General MBTA Topics (Multi Modal, Budget, MassDOT)

I can't find it now, but I saw a blog post once from the makers of the Transit app about how they went to some effort to do this right.
Google did have accurate lines at one point, several years ago. Not sure what algorithm changed to cause the screw-up.

Incredibly, at the time writing of the Transit App blog that ceo mentioned, it looks like both locations were rendered correctly on Google:
1_w6ffiQqAGbSAS4DvLFNHrQ.png

vs now:
google_boston_now.PNG
 
Maybe it's the transit equivalent of a trap street? Add a deliberate error to your map so if your rivals coppy it you notice?
 
Incredibly, Google Earth seems to think that the Fall River leg of SCR has been completed for some time now, albeit via Somerset.
 
They finally finished work on the C line last weekend and I can happily report they've somehow eliminated all the wheel screeching along the St. Mary's St portal.
It's so nice to just hear the woosh of the train car without the nail on a chalk board effect...

Platforms look minimally changed, but I assume the new curbs help accessibility.
 
As mentioned in another thread, the MBTA has released FY19 (July 2018-June 2019) ridership data for the gated subway stations. (Spreadsheet is here, feel free to download and play with the data.) Assembly is not in the FY13 data (not yet open); Wollaston was closed for reconstruction in the FY19 data. The only other substantial changes in service were the 2014 promotion of Bowdoin to full-time, and the 2018 addition of SL3 service.

Compared to FY13 (the 2014 Bluebook data), overall ridership at the 64 gated stations was down 7.4%. Change by line (change including transfer stations in parentheses):
  • Blue: +7.6% (-0.4%)
  • Green: -16.8% (-15.1%)
  • Orange: -7.8% (-5.8%)
  • Red: -8.9% (-7.6%)
  • Silver: +43.5% (+3.0%)
  • Subway transfer stations: -6.7%
  • CR transfer stations: -9.4%
  • Major bus transfer stations: -8.4% (excluding Blue Line stations, -10.3%)

Big losers
The biggest drops in ridership were Harvard (-4,671), Government Center (-3,151), Park Street (-3,082), Forest Hills (-2,622), and Back Bay (-2,454); these represented drops between -13.6% and -29.1%, and together were almost half of the decline in ridership. It's notable that three of the five (and the next three - Hynes, Kenmore, and Sullivan) are major modal transfer stations. Considering the greater-than-average drops for modal transfer stations, I would speculate that unreliable service means that transferring between subway and bus or CR has become undesirable.

Biggest percentage drops were Suffolk Downs (-53.7%), Government Center (-29.1%), Boylston (-22.9%), Hynes (-21.3%), and Harvard (-20.1%). Out of the top 15, 8 were Green Line stations. Not surprising given that one-sixth drop in ridership.

Biggest rank drops were Boylston (-9), Government Center (-7), and Hynes (-6).

Big winners
The biggest gains in ridership were Kendall/MIT (+1,585), Orient Heights (+1,467), North Quincy (+1,453), Courthouse (+1,247), and Maverick (+1,087). Maverick probably benefits from the new developments nearby, and Kendall/MIT and Courthouse are fast-growing areas as well. North Quincy was likely Wollaston ridership. The ridership growth on the SL was fueled entirely by Courthouse; WTC didn't change at all. Assembly ridership was 3,977 - not bad considering that before opening 2030 projections were 5,000 daily.

Biggest percentage gains were Courthouse (+97.2%), Orient Heights (+51.8%), Bowdoin (+39.4%), North Quincy (+20.8%), and Broadway (+14.4%). Six of the top ten were Blue Line stations. For Bowdoin, this was likely a return to its pre-reconstruction ridership, which was fairly steady in the ~4,000 range prior to that FY13 count.

Biggest rank gains were North Quincy (+10), Broadway (+8), Wonderland (+8), Courthouse (+7), and Orient Heights (+7).

It's worth noting that although transfer stations lost a lot, that's driven by the Green Line drop. DTX, State, and South Station had a collective drop of -0.2%, much better than the system average.
 
FY20 will be interesting: started off with Red Line derailment and subsequent signal fiasco, then the repetitive weekend and night shutdowns through the fall and winter, followed by the big shift to WFH in March. FY21, last I was aware, has been budgeted out to proceed as all things are normal (apart from hiring, which was slowed down/spread out to FY22), with additional cleaning @ $1 million/week, but I'm really worried about FY22+...
 
