CBTC - General Discussion

[Stephen Bauman]

A new GTFS timetable was released on May 9th. Here's a comparison of Main St to 34th St scheduled running times.

7 to Manhattan
5/18/16 "00:38:30.75";139.176673489183
5/17/17 "00:38:30.860656";137.501504103933
5/16/18 "00:38:33.84";135.902658289382
5/15/19 "00:38:34.5";138.059287247786

7X to Manhattan
5/18/16 "00:33:37.142857";74.907608597418
5/17/17 "00:33:37.142857";74.907608597418
5/16/18 "00:33:37.142857";74.907608597418
5/15/19 "00:33:40.465116";79.0705980022824

It's date, average running time in "hh:mm:ss" and standard deviation in seconds.

The GTFS schedules don't reflect any run time reduction as a result of CBTC's introduction (with or without ATO).

[RR503]

Do you have the old supplement schedules for the  on your computer? The  was on a long term supplement schedule before CBTC, against which runtime gains have been made. 

[Stephen Bauman]

26 minutes ago, RR503 said:

Do you have the old supplement schedules for the  on your computer? The  was on a long term supplement schedule before CBTC, against which runtime gains have been made. 

Here's the Manhattan bound data for 5/15/19 from the supplemented schedule that I downloaded yesterday morning (5/10/19).

7
5/15/19 "00:39:04.235294";161.586936722929
7X
5/15/19 "00:33:40.465116";79.0705980022824

As you can see, there's usually a fairly close relation between the supplemented and GTFS schedules.

There is an archive for supplemented schedules that dates back to September 2017. My experience has been that substantive differences between the supplemented and GTFS have been few and far between. 

[RR503]

10 minutes ago, Stephen Bauman said:

Here's the Manhattan bound data for 5/15/19 from the supplemented schedule that I downloaded yesterday morning (5/10/19).

 7
5/15/19 "00:39:04.235294";161.586936722929
7X
5/15/19 "00:33:40.465116";79.0705980022824

As you can see, there's usually a fairly close relation between the supplemented and GTFS schedules.

There is an archive for supplemented schedules that dates back to September 2017. My experience has been that substantive differences between the supplemented and GTFS have been few and far between. 

The post-4/26 supplements shouldn't show any runtime difference. If I'm not mistaken, before that date there was a supplement in effect that added a good bit of runtime to the . I'd be interested to know the extent to which that's changed, as with the  they pretty much left runtimes the same (which, IINM, causes some relatively significant operational issues with trains running ahead of schedule). 

[Stephen Bauman]

1 hour ago, RR503 said:

The post-4/26 supplements shouldn't show any runtime difference. If I'm not mistaken, before that date there was a supplement in effect that added a good bit of runtime to the . I'd be interested to know the extent to which that's changed, as with the  they pretty much left runtimes the same (which, IINM, causes some relatively significant operational issues with trains running ahead of schedule). 

Here are the run times from the supplemented from 4/10/19

7
4/10/19 "00:38:33.84";135.902658289382
7X
4/10/19 "00:33:37.142857";74.907608597418

 

[RR503]

12 minutes ago, Stephen Bauman said:

Here are the run times from the supplemented from 4/10/19

7
4/10/19 "00:38:33.84";135.902658289382
7X
4/10/19 "00:33:37.142857";74.907608597418

 

Very interesting. I’m surprised (but also somewhat not) to see no schedule difference. What does northbound look like? 

[Stephen Bauman]

3 hours ago, RR503 said:

Very interesting. I’m surprised (but also somewhat not) to see no schedule difference. What does northbound look like? 

7 from 34th to Main St
4/10/19 "00:39:06.419214";139.139961017714
5/15/19 "00:39:09.141631";130.222427854399

7X from 34th to Main St
4/10/19 "00:32:38.225806";66.3454494850382
5/15/19 "00:32:50.769231";81.2847558799406

 

[RR503]

Ugh. So they're padding the hell out of the schedules again. Lovely, just lovely. 

[Union Tpke]

9 hours ago, RR503 said:

Ugh. So they're padding the hell out of the schedules again. Lovely, just lovely. 

I had heard that runtimes would be reduced by 3 minutes, but I guess that was thrown out the window.

[Lance]

Flushing CBTC Temporarily Offline Following Multiple Station Overshoots

- Dan Rivoli - NY1

The MTA's latest technology, computers driving trains at the push of a button, got derailed on the 7 line, days after its full launch last week.

The reason: trains were overshooting stations — by four cars in one instance. Six trains — including three Friday, a day after the launch — overshot a station last week.

Pete Tomlin, the MTA's subway signals executive, called it an "anomaly," where trains didn't stop when they were supposed to and instead headed straight to the next stop. He notes there were at least 12,000 successful runs of the system.

"We've actually decided as a precautionary matter we would turn it off, get to the bottom of it," Tomlin told NY1. "If the operator has to emergency brake the train, or take control of the train, that's not a good thing. It's not unsafe."

