats Albany if you end up building over there and we get a mythical west side bus facility more fit for purpose. It's like 3x the size of Lake Street.

• New yard near Alewife Industrial Park, if the GLX Union Square branch is extended beyond Porter

Don't you sort of end up at Watertown Yard in this scenario?

I don't think Needham needs to be anything more than the what Medford/Tufts looks like today given how big Riverside is.

 

[TheRatmeister]

New yard near Needham Industrial Park

My personal favorite.

Green eats Albany if you end up building over there and we get a mythical west side bus facility more fit for purpose. It's like 3x the size of Lake Street.

That likely requires some amount of shuffling Albany routes over to Southampton assuming SL4/5 don't run anymore, but that's probably doable at least. One maybe possible way to get more yard space at Albany would be to stick some tail tracks under 93. You've got to work around the supports but I reckon it's still possible to fit 6 tracks that are each around 750ft long, which would get you space for roughly 36 type 10s.

 

[Teban54]

• The inside of the Needham Junction wye was proposed in the 1945-47 Coolidge Commission plan. End-of-line yards are somewhat preferable, but the industrial park wouldn't be bad either. There's currently a tree trimming business located inside the wye, which would be on the easier side to relocate.

I hadn't thought about using the Needham wye before. My initial impression was that it seems too small, but comparing it to Lake St Yard (lol) has now convinced me otherwise. So yeah, it does look better than Needham Industrial Park.

• I agree that an Alewife-area yard is only worthwhile for an extension past Porter. The Fitchburg ROW is pretty narrow west of Porter, so a tunnel as far as Sherman Street is likely needed - too much cost just for a yard. Porter is only a mile and a half beyond Union Square; the differential in needed LRVs is small enough that a new yard wouldn't be needed.

Yeah, this was under the assumption of extending the Union Sq branch to Belmont, Waltham, etc.

Don't you sort of end up at Watertown Yard in this scenario?

Somehow I wasn't considering Watertown Branch LRT at all, not to mention that such a branch (without Waltham) wouldn't even pass through Alewife anyway. But if a Watertown-Porter branch is built, then I agree, simply using Watertown Yard seems like the best bet.

My personal views on a Watertown-Porter branch are probably more pessimistic than most, with the main concern being the roundabout route. But that's less relevant to the discussion.

Green eats Albany if you end up building over there and we get a mythical west side bus facility more fit for purpose. It's like 3x the size of Lake Street.

Very interesting idea. Perhaps that's another argument in favor of an I-93 El to Nubian. While a surface connection from Washington St is feasible in engineering, I doubt it's politically feasible due to having to construct tracks on cross streets for no benefit to them.

 

[Riverside]

Since @Riverside made this post about the "Criss-Cross" GLR model almost a month ago, as opposed to the more traditional "Tripod" model, I had been doing some internal evaluations of the two, in an attempt to figure out which one is better. In this process, as hinted at in this comment of mine, I realized that we need a cleaner framework to reason about capacity.

So I've designed such a theoretical framework, which abstract away the engineering details (tracks, flying junctions, etc) and several operational details (e.g. outlying branches).

For the first time in almost a year, I actually managed to write a post on my own blog about it:


Routing the Green Line Reconfiguration (Part 1): A Theoretical Framework of Capacity and Service Patterns


TL;DR:

• The Green Line Reconfiguration (GLR) system likely has a theoretical maximum capacity of 80 TPH, up from today’s 32; but hitting this maximum is non-trivial.
• Most prior GLR proposals, namely the “Tripod” and the (naive) “Criss-Cross” models, only hit 64 TPH instead.
• Adding short turns terminating at Park to Criss-Cross gets you 80 TPH. It comes at the drawback of interlining — but as it turns out, interlining is inevitable for maximum capacity.
• There are (at least) 5 different service patterns that hit 80 TPH, not just limited to short-turning Criss-Cross.
• Evaluating them will be left for the future — but share your thoughts now!

View attachment 61661


Feel free to discuss and evaluate these alternatives or the framework itself! I plan to do my own evaluations at some point, but those will definitely be influenced by everyone else's thoughts.

As another reminder, I have a list of potential future blog posts in the last section of the article. Let me know which ones you are most interested in or want me to write!