The very end of this Globe article says that nine intersections on Beacon Street "were also upgraded . . . with technology that helps trolleys to squeeze through before the signal turns red". Conventional wisdom around here is that the signals have been TSP-capable for quite some time now, but that the necessary cooperation between Brookline (seemingly willing) and the MBTA (T...S...P?) was absent. Does anyone know whether this means that TSP will actually be activated on the C line, or is this just another step closer, somewhere on a long and interminable list of steps?
 

This was new to me, Codman Yard expansion project page with sparse details

That's been planned all along. One of the old FCMB slides diagrammed it preliminarily early on in the RLT effort, but I guess this is the real nailed-down design.

Codman's infield inside of the loop track is only two-thirds occupied (one track just being a throwaway dead line for the 4 deceased 01400-holdover work cars and the parted-out 01600's pair wrecked in last summer's Columbia Jct. derailment). Most of the rest is overgrown with brush, not even used for materials piles or anything like that. It has actually never been fully utilized ever since the '28 HRT conversion, with Historic Aerials showing progressively fewer and fewer storage tracks going back in time with the same loop tracing the outer edges. It only got as full as it is now because of the fleet expansion for 6-car trains in '88. It's not known why BERy so enormously overbuilt it by an order of magnitude for expansion that was flat-out not forseeable for a majority of the RL's existence. But their paranoia-level provisioning is just what the doctor ordered right now.

While train storage and neighborhood amenities like less distracting floodlights are the primary issues, would be interested to see after the 8/20 neighborhood presentation if they're going to manicure the perimeter at all for better materials storage. It's hella haphazardly-arranged right now with the dirt parking lot, dangerously makeshift staff truck grade crossings, unmanicured dirt perimeter road, and random disorganized piles-o'-crap on the southern perimeter. Wellington rebuild put lots of P's and Q's mindings into reorganizing the materials piles for neatness, quick access, and less-cumbersome grade crossing movements in the yard. Codman could use that pretty bad, too, because all the MOW functions are laid out like anti-efficient ad hoc slop...and pretty dangerously at that.
 
Last edited:
The very end of this Globe article says that nine intersections on Beacon Street "were also upgraded . . . with technology that helps trolleys to squeeze through before the signal turns red". Conventional wisdom around here is that the signals have been TSP-capable for quite some time now, but that the necessary cooperation between Brookline (seemingly willing) and the MBTA (T...S...P?) was absent. Does anyone know whether this means that TSP will actually be activated on the C line, or is this just another step closer, somewhere on a long and interminable list of steps?
It's already active on the C line. Equipment was installed in June.


And


And here they confirm the software was already up and running pre C branch closure

 
Been meaning to poke around, but has the T published data on the effect of TSP on the C line yet? Seems like it could be interesting.
 
Did the "Green Line Type Nine doors take way too long to close" thing turn out to be a nothingburger?

That question was asked on RR.net this week, and the general consensus from random-sample replies is that it's a non-issue.

Anecdotal from my only two rides to-date on a 9...it's def no worse than the shitty Breda doors with their over-sensitive cycling sensors.
 
Been meaning to poke around, but has the T published data on the effect of TSP on the C line yet? Seems like it could be interesting.

Too early. They don't have nearly a broad or varied enough dataset yet.

Even though this year is going to be only trace the traffic of a normal year, I would think that Allston Xmas/student locusts descending is the first stiffish test since during that last weekend of Aug. Beacon is full of wayward U-Hauls that don't know where the hell they're going.
 
TransitMatters quietly rolled out their Data Dashboard site this week: https://dashboard.transitmatters.org/

Internal release statement: https://transitmatters.org/blog/2020/8/3/rolling-out-our-data-dashboard

News story: https://mass.streetsblog.org/2020/0...eleases-data-dashboard-to-visualize-t-delays/


The dashboard graphs data dumps from the T's Performance API for travel times and delays on all 4 rapid transit lines, measured against agency OTP targets. Think your commute today was extra-hosed? Now you can plug it in and see if the actual travel clock jibes with your perception. Site has been in closed beta most of the summer pretty much only known to folks who discovered their "Labs" blog on the website; graduated to public beta start of month. Still some data gappage to fill (they note that Park St. & Kenmore are unavailable at present because they need add'l coding to translate the particular way the T API parses those multi-line stops), but it's now ready enough for prime time for the public unveil and is now listed on the TM homepage pull-down menu next to the New Train Tracker.
 
Last edited:
Oh boy, that is fun to mess around in.

A few stray observations already:

Blue Line -- unsurprisingly -- does pretty well staying on-target... during rush hours, where you can see the headway performance markers hugging the target line almost precisely. But during the mid-day off-peak -- and surprisingly frequently day-to-day -- things shift, with a lot more variability in headways, and in general more cases of missing the target. The evening does not seem to be beset by the same phenomenon, or at least not to the same extent.

