Why Your Subway Train Might Start Moving Faster

[AlgorithmOfTruth]

Can someone verify that the timers coming into Broadway Junction on the Manhattan-bound tracks of the / lines as being necessary?

[RR503]

It's hard to do that without having access to single line diagrams. A lot of the GTs on (especially older) parts of the signal system are there for stopping distance reasons

[Union Tpke]

5 minutes ago, RR503 said:

It's hard to do that without having access to single line diagrams. A lot of the GTs on (especially older) parts of the signal system are there for stopping distance reasons

Adding more insulated rail joints would really help!

[RR503]

10 minutes ago, Union Tpke said:

Adding more insulated rail joints would really help!

Yeah, or just more signals — more signals = more granular control of train position, speed, etc. 

I know you understand this, UT, but for the benefit of anyone who doesn’t: more IJs doesn’t necessarily mean more capacity; the reason that cutting in more here helps is that if you can cut in a new IJ at whatever the today-standard braking distance is from a given signal, there’s no need for a capacity-reducing GT or control line mod. Problem is, new IJs are expensive... 

Edited May 5, 2019 by RR503

[Union Tpke]

14 minutes ago, RR503 said:

Yeah, or just more signals — more signals = more granular control of train position, speed, etc. 

I know you understand this, UT, but for the benefit of anyone who doesn’t: more IJs doesn’t necessarily mean more capacity; the reason that cutting in more here helps is that if you can cut in a new IJ at whatever the today-standard braking distance is from a given signal, there’s no need for a capacity-reducing GT or control line mod. Problem is, new IJs are expensive... 

That is why the NYCT should have bitten the bullet and paid for it in the '90s.

Which do you think would make more sense?

Also, do you think that there any improvements should be made along with CBTC?

[RR503]

2 minutes ago, Union Tpke said:

That is why the NYCT should have bitten the bullet and paid for it in the '90s.

Which do you think would make more sense?

Also, do you think that there any improvements should be made along with CBTC?

Yeah, the mods contract was chronically underfunded—which, beyond making IJ mods hard, means that we got a lot more 1 shots than we probably should have (1s GTs are cheaper than 2s). 

I think your use scenario for more signals vs moved IJs is different. Leading up to interlockings, in areas with difficult geometry or at high dwell stations, you want more signals, while fixing some control line issue in a more average part of the system you’d want a new IJ. 

I think what’ll be key to good CBTC is a) maximizing speeds and minimizing braking distances wherever possible and b) focusing on installing CBTC, rather than CBTC with a bunch of other junk on top of it. So do max speed profiles, review emergency brake rates and then do work train compliance, etc to eliminate AWS. Because speaking of expensive IJs...

[R68OnBroadway]

26 minutes ago, RR503 said:

Yeah, the mods contract was chronically underfunded—which, beyond making IJ mods hard, means that we got a lot more 1 shots than we probably should have (1s GTs are cheaper than 2s). 

I think your use scenario for more signals vs moved IJs is different. Leading up to interlockings, in areas with difficult geometry or at high dwell stations, you want more signals, while fixing some control line issue in a more average part of the system you’d want a new IJ. 

I think what’ll be key to good CBTC is a) maximizing speeds and minimizing braking distances wherever possible and b) focusing on installing CBTC, rather than CBTC with a bunch of other junk on top of it. So do max speed profiles, review emergency brake rates and then do work train compliance, etc to eliminate AWS. Because speaking of expensive IJs...

How much would it cost though to retrofit all the old equipment (work trains, museum trains, etc.) ? If you had to do all of that, wouldn't it be better to just install AWS or is it that expensive?

[RR503]

Just now, R68OnBroadway said:

How much would it cost though to retrofit all the old equipment (work trains, museum trains, etc.) ? If you had to do all of that, wouldn't it be better to just install AWS or is it that expensive?

RPA pegs the cost of full work train compliance at 95 million (if axle counters are not used) or 50-35 million (if they are). Assuming the FTA's estimation of AWS adding a 30% premium onto CBTC installation costs is correct, then AWS added 145 million to the originally budgeted cost of Flushing CBTC. Compliance has been done before, and is actually recommended as -- beyond the upfront savings -- killing AWS means you have many fewer signal components to maintain, and have many fewer points of failure. 

[AlgorithmOfTruth]

57 minutes ago, RR503 said:

It's hard to do that without having access to single line diagrams. A lot of the GTs on (especially older) parts of the signal system are there for stopping distance reasons

Do those single line diagrams take into account the route's curvature and topography? The relationship between timers and stopping distance/speed restricted-areas is clear, however, the timers in question appear to be counterproductive in their placement. About 500 FT before you enter Broadway Junction, the slope of the track bed sharpens, naturally reducing speed.