The 7 line is the second, after the L line, to get a new signal system this year.

In April, the MTA started testing the 7's automated feature, slowly rolling it out to more and more trains. The computerized system lets trains run faster and closer together — up to 29 trains in a single hour at peak times. MTA leaders want to expand the technology to other lines within 10 years, after spending nearly a decade bringing it to the 7 line.

Officials say the results are clear: a major improvement in the number of trains running on time, reaching 89 percent this April during its slow rollout.

Passengers spent less time waiting on platforms and their trips were quicker, according to the MTA.

"Automatic train operation is a great add-on to every line that we have. It makes train travel predictable, people know where exactly every train is," said Andrew Albert, the chair of the NYC Transit Riders Council and an MTA board member. "We have to make sure that the glitches are worked out and obviously some of the glitches were not worked out."

MTA officials are working with the contractor to figure out exactly what went wrong, how to fix it, and get the system running again within days.

[Union Tpke]

@RR503 Are the dwell times on the 7 stringlines accurate? If so, there are very long dwell times.

Stringline_Southbound_7_Train_2Hours_5-17-2019_7-27-52-PM by Union Turnpike, on Flickr

[RR503]

9 minutes ago, Union Tpke said:

@RR503 Are the dwell times on the 7 stringlines accurate? If so, there are very long dwell times.

Stringline_Southbound_7_Train_2Hours_5-17-2019_7-27-52-PM by Union Turnpike, on Flickr

The MTA’s real time data feeds don’t measure dwell time so much as they measure time in station. Whenever the countdown clocks show “0 minutes” or “arriving now” or whatever, the viewer begins to register dwell, and whenever the last ‘stopped at’ message is sent in the feed, dwell is over. So take those figures to be a rough approximation of the time from when the train enters a station to when it begins to leave. 

[Around the Horn]

An update on Queens Blvd: they're running CBTC test trains now

[andreww]

Image failed to upload, I think.

[MHV9218]

Did they clean off the 30s pillar signs there like at W4? Looks a bit brighter than it used to be.

[Union Tpke]

@RR503 This had me laughing.

https://www.nydailynews.com/new-york/ny-signal-modernization-capital-plan-subway-service-20191007-j6w4cuthyvdqxdaelpga3rm3ta-story.html

Quote

Modern signals will allow the MTA to run up to 31 trains per hour on the Lexington Ave. line, which carries the 4, 5 and 6 trains and serves roughly 1.3 million daily riders. That would allow trains on the line to run less than two minutes apart.

The MTA can currently run up to 27 trains per hour on the line, according to [MTA spokesman] Minton.

Subway service capacity improvements will also include:
The Queens Blvd. line, which serves the E and F trains, will be able to handle 36 trains per hour, up from 30.
The G Line will be able to handle 18 trains per hour, up from 15.
The A, C and E lines in Manhattan will be able to handle 30 trains per hour, up from 24. The A and C lines in Brooklyn will be able to run 31 trains per hour, up from 26.
The Astoria Line, which serves the N and W trains in Queens, will be able to run 18 trains per hour, up from 15.

Minton pointed out that the subway may not actually hit the increased service limits.
The Lexington Ave. and Queens Blvd. lines are two of the only stretches of track where the MTA even comes close to pushing the signal system to its limit. Minton said that low demand on some lines leads MTA officials to run far fewer trains than the system is capable of handling.

MTA officials run just nine trains per hour on the G line, despite its current signals being able to handle up to 15 per hour, Minton said.
Fried said that fewer subway delays will likely lead to a boost in ridership that would justify more frequent service.
“Broadly speaking, with a reliable subway system, we would expect ridership to resume the upward trajectory it was on until 2014,” said Fried. “Peak capacity on these lines won’t be the only factor affecting ridership. Improved reliability — fewer signal problems — will reduce the worst, most time-consuming delays.”

What a joke. Good luck trying to get 31 on Lex. 36 on Queens Boulevard is hysterical! I guess they are going to run more trains than can fit on the line to get 30 TPH. This will be awful for crews. This contrasts with another document, which stated that CBTC would allow for a 10% increase in capacity to 33. As it is, only 25 E and Fs actually run during rush hours, so that would get them to 30. This will be a scheduling nightmare. Can't wait for gaps in the schedule!

Here is what happened when they tried to schedule 35 TPH in the 1960s.

http://www.laguardiawagnerarchive.lagcc.cuny.edu/FileBrowser.aspx?LinkToFile=FILES_DOC/Koch_FILES/08.005.0000.0000261.02.PDF#undefined

Screen Shot 2019-10-07 at 8.03.45 AM by Union Turnpike, on Flickr

Screen Shot 2019-10-07 at 8.03.50 AM by Union Turnpike, on Flickr

Screen Shot 2019-10-07 at 8.03.55 AM by Union Turnpike, on Flickr

It seems like that the MTA chose random numbers for the new capacity figures.