July 27 gives a good example. In this case, the mid-day variability isn't particularly problematic, but there are other days where the mid-day seems to unravel. By contrast, those peak headways are remarkably consistent, more so than any other HRT service I've seen so far.

(Also of note: as you'll see below, the Red and Orange generally do pretty well in terms of targeted travel times [at or faster than target], but poorly in terms of on-target headways; the Blue Line has the reverse, with travel times that are often barely on target, and frequently well above, though not egregiously.)

Screen Shot 2020-08-13 at 1.43.26 PM.png


Orange Line headways in the core struggle hard. For example, check out Back Bay inbound to State Street on July 9. This is a case where the visualization is very valuable -- my guess (from reading the chart) is that the average headway here probably hits the benchmark of 6 min peak/8 min off-peak... but does so because for every train that is late by two minutes, there's another that is early by two minutes.

Screen Shot 2020-08-13 at 1.27.35 PM.png


In this specific example, there was a service advisory at 9:38am about a mechanical problem with a train approaching Sullivan Square. It's possible that that delay kicked everything off the rails for the rest of the day. But...

If we look at July 10, we see that the AM peak has similar unevenness, and though there was a reported incident in the early afternoon, nothing in the morning that we can tie things back to. (Nothing public, at least.)

Screen Shot 2020-08-13 at 1.33.25 PM.png


(Note that in my random clicking, I couldn't find a single day without some sort of service alert on the Orange Line.)

I haven't looked as much at the Red Line, but I was surprised to see some similar effects there. Again looking at July 10, we see a lot of dancing around the headway target on inbound trains, although I'd say the impact is reduced since the headways themselves are shorter -- very few accidental 10-minute headways, unlike the Orange Line above:

Screen Shot 2020-08-13 at 1.38.52 PM.png


But the outbound service isn't so fortunate, with a lot of dancing, and a fair few "surprise" 10+ minute waits, during peak, off-peak, and reverse peak alike:

Screen Shot 2020-08-13 at 1.41.01 PM.png
 
Oh boy, that is fun to mess around in.

A few stray observations already:

The Orange plots are a screaming obvious tell for the car shorage. Same 120-car fleet that was running more frequent 4-car trains over the El in 1987 stretched to its absolute limit by longer station dwells today, so you get funhouse-mirror games like 'technically correct' adherence with a buried lede that everything is either running late or early on its assigned schedule. A graph like this would suggest that OL is likely also subject to more "stealth cancellations" than the other lines because that's the variable they're playing with to game a 'technically correct' headway adherence. That's how it works when fleet's so threadbare you have no other options. At least the +32 car / 7-set fleet expansion and signal reboot slays this game once and for all.


Red's never had a car shortage, but the increasing downtown station dwells have definitely had punishing cumulative effect over the last 25 years on adherence and creates its own set of dispatch sleight-of-hand bunching + gapping games in the name of 'technically correct' headway adherence. Part of the blame here also goes to the 1988-install ATO signal system, which made some incorrect assumptions that the move to 6-car trains would flatten downtown growth such that the same uniform signal blocks would work end-to-end. That quickly turned out not to be true, as Kendall-SS platform dwells became the exploit where schedules went to die. The old Blue Line-style wayside lights + trip-arm stops system segmented downtown more fine-toothed than the currrent system and allowed closer-packed trains as result...with the all-or-nothing enforcement of the mechanical trips allowing shrewd operators the leeway to gun it between blocks to make up lost time if they found themselves bleeding any schedule from platform dwells (you can still watch Blue operators pull this trick where they go leadfoot between signal lights...but throttle back in the nick of time before getting a stop penalty). ATO's constant speed enforcement removed the cheat (for wise reasons, too) but went too far at trying to fit square-peg Downtown dwells into round holes. Thus the only troubleshoot trickery is more dispatch-side gapping/bunching games in the name of overall headway adherence...and probably also more outright canceled runs than normal when they find themselves too far on a limb with those games and have to reset. Similar fix is coming with the RLT signal reboot which direct-targets Downtown's problematic block spacing for more nimble packing that treats downtown as a distinctly more complex dispatching area than Cambridge.