[RR503]

Just now, AlgorithmOfTruth said:

Do those single line diagrams take into account the route's curvature and topography? The relationship between timers and stopping distance/speed restricted-areas is clear, however, the timers in question appear to be counterproductive in their placement. About 500 FT before you enter Broadway Junction, the slope of the track bed sharpens, naturally reducing speed.

Yes, for obvious reasons stopping distance calculations take into account geometric properties. Single line diagrams, btw, don't show stopping distances, but instead the layout of the area signal system. A write up (with examples) can be found here:

https://www.nycsubway.org/wiki/Subway_Signals:_Single-Line_Signal_Diagrams

[Union Tpke]

There seems to be more to come:

Screen Shot 2019-05-17 at 2.27.21 PM by Union Turnpike, on Flickr

[R68OnBroadway]

Glad to see the push for more TPH on the  , I always found it ridiculous that they didn't look to have 30+ tph given CBTC and high ridership. Hope they add tail tracks at 8th and maybe some turnback ones at Atlantic to faciliate this...

[Union Tpke]

44 minutes ago, R68OnBroadway said:

Glad to see the push for more TPH on the  , I always found it ridiculous that they didn't look to have 30+ tph given CBTC and high ridership. Hope they add tail tracks at 8th and maybe some turnback ones at Atlantic to faciliate this...

We need more trains! The  should not be crowded! 28 TPH are possible without tail tracks!

[Union Tpke]

Screen Shot 2019-05-17 at 7.47.40 PM by Union Turnpike, on Flickr

Screen Shot 2019-05-17 at 7.47.45 PM by Union Turnpike, on Flickr

[Union Tpke]

Screen Shot 2019-05-17 at 7.47.45 PM by Union Turnpike, on Flickr

@RR503 How would infrastructure improvements affect the  's OTP?

[RR503]

53 minutes ago, Union Tpke said:

Screen Shot 2019-05-17 at 7.47.45 PM by Union Turnpike, on Flickr

@RR503 How would infrastructure improvements affect the  's OTP?

I’m not aware of any infrastructure improvements — just the previously reported sign changes...

[Lex]

24 minutes ago, RR503 said:

I’m not aware of any infrastructure improvements — just the previously reported sign changes...

Did they install CWR?

[RR503]

22 minutes ago, Lex said:

Did they install CWR?

Yeah, but that shouldn’t have any impact on speed — aside from it making a cooler noise, of course. 

[Union Tpke]

@RR503 This shows runtimes between Lexington-63rd and Kew Gardens. 1 is between Lex and Roosevelt Island, 2 is between Roosevelt Island and 21st, 3 is between 21st and Roosevelt Av, 4 is between Roosevelt and Forest Hills, 5 is between FH and 75, and 6 is between 75 and Union Tpke.

Look at the extreme variability.

The variability at 3 has to do with the merge at 36th Street.

RuntimeNorthbound_Train_ by Union Turnpike, on Flickr

RuntimeSouthbound_Train_ by Union Turnpike, on Flickr

 

[Union Tpke]

Timers on CPW Express between 125th and 59th are clearly a factor here. Six minute variability!

RuntimeNorthbound_Train_-1 by Union Turnpike, on Flickr

RuntimeSouthbound_Train_-1 by Union Turnpike, on Flickr

[RR503]

Yeah, the variability charts are really something. When my friend is done fixing the X axis, I'm going to post some of my favorites -- these are really good at highlighting problem areas.

A methodological note: the big range is the 5th to 95th percentile runtimes (not 0-100 to help cull some data failures), and the little box is 25-75. Middle line is 50. The site that makes these will eventually be public.

[Union Tpke]

1 minute ago, RR503 said:

Yeah, the variability charts are really something. When my friend is done fixing the X axis, I'm going to post some of my favorites -- these are really good at highlighting problem areas.

A methodological note: the big range is the 5th to 95th percentile runtimes (not 0-100 to help cull some data failures), and the little box is 25-75. Middle line is 50. The site that makes these will eventually be public.

Box and whisker plots put to good use!

[Union Tpke]

@RR503 The program measures runtimes between departures and arrivals, so you cannot easily show delays from merges, as these would be in the form of increased dwell times. The awful switch layout at Jamaica Center really delays  service. To show this I showed the running time from one station further west, Jamaica-Van Wyck. The comparison between Southbound and Northbound  should show the impact of the lousy switching layout:

 

 

       

 

[RR503]

Yup! My entry to chartfest is this: look how the GTs from Franklin to Atlantic, the put-ins at Bowling Green, and crappy signalling and long dwells at 14th kill Lex service:

[Union Tpke]

Last one for tonight. The merge at 50th slows  service. Look at the difference here: