Crazy Transit Pitches

[TheRatmeister]

Wait, I'm sorry, can you clarify, are you suggesting that automating the Orange Line is the better deal, or building a Sullivan <> GJ <> Longwood service is better?

I'm suggesting that the UR is the better deal.

 

[ritchiew]

@TheRatmeister Thanks for pulling out that data. But I'm not sure I follow some of your conclusions

Therefore, I think it can reasonably be said that if an Everett Subway was built, Wellington ridership would plummet, Sullivan would suffer both in terms of bus transfers and subway ridership but it probably wouldn't be catastrophic, and Malden Center would be basically unscathed.

In that hypothetical, I'm not sure what's wrong with ridership numbers dropping at those specific stations. The goal isn't to keep ridership numbers high at Wellington Station. The goal is to move people from where they are to where to need to go. If an Everett Branch captures a lot of those riders, that's a good thing.

That would generally support the conclusion that branching the OL with basically anything worse than 90 second trunk headways is not a great idea.

I don't really understand how this follows from what you said. And I'm really not sure where the 90 second cutoff is coming from. Pre-covid, Malden was getting 4-5 minute headways. That's the current target again, once the new signaling is done. That would seem to be totally sufficient for Malden. So if there were an Everett Branch, why wouldn't it be sufficient to have 2-2.5 minute trunk headways? I don't want to sound like I'm nitpicking here, but that difference could be big. Running 1.5 minute headways will for sure require total automation. But 2.5 minute headways might not. The difference in required infrastructure could be $100s of millions.

I know the MBTA studied at least CBTC for the Red Line before they started this latest signaling project and found it wouldn't be worth it. I thought they did a similar study for the Orange Line, but I've never seen the specifics of that. Does anyone know where to find that?

90 second headways basically demand full automation which has proved to be very expensive. The full automation of the 10 mile long Paris Line 1 (About the same length as the OL) cost €750,000,000 (inflation adjusted),

Calling a project "expensive" gets trickier when talking about automation like this. In the long run, operating costs dwarf the capital costs, and automation is a way to drastically reduce those huge operating costs. Full automation and driverless trains would be a massive upfront investment. But if we then run twice as many trains, all day, with almost none of the associated labor costs, that's a massive win in the long run.


Just to be clear, I'm not really advocating for an Everett Branch of the Orange Line. I would not be surprised if it couldn't work out. But you seem a little too quick to dismiss it, and I don't really follow your logic why.

 

[Riverside]

Therefore, I think it can reasonably be said that if an Everett Subway was built, Wellington ridership would plummet, Sullivan would suffer both in terms of bus transfers and subway ridership but it probably wouldn't be catastrophic, and Malden Center would be basically unscathed.
In that hypothetical, I'm not sure what's wrong with ridership numbers dropping at those specific stations. The goal isn't to keep ridership numbers high at Wellington Station. The goal is to move people from where they are to where to need to go. If an Everett Branch captures a lot of those riders, that's a good thing.

I think the point is that continued service to Malden Center is the highest priority, as the data suggest that its ridership demands would be unchanged by an Everett Subway (as opposed to Sullivan, which might become somewhat less critical, and as opposed to Wellington, which would become a much less important station if there were an Everett Subway).

That would generally support the conclusion that branching the OL with basically anything worse than 90 second trunk headways is not a great idea.
I don't really understand how this follows from what you said. And I'm really not sure where the 90 second cutoff is coming from. Pre-covid, Malden was getting 4-5 minute headways. That's the current target again, once the new signaling is done. That would seem to be totally sufficient for Malden. So if there were an Everett Branch, why wouldn't it be sufficient to have 2-2.5 minute trunk headways? I don't want to sound like I'm nitpicking here, but that difference could be big. Running 1.5 minute headways will for sure require total automation. But 2.5 minute headways might not. The difference in required infrastructure could be $100s of millions.

One question to ask is whether 4-5 minutes for Malden is/was sufficient. And I genuinely don't know the answer to that. I found a 2016 MPO Study that looked at crowding on the T. Their Orange Line section (infographic) says:

Northern Section​

The northern section of the Orange Line brings commuters from the north including Charlestown, Everett, Somerville, Medford, Malden, and Melrose. Many of these commuters connect with buses, notably at Wellington, Sullivan Square, and Haymarket.