I'm sorry that it took me so long to reply to this. Great analysis, and I really like the theoretical framework, particularly the graphs -- I've never been able to sketch them out particularly clearly, but they definitely align with my internal thought process on all this. (And an excellent overview of many of the moving parts here, including the practical questions of capacity, and what actual constraints capacity would and would not impose.)

For my part, I would not treat Nubian-Park St Inner (N-PI) as physically impossible. The crossovers for that already exist, and while it's true that it isn't physically possible to wholly isolate N-PI trains the way you can some of the other permutations, it's certainly well within the realm of feasibility.

I think that you excluded N-PI because there's no way to run those trains without having them do a flat crossing of other services, even theoretically. Annotating your graph:

Two of the crossings (green circles) are trivial. A H-PI + K-GC service could theoretically avoid crossings if the Kenmore trains do a wraparound via Bay Village while Huntington trains run via Arlington; in that scenario, Huntington trains would arrive at Boylston Inner while Kenmore trains would arrive at Boylston Outer.

But, absent one of the more extreme workarounds discussed upthread (e.g. rerouting Nubian trains under Charles St in order to access Boylston Inner from the west), any service from Nubian to Park St Inner would need to cross GC services -- no way (even theoretically) to avoid it. So I do see why you'd treat it a little bit differently from the others.

That being said, I disagree with this framing:

This is an example of interlining, where we have two largely independent lines (Brown and Gold), but some trains switch from one line to the other(Magenta).

• I myself call this the “N-merge”, both because the lines form an “N” shape (as seen in Figure 5), and also because one of the most prominent examples is New York City’s Broadway Line, where the N train switches between the local and express tracks. (This video shows delays on the Broadway Line due to the N merge.)
• Meanwhile, another kind of interlining is what I call the “X-merge”, where a fourth set of trains cross the two lines in the opposite direction, forming a full 2×2 service pattern. (In Figure 5, that would correspond to adding hypothetical Nubian-PI trains.)

An obvious benefit of interlining is convenience, as (some) riders get service to two destinations from the same origin. But it comes at a pretty significant cost: Reliability. If a (delayed) Magenta train and a Gold train arrive at the same time, only one of them can go ahead — and no matter how you do it, delays have been propagated from the “Brown Line” to the “Gold Line”.

I think describing this as "N-merge interlining" is a bit of a mischaracterization. In a situation like New York's, both northern branches continue (along separate paths) in revenue service. But here, the Park St Inner branch terminates, which signficantly reduces the exposure of the network. It's interlining, yes, but it's essentially just a "λ-merge", with some trains being pulled out of service before actually reaching the merge point. "λ-merges" are relatively simple; case in point, we don't see it as particular urgent to deinterline east of Kenmore, even though it also has a λ-merge.

All things being equal, it would be most parsimonious if the Nubian branch(es) could feed directly into Boylston Inner, and turn at Park St Inner without interruption. If we were building the system today, I think that's what we would do.

But since that's not possible, I think the alternative is to leverage the crossovers to create a flexible system:

With the restoration of a northbound track out of the Fence Track, trains from both branches can be shunted on to either the Inner or Outer tracks "live". This in turn means that trains can be held to restore headways, and can be short-turned ad hoc as needed. I think this would offer a good counterweight to the complexity of the λ-merge, enabling many of the benefits of interlining while offering an antidote for some of its pitfalls.

In any case, I know you weren't specifically arguing in favor of one proposal or another; I just think it's worth bringing Nubian-Park Street Inner into the mix, as it doesn't seem to me (at the moment) that there's particularly strong reason to exclude it.

(But yes, this is why I favor "Criss Cross with optional short-turns" -- 16 tph from GC to Nubian, 16 tph from GC to Huntington, with up to 8-16 tph additionally sent from Park Street Inner to Huntington, depending on need -- peak demand during the day, an extension to Hyde Square, etc -- with the remaining Park Street Inner slots being available for short-turning or holding trains to correct headway issues. Use both Inner and Outer tracks for all through-services, distributing as needed.)

 

[Teban54]

Great thoughts from @Riverside (and @TheRatmeister) earlier re: the GLR Capacity Framework. This reply of mine is venturing somewhat into the 2nd and 3rd posts in this series, but they're worth addressing now:


Nubian - Park Inner via original Tremont St?

