Why Your Subway Train Might Start Moving Faster

[RestrictOnTheHanger]

Those long dwells at Roosevelt  are usually due to people holding doors and/or expresses waiting for arriving locals, which causes more door holding. 

I did the from 14th to 71 ave I think last Weds and it was mostly smooth sailing. I dont go Roosevelt to 71 on the express more than once or twice a month though. 

Edited April 1, 2019 by RestrictOnTheHanger

[Union Tpke]

My  today definitely did not go at the new speeds.

[VIP]

7 hours ago, RestrictOnTheHanger said:

Those long dwells at Roosevelt  are usually due to people holding doors and/or expresses waiting for arriving locals, which causes more door holding. 

I did the from 14th to 71 ave I think last Weds and it was mostly smooth sailing. I dont go Roosevelt to 71 on the express more than once or twice a month though. 

I thought Conductors were instructed NOT to keep doors open to make Express/Local connections as per Save Safe Seconds... 

[Bay Ridge Express]

1 hour ago, VIP said:

I thought Conductors were instructed NOT to keep doors open to make Express/Local connections as per Save Safe Seconds... 

Yet they do that every morning on the  at 36 St... I'm not complaining though if the next express train would be 10 minutes away during the early rush

[RestrictOnTheHanger]

Some changes along the Queens bound

The curve from 47th-50th to 5th/53rd looks like it's had its timers fixed. 

The 2 shot GTs appear to clear at the posted speed (20 or 25). No trains were ahead of us at 5th so I couldnt see if the STs were fixed too. 

The 53rd St tube  had one of the 2 shots disabled or fixed, I only saw one active. 

Finally a new 30 miles sign replaced the old 23 miles sign on the QBL local leaving Steinway. 

Based on riding the both yesterday and right now. 

 

Edited April 9, 2019 by RestrictOnTheHanger

[N6 Limited]

What's a two shot timer again?

[RR503]

2 minutes ago, N6 Limited said:

What's a two shot timer again?

2 shot timers basically give ops 2 chances to get their train's speed right. They have yellow over S aspects, and if you run through the block before the timer at or below its speed, it will clear to green and the next signal (which was red) will go to yellow/S. If you pass this first signal over speed, no matter, the next signal can be cleared as a one shot -- it's only if you pass that second signal over speed also that you get tripped. They're generally used in areas where restrictions go on for a longer distance, or in areas where restrictions do not have to be so granular as they cause less operator variability as, during normal operation, an op will never be facing a red.

[Union Tpke]

@RR503 It took me about 23 minutes to get from Grand Central to Kew Gardens. Service on Queens Boulevard has sped up!

[RR503]

34 minutes ago, Union Tpke said:

@RR503 It took me about 23 minutes to get from Grand Central to Kew Gardens. Service on Queens Boulevard has sped up!

I'm just waiting for rush hours to stop getting killed -- take a look at how runtimes progress here:

[Union Tpke]

2 minutes ago, RR503 said:

I'm just waiting for rush hours to stop getting killed -- take a look at how runtimes progress here:

 

There are few caveats to my great trip–the 6 came after a minute, the E was just in the station at Lex, and it is Good Friday today. However, I have noticed the speed up over the past week. E and Fs move through 75th Interlocking without any issues during the off-peak.

[RR503]

3 minutes ago, Union Tpke said:

There are few caveats to my great trip–the 6 came after a minute, the E was just in the station at Lex, and it is Good Friday today. However, I have noticed the speed up over the past week. E and Fs move through 75th Interlocking without any issues during the off-peak.

This is good news. I rode through to FH on Monday just before the evening rush, and we really flew down the express despite us being pretty near the train in front. It's really heartening to see things like this change. A lot still to be done, but it's something!

The gains feel especially poignant in light of today's Times article...

Edited April 19, 2019 by RR503

[T to Dyre Avenue]

Hopefully, we don’t lose Andy Byford. He played a big part in getting the trains to speed up. Losing him (due to the other Andy’s huge ego) could undo much of the progress being made on this front.

[RR503]

I wanna post these two RFWs from the mid-90s, not just because they’re some high quality foam, but also because they really give you a good sense for how the system once performed. In video one, our  makes it from Dyre to 42nd (via Bx local) in 42 mins; today that is scheduled at 52. In the second, we do Main to QBP in just over 14 mins. It’s of course good to recognize that this degree of speed was actually unsafe in some instances, but all the same, as we head towards conversations about systemwide resignalling, it’s important to see what fixed blocks once did — if for no other reason than I don’t think Flushing CBTC can beat that  runtime.

Take a watch (and enjoy lots of old IRT signals in the first one!)

 

Edited April 24, 2019 by RR503

[RestrictOnTheHanger]

The video, 7:52 onwards even has workers on the track! That shows the chokehold that flagging rulea have on capacity

[Union Tpke]

7 hours ago, RR503 said:

I wanna post these two RFWs from the mid-90s, not just because they’re some high quality foam, but also because they really give you a good sense for how the system once performed. In video one, our  makes it from Dyre to 42nd (via Bx local) in 42 mins; today that is scheduled at 52. In the second, we do Main to QBP in just over 14 mins. It’s of course good to recognize that this degree of speed was actually unsafe in some instances, but all the same, as we head towards conversations about systemwide resignalling, it’s important to see what fixed blocks once did — if for no other reason than I don’t think Flushing CBTC can beat that  runtime.

Take a watch (and enjoy lots of old IRT signals in the first one!)

 

Are there specific locations that you can find in these two videos where the speed was actually unsafe?

[RR503]

9 hours ago, Union Tpke said:

Are there specific locations that you can find in these two videos where the speed was actually unsafe?

Hard to say exactly where the high performance of car equipment would cause control line issues, but I can tell you that starting GT protection so shortly before sharp curves as you see there at Simpson and Tremont is frowned upon, as would be making any signal calc assuming T/Os reduce speed at stations — which I imagine would get you GTs at Freeman sb, as (IINM) there are today. 

Another fun video: the Williamsburg Bridge seen through RFW before the crash: 

 

[R68OnBroadway]

Did they remove any timers on the south side of the Manhattan? The  seems to go much faster now (but DeKalb seems worse; the  now crawls and stops where it use to run through). Why don’t they just use punch boxes or install train-dispatch center transponders so they can see the train and route it without CCTV?

[RR503]

8 minutes ago, R68OnBroadway said:

Did they remove any timers on the south side of the Manhattan? The  seems to go much faster now (but DeKalb seems worse; the  now crawls and stops where it use to run through). Why don’t they just use punch boxes or install train-dispatch center transponders so they can see the train and route it without CCTV?

There already are punches at all entrance stations; it's just a matter of trusting T/Os to punch the right thing on them. Read the box at the bottom of this diagram carefully. 

[bobtehpanda]

2 hours ago, RR503 said:

There already are punches at all entrance stations; it's just a matter of trusting T/Os to punch the right thing on them. Read the box at the bottom of this diagram carefully. 

So it's basically just "trust but verify" gone haywire in the worst possible place.

I'm surprised there can't be a small device hooked up to the display board/rollsign that just verifies. Remove the human element entirely.

[RR503]

58 minutes ago, bobtehpanda said:

So it's basically just "trust but verify" gone haywire in the worst possible place.

I'm surprised there can't be a small device hooked up to the display board/rollsign that just verifies. Remove the human element entirely.

They had that on the  for years -- called IDENTRA. There was a funny looking antennae on the front of all the trains that 'told' the towers what to do with the trains, with the antennae's setting being tied to a set train ID. Today, the back-end of ATS and the coming implementation of some sort of ISIM-B replicate that train ID function, but those IDs aren't always totally correct. 

The loop protruding out of the front of the train in this pic is the IDENTRA antennae. 

Edited April 25, 2019 by RR503

[Jsunflyguy]

22 hours ago, RR503 said:

In video one, our  makes it from Dyre to 42nd (via Bx local) in 42 mins; today that is scheduled at 52. In the second, we do Main to QBP in just over 14 mins.

My copy of the A Division time table shows Dyre to Grand Central consistently at 42-48 minutes. Which is the same range that was scheduled in Fall 2004, it's not a fair characterization to say its scheduled at 52 minutes  today, that compares an off peak train with a rush hour train (there are also two trains in the midday when track work begins, once track work is out and about the timetable goes back to a 42-46 minute range). Apples and Oranges. 

Also I can't support the assertion that the trip was made in 42 minutes. There are a number of dwell times cut from the video so it may well fall into the 45 minute period which is the median in today's schedule.

Edited April 25, 2019 by Jsunflyguy

[bobtehpanda]

53 minutes ago, RR503 said:

They had that on the  for years -- called IDENTRA. There was a funny looking antennae on the front of all the trains that 'told' the towers what to do with the trains, with the antennae's setting being tied to a set train ID. Today, the back-end of ATS and the coming implementation of some sort of ISIM-B replicate that train ID function, but those IDs aren't always totally correct. 

The loop protruding out of the front of the train in this pic is the IDENTRA antennae. 

Still, if you have both IDENTRA and the TO punching and they match, that should be enough to not require the CCTV. Two sources is good enough for Boeing, apparently.

[RR503]

1 hour ago, Jsunflyguy said:

My copy of the A Division time table shows Dyre to Grand Central consistently at 42-48 minutes. Which is the same range that was scheduled in Fall 2004, it's not a fair characterization to say its scheduled at 52 minutes  today, that compares an off peak train with a rush hour train (there are also two trains in the midday when track work begins, once track work is out and about the timetable goes back to a 42-46 minute range). Apples and Oranges. 

Also I can't support the assertion that the trip was made in 42 minutes. There are a number of dwell times cut from the video so it may well fall into the 45 minute period which is the median in today's schedule.

Are we looking at the same timetable? The train in the video seems to be a weekday off-peak Dyre-BG. I pulled a bunch of trips (the 12:00, 12:08, 12:16 and 12:24 Dyre-Flatbushes, to be precise) from today's (or yesterday's, now) A division supplements. They all show 53:30 from Dyre to GC. You're right that they're padded for flagging, but excluding trackwork-caused runtime losses from the equation here I think misses a crucial point, which is that ATF/a lack of productivity in NYCT maintenance has a serious negative impact on system performance. Regardless, to the point about unpadded schedules: the trips leaving Dyre s/b between 8:30 and 9 and 3:30 and 4 all give runtimes around 50 mins...remember they just holistically padded the A division. 

I hadn't noticed cut dwells in my first time through -- but my distraction is biting me, I see some stops where the camera changes POV and probably shaves a few seconds off dwell. I doubt it adds up to 3 mins (look how short the uncut dwells are) but thanks for catching. 

55 minutes ago, bobtehpanda said:

Still, if you have both IDENTRA and the TO punching and they match, that should be enough to not require the CCTV. Two sources is good enough for Boeing, apparently.

Hell, just punching is enough. 90% of the non-ATS system works like that.

EDIT: schedules suck, so I pulled the runtime data for that grace period between the peaks and flagging time. Trips seem to hover in the 44-50 min range, so a little bit higher than before but not as high as I’d said. But of course, this is excluding flagging.

Edited April 25, 2019 by RR503

[Jsunflyguy]

22 hours ago, RR503 said:

Are we looking at the same timetable? The train in the video seems to be a weekday off-peak Dyre-BG. I pulled a bunch of trips (the 12:00, 12:08, 12:16 and 12:24 Dyre-Flatbushes, to be precise) from today's (or yesterday's, now) A division supplements. They all show 53:30 from Dyre to GC. You're right that they're padded for flagging, but excluding trackwork-caused runtime losses from the equation here I think misses a crucial point, which is that ATF/a lack of productivity in NYCT maintenance has a serious negative impact on system performance. Regardless, to the point about unpadded schedules: the trips leaving Dyre s/b between 8:30 and 9 and 3:30 and 4 all give runtimes around 50 mins...remember they just holistically padded the A division. 

I hadn't noticed cut dwells in my first time through -- but my distraction is biting me, I see some stops where the camera changes POV and probably shaves a few seconds off dwell. I doubt it adds up to 3 mins (look how short the uncut dwells are) but thanks for catching. 

Hell, just punching is enough. 90% of the non-ATS system works like that.

EDIT: schedules suck, so I pulled the runtime data for that grace period between the peaks and flagging time. Trips seem to hover in the 44-50 min range, so a little bit higher than before but not as high as I’d said. But of course, this is excluding flagging.

Aha, that is the problem, I had accidently loaded the schedule from last year. The 1200 from Dyre tipped me off as last year it was an 1159 departure, covering the trip to GC in 46 minutes (with a 2 minute hold in Dyre). Loading the correct one...quite a difference in a short period of time.  There are a lot more midday holds in the new schedule than in previous years, very interesting...that explains the miscommunication here.

As for the ATS and punching, the NYCTA has a well established pattern of treating their employees like incompetent morons, punching is indeed enough. 

[RR503]

56 minutes ago, Jsunflyguy said:

Aha, that is the problem, I had accidently loaded the schedule from last year. The 1200 from Dyre tipped me off as last year it was an 1159 departure, covering the trip to GC in 46 minutes (with a 2 minute hold in Dyre). Loading the correct one...quite a difference in a short period of time.  There are a lot more midday holds in the new schedule than in previous years, very interesting...that explains the miscommunication here.

As for the ATS and punching, the NYCTA has a well established pattern of treating their employees like incompetent morons, punching is indeed enough. 

They redid (read: lengthened) most A div schedules last year. One of the big drivers in OTP gains. 

Supplements these days are crazy. They throw 5 min holds at literally everything -- on weekends, it's not uncommon for  and  to be padded 20-25 mins above their base to control flagging delays. It's a wonder OTP isn't higher...

And yeah, NYCT's labor relations are just awful -- a massive causation behind our current set of issues, if I may say so. Speaks to the generally idiotic way labor/mgmt relationships have grown in the US (carefully externalized unions, combative management seeking to make workers into parts of assembly lines, poor governmental decision making) moreso than a narrowly MTA issue, though -- not to excuse MTA, of course. A real shame.