As seen in Figure 11, in the Base Year the Orange Line is overcrowded during the last half hour of the AM peak period between Sullivan Square and Downtown Crossing, and reaches an unacceptable level of crowding between Haymarket and State. Base-Year crowding is much less severe during the PM peak period, as seen in Figure 12.

The widespread regional growth projected for the No-Build scenario will add substantial ridership to the Orange Line, significantly increasing the Base-Year crowding patterns. Crowding will be unacceptable between North Station and Haymarket for an entire hour, and between 8:30 and 8:45 AM between Assembly and Chinatown.

Many of the 72 large-impact projects summarized in Table 5 and shown in Figure 4 will be served directly by the Orange Line. These include Assembly Row, Assembly Square, North Point, West End, Old Boston Garden, Downtown Crossing, and Northeastern University. The combined impacts of these projected developments would impact the Orange Line severely if it is still operating with today’s capacity, as depicted Figure 11. Indeed, if peak-period Orange Line capacity is not expanded meaningfully, then more than half of the congested situations in the AM peak period will be congested at the unacceptable level. The reciprocal congestion during the PM peak period mirrors the AM peak congestion but at a slightly lower level of severity, as seen in Figure 12.

The MBTA is planning to procure new Orange Line vehicles jointly with the new Red Line vehicles, which will expand its vehicle fleet by about 25 percent, and vehicle capacity by 10 percent. With better equipment utilization resulting from a much lower average vehicle age, the Orange Line might be able to move 40 percent more passengers during peak periods. This increased capacity will reduce the duration and severity of crowding greatly, but even this amount of added capacity would not eliminate unacceptable levels of crowding completely.

The Orange Line vehicle fleet could be expanded further and operated safely beyond what is planned currently. However, with a more ambitious expansion program, equipment storage and maintenance capacity also would need to be addressed.

Spoiler: explanation of infographic

The three scenarios, Base Year, No-Build and Build appear at the top. The three AM peak-period hours are shown below each scenario name, and beneath the hours, there are arrows indicating the direction of travel. Alewife station appears at the top of all Red Line figures, but the direction of travel, indicated by the arrows, is away from Alewife in Figures 7 and 10 and towards Alewife in Figures 8 and 9.

Within each one-hour column are four positions where a crowding icon might be placed. The crowding icons are shown in the legend and indicate levels of crowding specific to Red Line vehicles as defined in Table 15. For instance, in the Base Year for travel between Central and Kendall/MIT crowding is apparent, with more than 114 passengers per vehicle between 8:30 and 8:45 AM, and between 8:45 and 9:00 AM. Two icons indicating “Overburdened” are placed in these two positions.

More crowding icons appear between 8:00 and 9:00 AM in the No-Build scenario, reflecting substantial regional growth and a large increase in Red Line ridership, as shown in Table 13. In this scenario, crowding between Central and Kendall/MIT begins at 8:15 AM, and between 8:30 and 8:45 AM there are more than 160 passengers per vehicle, indicated by the “Unacceptable” crowding icon. Detailed information about 2040 No-Build scenario passenger loads by station-pair and 15-minute interval are found in Appendix G.

The Build scenario includes trips generated by the 72 large-impact projects in the 20 sample TAZs. Both the duration and severity of AM congestion in this direction are greater than in the No-Build scenario; and the “Overcrowded” icon appears, indicating that the average vehicle is carrying more passengers than the “acceptably full” level shown in Table 15. Detailed information about 2040 Build scenario passenger loads by station-pair and 15-minute interval are found in Appendix H.

Based on the above, it sounds like 4-5 min headways were indeed sufficient for Malden, which is good news for an Everett Branch (although the data do also suggest significant crowding at Assembly, which does point to a need to branch north of there, rather than at Sullivan). If 2.5-min headways can be achieved in the core, then yeah, a branch could be doable.

2.5-min headways in the core, though... man I just don't know. (And really, if we are matching "4-5 minutes" at Malden, that points to 2 minutes in the core.) I'm not saying that it can't be done, but it also seems like it's far from certain that it can be done.

 

[jbray]

It feels like there should be better consensus on the order of projects in the advocacy community or even what projects. We don't even have consensus at to whether the grand junction in Cambridge should be urban ring or radial. Here are the projects (as I see them), and I'm going to rank them in the timeline by start date (there's no reason they can't be concurrent):

Major "New" Projects
North-South Rail Link
East-West
Urban Ring (Grand Junction all the way)
South Coast Phase 2
Double Tracking the Old Colony Pinch (Cape Rail?)
Stuart Street Subway
South End/Roxbury/Blue Hills Ave Rail Corridor (Nubian)
Chelsea Subway (to North Station Only?)
Everett Subway (to Sullivan Only? North Station via New Rutherford? OL Branch? Cambridge centered line?)
Congress Street Subway (Chelsea? to...? Nubian? RedX?)

Extensions
Red-Blue Connector
Blue to Lynn
GLX to Route 16 (West Medford?)
Roslindale OLX
RLX to Arlington Heights
GLX to Porter/Harvard
OLX West Roxbury/GLX Needham
RLX to Mattapan
RLX to Lexington
BLX Kenmore/Salem

Some of those are permutations but you get the idea.

 

[F-Line to Dudley]

2.5-min headways in the core, though... man I just don't know. (And really, if we are matching "4-5 minutes" at Malden, that points to 2 minutes in the core.) I'm not saying that it can't be done, but it also seems like it's far from certain that it can be done.

This is the intractable problem. You've gotta get transfer dwells in the core down before you can be talking headways like that. And with Red and Blue both aiming for 3-minute headways to flush downtown even fuller with riders, very old and access-constipated State and DTX won't be able to get Orange's doors closed on those transfer swells in time to clear the adjacent signal block without cascading delays at peak. The systems that use automation to achieve incredibly dense frequencies usually have their transfers spread over a LARGE number of stations so the transfer swells aren't singularly severe at any one of them even at max loading, because at that level of service density literally any dwell delay will mess with the tighter-than-tight frequencies. We've got the unfortunate gravitational singularity in the Big 4 downtown transfer stations, with a largely structurally unimprovable egress situation making the flushing of a platform between trains much more difficult. The only way to make it possible within those constraints is building radial transit first to take a big bite out of the transfers by redistributing them across the system.

Like...I could see Orange shooting for 1.5-2 minutes if we built the UR to Sullivan + Ruggles and/or if NSRL really gutted the Back Bay-North Station core with lots more people staying on Purple Line run-thrus. But the Red Line pretty much faces a 3-minute frequency cap because of Harvard curve (and likely gets hammered with tons of brand new core-to-Cambridge ridership in an NSRL universe), so the transfer swells at DTX are always going to be a bit outsized and impose a practical limiter on how fast you can get Orange's doors closed at peak. It honestly might not be enough for sustaining OLT base service levels to major bus hubs on multiple branches.

 

[Tallguy]

View attachment 59606
Keep in mind this does not take into account bus-bus transfers, and doesn't isolate only the Everett stops, just routes that go through Everett. I highly doubt that 65% of Sullivan's ridership comes from people coming from Everett, for example. Here's the list of routes for each station:

Sullvan: 104, 105, 109 (Pre BNRD just keep that in mind)
Wellington: 97, 99, 105, 106, 110, 112
Malden Center: 97, 99, 104, 105, 106

General conclusions:

• If we assume that 50% of bus riders at Everett transfer to buses to go to Kendall, while 50% go to Downtown or Longwood via the OL, then Sullivan is roughly 1/3 dependent on Everett ridership, although potentially less now that the 104 has been rerouted.
• No matter how you slice it Wellington is highly dependent on bus transfers from Everett. An Everett subway would pass through Sweetser Circle and no-doubt steal much of this ridership.
• That 20% figure for Malden Center is probably very optimistic. It includes the 105 and 106 which I would highly doubt many people are riding from Everett on. If they are omitted then it drops down to 11%.

Therefore, I think it can reasonably be said that if an Everett Subway was built, Wellington ridership would plummet, Sullivan would suffer both in terms of bus transfers and subway ridership but it probably wouldn't be catastrophic, and Malden Center would be basically unscathed. That would generally support the conclusion that branching the OL with basically anything worse than 90 second trunk headways is not a great idea.