The problem with Red vs. Orange is that the immediate improvement won't be as long-lasting with Red if we don't do something to fix the platform crowding growth which is simply outpacing the availability of ops tricks. Orange fleet expansion + signal revamp is a legit perma-fix that won't be overtopped for decades because it finally gets the line running up to its incumbent infrastructure's capabilities. We've been punished for 33 years by the decision to not buy an expansion order of OL cars for '87, so that's pretty open-shut in what the fleet fix + signal refresh does for implanting headways a few steps ahead of the oversaturation threshold. With Red we're unfortunately buying maybe 10 years of pretty substantial relief but only enough kick-can time for getting Red-Blue unstuck , addressing second Park St. Red egress and crowd-spreading, and addressing DTX platform/egress flow as primary irritants. The problem comes immediately roaring back as bad as ever if you can't wrestle down how long those doors are staying open at Park/DTX with platform-side load-spread options. We'll be at the Harvard-curve throughput limit of the line once it's doing 3 min. headways, so there's no additional "optimization" tricks left and it's either build your radial relief or bust. "Or bust" a spot-on guarantee solely from the SS transfer swells if we introduce southside RUR.


No idea why Blue performs so counterintuitively with which target it's hugging. Definitely not a capacity-side problem, as they currently have more cars than they need and the '04-07 six-car extension has stayed well, well ahead of Downtown dwell growth. The incumbent ops post- fleet expansion are basically what was required for absorbing new Red-Blue ridership, so there's some slack. It's thus curious that they seem to be lollygagging their adherence like this. Maybe playing it tight is just not enough of a priority for central dispatch right now??? Anyway...since the BLT study is going on right this second for infrastructure modernization and climate change resilience, these findings would be a good thing to ask at the first slate of community meetings. The old permissive signal system is almost certainly going to be a first-priority target for replacement solely as part of the climate fortifications, because the quantity of wayside lights + ancient mechanical trips + trip-arm heaters are extremely hardware- and electricity-intensive compared to the solid-state ATO on Red/Orange. They're going to rip it all out and replace solely to lower their water threat exposure threshold. Could even be that they skip right over Red/Orange ATO and lay waterproof fiber for the T's first CBTC signal system...since that radio-heavier system requires even less total field hardware (and less water-vulnerable hardware for what it does require in the field) than ATO and Blue is far and away the easiest of the lines to use as a tech testbed for that. (Note: while Red/Orange aren't going full-on CBTC, all that fresh new fiber backplane the existing system is getting for RLT/OLT makes them future-upgradeable to the Jetsons Shit signals at minimal pain and suffering...so Blue is still the consensus first-dip line for perfecting a CBTC implementation before it gets deployed wider).

Therefore, it's going to be eminently fair ask in the public meetings what the schedule optimization angles are going to be for BLT. Those haven't been defined at the macro- bullet level in any FCMB presentation yet because any Blue schedule adherence issues pale so dramatically in comparison to what's killing Red/Orange and getting five-alarm fixes. Asking them to explain a persistent schedule adherence chart pattern like that and how their pu-pu platter of infrastructure refreshes make it more precision-controllable would be a good place to start. I mean...if you live on the actual North Shore the climate-change fortification bread-and-butter of BLT is of enough interest because you're inundated with those concerns as a function of living there. But the at-large commuter probably needs a service reliability hook to get as interested in BLT outreach as they are with built-in interest in de-crapifying Red/Orange. Centering a lot of the discussion on schedule adherence oddities and fixes therein probably helps for giving the BLT process wider rider participation.
 
Last edited:
Regarding the Blue Line... it may just be meaningless noise in the data. To momentarily put my armchair data scientist cap on:

Even during the off-peak segments, the variance is pretty small: we're talking about 8 minutes or 10 minutes instead of the on-target 9 minutes. My gut says that the average rider isn't going to notice that difference. (Compare to the Orange Line, where 6 minutes regularly turns into 10 minutes.) The off-peak headways are pretty much 9 minutes ± 11%.

I think the variance on the on-peak headways is actually pretty much the same. It looks smaller because, numerically, it is. But the target is 5 minutes, and 5 minutes minus 11% is 4.45 minutes, and if you look at the headways, you do indeed see them ranging from ~4.5 minutes to ~5.5 minutes.

@TransitMatters -- it would be awesome to have a "Download as CSV" feature to be able to dump the data in the on-screen visualization into a download that could be pulled into Excel or the like. I know this data is accessible elsewhere, so it's hardly an urgent request, but could be cool!

That all being said: looking at that off-peak section, I wonder if the T's internal timetable actually alternates between 9 minute and 10 minute headways for the off-peak Blue Line. Because that pattern is extremely regular. Moreover, the variance against alternating 9 minute/10 minute headways off-peak is very consistent with the variance for the peak 5 minute headway, which would make sense if both were driven by the inevitable variability in turning trains.

So, if that's the case, then there really isn't anything different going on off-peak. The Blue Line typically stays on schedule, give or take 30 seconds, throughout the course of the day; the schedule itself varies, but the performance, not so much.
 

Back
Top