 

Edited October 7, 2019 by Union Tpke

[RR503]

1 hour ago, Union Tpke said:

It seems like that the MTA chose random numbers for the new capacity figures.

Yeah, as far as I can tell they just multiplied existing throughputs by 1.2 and decided that'd be their new target capacity. Mark me down as extremely skeptical of those figures. CBTC does not do well with merges and long dwell times, and, uh....

[LGA Link N Train]

6 hours ago, RR503 said:

Yeah, as far as I can tell they just multiplied existing throughputs by 1.2 and decided that'd be their new target capacity. Mark me down as extremely skeptical of those figures. CBTC does not do well with merges and long dwell times, and, uh....

I was about to question the same things.

Its good that they want to increase service on all lines receiving CBTC. However, with the current circumstances that we have to deal with the Subway now, would not allow for it. 

7 hours ago, Union Tpke said:

Minton pointed out that the subway may not actually hit the increased service limits.
The Lexington Ave. and Queens Blvd. lines are two of the only stretches of track where the MTA even comes close to pushing the signal system to its limit. Minton said that low demand on some lines leads MTA officials to run far fewer trains than the system is capable of handling.

This quote speaks for itself. 

[LexAveExp5]

I don’t understand why the G is such a high priority for CBTC

[RR503]

10 minutes ago, LexAveExp5 said:

I don’t understand why the G is such a high priority for CBTC

The signal system on xtown is original. The interlockings are still fully manual US&S machines with L/R signal numbering.

http://www.trainsarefun.com/lirr/lirrextralist/MTA Transit Diagrams/Schematic Track & Interlocking Diagrams (2019-01).pdf#page=235

Edited October 16, 2019 by RR503

[P3F]

On 10/16/2019 at 3:25 PM, RR503 said:

The signal system on xtown is original. The interlockings are still fully manual US&S machines with L/R signal numbering.

http://www.trainsarefun.com/lirr/lirrextralist/MTA Transit Diagrams/Schematic Track & Interlocking Diagrams (2019-01).pdf#page=235

At the same time, I'm not sure if it should really be a focus. According to the quote above, Crosstown would be able to run 18 tph instead of 15. But it barely runs 10 tph of 300 foot  trains at its peak right now, and it's unlikely that ridership will increase to the point that 15 tph would become necessary.

So unless the current signals are very unreliable, CBTC won't really make a difference on Crosstown other than being able to better handle occasional reroutes.

[RR503]

1 hour ago, P3F said:

At the same time, I'm not sure if it should really be a focus. According to the quote above, Crosstown would be able to run 18 tph instead of 15. But it barely runs 10 tph of 300 foot  trains at its peak right now, and it's unlikely that ridership will increase to the point that 15 tph would become necessary.

So unless the current signals are very unreliable, CBTC won't really make a difference on Crosstown other than being able to better handle occasional reroutes.

Totally agree with you that optimally the funds would be spent elsewhere, but capacity isn’t the only metric of signal system functionality. If we don’t do Crosstown, it’ll be the only stretch of track left with original IND signals — the next oldest stuff to it is a full 25 years younger, and relies on different interlocking technology. Sure it isn’t a total dealbreaker, but these ancient signals are pretty expensive, and they only become more so when you’re spreading fixed expenses (like keeping the tooling for DIY vacuum tube relays) over a smaller set of active signals. I have many problems with how subway signal issues were reported, but this is something folks really got right: maintaining 1930s era signaling is quite pricy. This is all to say the agency can expect a good RoI through doing replacement here — and that’s on top of the ops RoI inherent in CBTC installs.

And anyway, who doesn’t want a  that can consistently beat an  from Bergen to Queens? 

[34bluebird]

 could always beat the now.

[bulk88]

I took a look at the Piper UWB equipment on 7 line. Few or none of the nodes have a backhaul. Just 120v. They are radio data packet repeaters, not Access Points with their own fiber optic backhaul. I also suspect the next train UWB itself is a rolling radio repeater mesh node. So even though it is NOT AV, if there is a poor to low signal, a train will talk to another train that talks to a repeater node that talks to a backhauled node to a zone controller for CBTC packets. On QBL there are radio nodes ever 600-1000 feet. Almost as bad automatic block track circuits in sheer equipment count, And each legacy radio node must backhaul. Piper equipment just requires 120v, and almost no backhaul cables since it is a mesh network, not a hub spoke wifi system.

[Kamen Rider]

 

This started as an event at Penn Station for OMNY's expansion. But, at the roughly 30 minute mark, after they did the ceremonial tap-in, Andy Byford came back to the mic and he was LIVID. The Flushing CBTC system was malfunctioning in the slushy snow for the second time this month and he made it publicly known he was NOT a happy camper with Thales.