I think the fact that an Everett Subway could easily connect to an Urban Ring serving Kendall and Longwood also hurts the merit of such a proposal, 90 second headways basically demand full automation which has proved to be very expensive. The full automation of the 10 mile long Paris Line 1 (About the same length as the OL) cost €750,000,000 (inflation adjusted), while the construction of the Line 14 extension cost around €360,000,000 per mile. The Sullivan-Longwood segment of the Urban Ring (Via the GJ) would be about 4.2 miles, for a price of roughly double what it would cost to automate the Orange Line, assuming the construction costs scale evenly from Paris to Boston. That seems like a better deal personally.

The line presently had a six minute frequency at best. Three minute trunk frequency would mean that Malden would get the same frequency it gets now. And, as much of the Everett ridership would be coming from Sullivan and Wellington, there should be significant overcapacity on the line as well. 90 second frequency is overkill

 

[TheRatmeister]

If we assume a normal capacity of 900 pax per train, and 10 TPH, then in 2019 the northern OL was around 2/3 of its maximum capacity of what it would have as a branch, assuming three minute headways on the trunk. If we use the crush capacity of 1350 pax per train it was at a bit less than 50% of its max capacity. In other words, my required 1.5 minute headways estimate was too pessimistic, 3 minute headways on the trunk would be fine.

But the headways have to be 3 minutes or less. If they drop to 4 minutes on the trunk, then after Community College the trains would be eating into their crush capacity. So, can the OL handle three minute headways? I'm not sure. According to the TM data dashboard, in September of 2019 dwells as high as two minutes were common in rush hour even with headways of <5 minutes. That would suggest that no, it's not feasible and the maximum OL headways are more like 4-5 minutes. If that's the case, there is essentially no (comfortable) capacity to spare on the OL for continued growth after a branch, assuming that passenger numbers return to 2019 levels at some point. If headways are more like 5 minutes, even at crush loads there's less than 25% of the line's capacity left. That's not a lot.

 

[Riverside]

On the topic of a subway through Everett, I do want to draw our attention back to the question of "where would the stations be?", because I think that will drive these decisions more than we might expect (like I described upthread).

 

[Riverside]

It feels like there should be better consensus on the order of projects in the advocacy community or even what projects.

So I'm gonna be difficult here and nudge a little bit on this; why should there be a better consensus?

From my perspective, having multiple projects that are "advocacy-ready" is beneficial because it allows us (as advocates) to respond to evolving political situations and public sentiments, capitalizing on moments as they come. I do think there's need for consensus around the most immediate next project, and I think we've got that: Red-Blue Connector, supplanting (at long last) GLX. But it's not obvious to me that consensus beyond that is necessarily desirable (to say nothing of likely not being possible).

That all being said, it's always fun (and worthwhile) to discuss and compare potential projects. To that end, I'd like to suggest a "creative constraint": set a limit of overall route-miles, and rank projects within constraints of that list.

I've been working on a blog post, analyzing the different eras of transit expansion and construction in Boston, and I think they are illustrative (these do not look at mainline rail/commuter rail):

• Initial Builds (1897-1922): 22 miles of ROW, all greenfield
• First Extensions (1923-1950ish): 7.5 miles, mostly reusing existing ROW
• Post-War Suburban Expansions (1950ish-1987): 34.8 miles of new transit, overwhelmingly reusing existing ROW
• [Big Dig {1990-2000ish}: Notable because it sucked all of the oxygen out of the room]
• Post Big Dig (2000ish-present): 6.25 miles, mostly existing ROW

So, an interesting (and I'd argue useful) constraint to a discussion about ranking projects is to ask ourselves which era we want to compare ourselves to. The heady days of Initial Builds where 22 miles of grade-separated transit was built over 25 years? The reactionary days of First Extensions or Post Big Dig, which saw less than 8 miles each of new construction over similar timeframes? Or the Post-War Suburban Expansions, which saw rates similar to Initial Builds but almost exclusively reusing existing ROWs?

(I'll try to post some thoughts on project rankings with these constraints later in the week, but for now I need to go to bed!)

 

[TheRatmeister]

On the topic of a subway through Everett, I do want to draw our attention back to the question of "where would the stations be?", because I think that will drive these decisions more than we might expect (like I described upthread).