For my part, I would not treat Nubian-Park St Inner (N-PI) as physically impossible. The crossovers for that already exist, and while it's true that it isn't physically possible to wholly isolate N-PI trains the way you can some of the other permutations, it's certainly well within the realm of feasibility.

I believe this is impossible with existing infrastructure, particularly for southbound. Consider the track diagram:

Any train from southbound Park St Inner only has one switch available, to the opposite direction (northbound Boylston Inner), circled in red. Otherwise, it's locked on to the Boylston St subway.

Northbound Boylston Outer trains can indeed switch to Park St Inner by reactivating the full diamond crossover circled in yellow. But doing so for southbound trains would require a new switch -- one of the several in pink below:

I haven't looked into the feasibility of installing such a switch (possibly by converting one of the Park Outer -> Boylston Inner switches to a diamond). As vanshnookenraggen didn't draw any provisions for it, it's prudent to not take it for granted -- even though there's a good chance it might work.


However, it turns out that your understanding is ultimately correct -- it doesn't matter, because a flat diamond crossover is still too restrictive:

I think that you excluded N-PI because there's no way to run those trains without having them do a flat crossing of other services, even theoretically.

Details below:

Spoiler: How does a full diamond crossover change the calculus?

The original N-merge illustrated in my post is more aptly described as a |/|-merge, where northbound trains switch from inner to outer, but not the other way round. Having a full diamond crossover in both directions would allow two other kinds of merges:

• |\|-merge, where northbound trains switch from outer to inner only. This means Boylston Outer gets more trains than Boylston Inner.
• A full X-merge, where trains may switch from either inner track to either outer track.
  • Note this can be a |X|-merge.

From the perspective of efficiency -- and hence capacity -- a full X-merge is still inefficient, even with diamond crossovers. Two trains that need to cross in opposite ways can't depart Boylston at the same time.

So we're left with |\|-merge... But it doesn't make sense operationally. You're removing trains to GC in exchange for extra trains to Park Inner, but then you're capped by our assumed capacity constraint of Park Inner.

Put it another way:

• If you have 16 TPH at northbound Boylston Inner and 32 TPH at Boylston Outer, your best bet is to just have all trains head straight without switching, anyway. Anything else would trigger an X-merge, which isn't supported.
• If you have 32 TPH at Boylston Inner and 16 TPH at Boylston Outer (or 24/24), then you only need (and is essentially only allowed to do) an |/|-merge.

Rebuttal against Riverside's "Criss-Cross with N-PI"

All things being equal, it would be most parsimonious if the Nubian branch(es) could feed directly into Boylston Inner, and turn at Park St Inner without interruption. If we were building the system today, I think that's what we would do.

But since that's not possible, I think the alternative is to leverage the crossovers to create a flexible system:

View attachment 62230

With the restoration of a northbound track out of the Fence Track, trains from both branches can be shunted on to either the Inner or Outer tracks "live". This in turn means that trains can be held to restore headways, and can be short-turned ad hoc as needed. I think this would offer a good counterweight to the complexity of the λ-merge, enabling many of the benefits of interlining while offering an antidote for some of its pitfalls.

The issue is that this doesn't allow full capacity. I've explained it in the spoiler above, but to summarize: This requires an X-merge that's not supported by diamond crossovers.

Assume that you have 32 TPH from Huntington and 16 TPH from Nubian. Under your proposal, all trains need to switch north of Boylston (Huntington trains from Boylston Inner to Park Outer, Nubian trains from Boylston Outer to Park Inner). All trains cross each other's paths, so you'd then need to slot every Nubian train in-between adjacent Huntington trains -- but if Huntington is operating at full capacity, there won't be any such gaps.

You need to grade-separate BI-PO and PI-BO trains somehow, so that they can depart Boylston together. Which is theoretically possible if you're willing to do the "Boylston-Park Flip", modifying the tunnels between Boylston and Park. But that's obviously well beyond your expectations of cost.


N-merge vs. λ-merge

@Riverside made a good point that Park Inner being a terminating loop. However, I'd argue that they're still not as "pristine" as standalone λ-merges -- the benefit is smaller than you might think.

My main thoughts are:

• Indeed, λ-merge (or perhaps Y-merge?)is pretty much okay -- and it's inevitable for any designs with branches
  • Even the NYC de-interlining fans are generally fine with it
• Essentially, this requires treating the northbound "Magenta D/E trains" in my Figure 6 not as "coming from a branch that's mixed with Brown N/R trains", but "coming from a branch with only Magenta trains but more prone to delays".
• However, the fact remains that Magenta trains are still more prone to delays than an isolated track with 16 TPH. Any delays from an earlier Brown train can then delay a Magenta train, at an earlier λ-merge.
  • The fact that nobody will board a northbound Brown train at Park St is irrelevant to Brown's effects on Magenta and Gold trains.

Where this gets a bit better is southbound: departing Brown trains on Park Inner can hold in the station until a (delayed) switching Magenta train has passed. But that's not delay-free, either: the Brown train would still be departing behind schedule.

The only concrete positives I see from this, compared to a typical N-merge, is primarily due to how short the "Park Inner branch" is. Northbound, nobody is taking a Park Inner train north of Park. Southbound, the Brown trains have more even headways due to this being the first stop, and you also avoid the optics of "a full Brown train carrying passengers from Somerville is held in the station while a Magenta train crosses ahead". But the Brown train can jolly well be full anyway, and its passengers still feel the delay.

Ultimately, I don't think this difference is substantial enough.

 

[Teban54]

How I didn't think through Park Inner enough

This discussion does reflect a high-level point though: My original graph didn't fully consider Park St Inner's connections.

Originally, I claimed that Kenmore-PI and Huntington-PI are feasible, whereas Nubian-PI is not. H-PI was primarily due to Riverside's proposal of having Huntington trains take over Arlington; N-PI was primarily due to geometry (of PI pointing "the wrong way"). However, the lack of the "pink diamond" at Boylston, as well as earlier discussions with @TheRatmeister, made me revisit this claim. The main reason: Huntington-Park Inner is less trivial than I thought.

Unlike GC, Park Inner only has one existing connection: Kenmore. I've argued above that Huntington-PI and Nubian-PI via the flat "pink diamond" lack real value. Absent that, here are all semi-realistic ways for Huntington-PI and Nubian-PI:

• (a) (H-PI only) Merge onto the Central Subway before or after Copley (at arbitrary Point X), and stops at Arlington
  • Requires K-SP (which is mandatory in any alternative with H-PI) to diverge out of Central Subway before then
  • This includes "quad-track Arlington"
  • Not applicable for N-BI
• (b) Build a new north-south tunnel around Public Garden, which merges into (eastbound) Central Subway
  • Options: Charles St; Arlington St; or Columbus Ave.
  • Still requires K-SP to diverge out of Central Subway before then
  • Charles St and Arlington St are plausible for N-BI
• (c) The "Boylston-Park Flip": Modify the quad-track layout between Boylston and Park, so that the two tracks in each direction are swapped in a grade-separated manner (with or without ability to switch)

All of them face additional engineering challenges if you want "Full Criss-Cross" (Alternative 5), particularly enabling all of: K-SP, K-PI, H-SP, and H-PI.

We can now imagine a few scenarios:

• World 1:We're not willing to build any of (a), (b) and (c)
  • Only K-PI is enabled, not H-PI or N-PI
  • Only Alternatives 2 and 3 are valid
• World 2:We only build (a)
  • K-PI and H-PI are enabled, but not N-PI
  • All 5 alternatives are fair game
• World 3:We build (b) or (c)
  • K-PI, H-PI and N-PI are all enabled
  • More alternatives now enter the picture, beyond the original 5

I believe all of (a), (b) and (c) have plausible builds. Some are easier than others, and it also depends on how much flexibility you want to allow. The details are still being worked out, and will be discussed in future posts.

 

[TheRatmeister]

Good points above. I will just say that I think Park Inner to Nubian (or Huntington) service is highly dependent on how the seaport connection is done. If you go from Bay Village to Seaport via Lincoln St or Hudson St, not only does I think that make the connection less likely, but also significantly less useful since you can just send the extra capacity that way. If however you build "SL Phase III modified", I think it becomes more useful as a way to unlock more service to Hyde Sq or extra branches beyond Nubian.