Here's the 109 outbound data. If we just put stations at the 'spikey' stops, then the stations would be at:

• Dexter St (Encore)
• Sweetser Circle
• 2nd St/School St (Everett Sq or renamed, I vote Carrington after Walter Carrington.)
• High St/Hancock St
• Ferry St (Glendale)
• Lynn St (Woodlawn)
• Eastern Ave (Faulkner?)
• Linden Sq (Linden)

Thoughts:

• High St/Hancock St is probably redundant. If the Everett Sq/Carrington stop is moved up to Everett City Hall, that leaves a pretty even stop spacing between Sweetser Circle, Everett Square/Carrington, and Glendale stations.
• As you get towards Linden ridership drops, and these station sites become more speculative, designed for future development rather than existing ones.
• I'm not sure that Linden and Faulkner are both necessary, if Faulkner is eliminated then a more direct route could be taken to Linden.

 

[ritchiew]

On the topic of a subway through Everett, I do want to draw our attention back to the question of "where would the stations be?", because I think that will drive these decisions more than we might expect (like I described upthread).

Hehe. I brought up the idea of automation as I was thinking over your question. I seem to have derailed the conversation with that. Sorry about that.

I think the Orange Line deviation is a neat idea, but agree with most of the criticism. Here are some general thoughts:

I'll guess you've come across this, but here's a nice article on stop spacing. He bases most of his logic on how far people are willing to walk. He says there is a sharp drop-off in ridership past a 1/4 mile walk for slower, more local transit, like a bus. In that article he says people are generally willing to walk 3/5 mile for rapid transit. In the most recent edition of his book Human Transit, he changes that to 1/2 mile for rapid transit. (The book is wonderful, btw. I highly recommend. Get the new 2024 edition.) Everett isn't a super dense downtown where you have to also consider shorter spacing to spread heavy passenger loads across multiple stations. This also isn't where you'd want big park-and-rides, which would justify wider stop spacing. So using walking distance is pretty good model to figure reasonable stop spacing, I think.

Assuming people will walk 1/2 mile to get to a transit stop, then ideal stop spacing is..... there is no ideal stop spacing. It depends on specific goals, values, thinking on ridership v coverage, costs, and on and on (really, I recommend the book Human Transit). But I'm going to throw out 3/4 miles as a pretty decent heuristic. That creates some wasteful overlap in walkshed. But it also catches more people off the direct corridor who would otherwise be too far from a station. Natural station spots are unlikely to happen at exactly 3/4 mile intervals, so I'll give a range. Following the logic in that article, stop spacing of less than 1/2 mile in a place like Everett is pretty bad. That would create lots of wasteful overlaps in the walksheds, but add little extra coverage. Having stops that close together at that extreme would require some real justification. Stops further than 1 mile would also be bad. That would leave a coverage gap along the main corridor. That would also need some justification, like a big drop in density.

Then for actually picking stations in that range, there are a few other things to consider. One is multimodal access, which really drives ridership. That might mean prioritizing a place where you could build a whole bus station. At the very least it means prioritizing intersections that could easily facilitate busses, which means the intersection of two or more major streets. If Everett were a grid, that could be just abound anywhere, but it's really, really not. All those smaller, shorter residential streets that don't connect very easily would be a pain for lots of buses.

The other thing to consider is trip generators. Areas with businesses generate more trips then areas that are more purely residential. Schools, hospitals, government buildings, etc draw more. (These two factors, bus access and business proximity, are pretty big. I've been thinking about that since the recent quiz on least busy Red Line station. Shawmut and Fields Corner aren't that far apart, but Fields Corner has double the AM ridership.)

(continued)

 

[ritchiew]

With that long winded explanation of my thinking out of the way, here's how I would evaluate station locations:

Everett Square - Yup. Easy. Dense residential. Apartment buildings. Businesses. City Hall and other government offices. Medical clinics. Good bus connections from along Broadway. Less-good-but-doable bus connections possible for buses coming from cross streets and adjacent neighborhoods. This is probably the single most important spot to hit if you were to do an OL deviation. And of course you'd stop here if you did an OL branch, GL branch, or some modified UR through Everett.

Glendale - You weren't sure if this spot is good enough, but yes it totally is. From my crude measuring, the immediate area is even slightly denser than Everett Square. Numbers from @TheRatmeister bear that out, since the intersection of Broadway and Ferry actually has the second highest number of alightings (just barely). Buses along Broadway or Ferry would give basically straight shots to the station. Also, it's roughly 3/4 miles from Everett Square. Perfect.

Parkway - Depends. The walkability there is bad. Job and residential density is really low compared to most of the rest of the city (though that might be changing with new, planned developments). Presumably this station is to be an intermodal hub. It's a good place to catch buses from Broadway or the Parkway. TransitMatters also floated the idea of putting a Commuter Rail infill station there in their report on modernizing that line. It's only 1/2 mile from Everett Square, but you could justify that short distance because this could be a major intermodal hub. Plus the 1/2 mile walk to Everett Square has a big manmade obstacle in the way (f'n parkway). If you were building an Everett branch of the OL, or GL, or UR, then yes, you would put a stop here. However, this station isn't very compelling in your plan for an Orange Line deviation into Everett. In each of those plans, this replaces or duplicates the existing intermodal hub at Wellington. Moving the hub from Wellington to Parkway would be better for Everett riders, but worse for Medford riders, and then it's really splitting hairs which is better. It seems pretty close to a wash. It's certainly not so big an advantage that you'd do an expensive deviation.

Casino/Potential Stadium - Yes, but again, depends. The walkshed is currently bad, and the casino isn't actually that big a draw on its own. But there are some big plans for the area, including the potential stadium and maybe a whole new neighborhood on disused industrial space. I don't know all the details what is getting redeveloped, so I can't pick an exact spot for you. But there is some place on Alford that would make sense, roughly 3/4 miles from Parkway. I did said "depends," though. Again, this would be a good stop if you were building a new line, and so it's already along the way. But I don't think netting this stop is a compelling reason to do an Orange Line deviation. The people developing that area are doing it knowing full well there's no rapid transit there. That might hurt their businesses, but that's kind of just on them. They could have paid the premium to build closer to transit, but they didn't. I don't think we need to now bend over backwards to reroute a train closer to them.

Infill along Main Street - This could be a good addition. There's stuff going on on Main Street, but a little sparse on businesses and bigger apartment buildings. This stop would make sense if you were doing an OL deviation. However, if you were just looking at ideal spots for stations in Everett, this would rank kind of low. Anyone from Everett correct me, but Broadway just has more going on than Main, by a lot. Glendale (Broadway and Ferry) would be a more important stop than anything along Main Street. And I don't know how to serve them both with one project.

Infill along Broadway - So, somewhere between Everett Square and Glendale. For reasons I explained above, I think that spacing is wastefully tight. Unless there's some big trip generator I'm missing there, it's overkill in my opinion.

Infill along Ferry Street - I'll assume this isn't possible with any OL deviation. But for a new line, it makes sense. The new line would run up Broadway, then turn northwest onto Ferry. This station would be at maybe Cross Street. This looks like Main Street, in that it has some businesses and other ridership generators, but much less than Broadway. A line could serve this while also hitting Glendale, which is good. @TheRatmeister seems to be throwing out the idea of following the 109 up Broadway instead, and I think I agree.

Broadway and Lynn, or Broadway and Eastern - Glendale to Lynn Street is kind of short (<1/2 mile) and Glendale to Eastern is kind of far (>1 mile). @TheRatmeister lists these and suggests dropping Eastern, in part because then the route (if you want to follow the 109) would go more directly to Linden Square. That's pretty reasonable, but I think I'd do the opposite. Skip the intersection of Broadway and Lynn. This area is not very dense, especially with the big cemeteries adjacent, and that justifies wider stop spacing. Also, a Broadway and Eastern station could be a decent bus hub. Multiple bus routes run on nearby Salem Street, and a bus network redesign could funnel them to this station, instead of having to go the long way to Malden Center.

Linden Square - We're now well out of Everett, and the density is dropping off considerably. Everything past Glendale is really dropping off in priority. But this is still a good stop. The route from the last stop is a bit under a mile and could be done cheaply cutting under a historic rail line (current bike path). This stop has the most alightings on the 109, apparently, likely because of the other good bus connections.

Northgate Shopping Center - This is Crazy Transit Pitches, right? Past Linden Square, head under Route 1 for a terminal station at Northgate Shopping Center. That could be a great bus terminal for people coming in on Route 1. That's also like 100 acres of parking lot and low rise commercial that could be redeveloped for TOD.