NYCT - Bottlenecks Discussion Thread

[Jemorie]

4 hours ago, RR503 said:

As y'all can imagine, I have a _lot_ to say on this subject, but alas I've been under a pile of work this week. Weekend, maybe.

Great. I look forward.

[Lawrence St]

Can we PLEASE discontinue service to 49th St? It's not needed since the already serves that station and causes merging delay madness at 34th St...

[Jemorie]

12 minutes ago, Lawrence St said:

Can we PLEASE discontinue service to 49th St? It's not needed since the already serves that station and causes merging delay madness at 34th St...

You’re just moving the delay north of 57 St-7 Av. It is not going to make a difference.

But yeah, the  does not need to stop at 49th Street when the combined frequencies of the  and  are enough. On top of that, the  no longer originates and terminates at 57 St-7 Av.

Edited November 1, 2019 by Jemorie

[Lawrence St]

1 hour ago, Jemorie said:

You’re just moving the delay north of 57 St-7 Av. It is not going to make a difference.

But yeah, the  does not need to stop at 49th Street when the combined frequencies of the  and  are enough. On top of that, the  no longer originates and terminates at 57 St-7 Av.

Isn't the 57th St switch at a higher MPH then the 34th St switch?

[Jemorie]

7 minutes ago, Lawrence St said:

Isn't the 57th St switch at a higher MPH then the 34th St switch?

The miles per hour does not count. It’s trains physically crossing in front of one another if they arrive there at the same time.

[Lawrence St]

1 hour ago, Jemorie said:

The miles per hour does not count. It’s trains physically crossing in front of one another if they arrive there at the same time.

Which tends to happen more at 34th St since 42nd St is a high-usage station, leading to longer dwell times. 57th St does not have this problem.

[Jemorie]

40 minutes ago, Lawrence St said:

Which tends to happen more at 34th St since 42nd St is a high-usage station, leading to longer dwell times. 57th St does not have this problem.

You’re seriously basing merging delays on crowding levels....

[Lawrence St]

5 minutes ago, Jemorie said:

You’re seriously basing merging delays on crowding levels....

Am I wrong?

Let's say a is scheduled to leave 42nd St at 8:04 AM, but because its rush hour, it dosent leave until 8:07 AM. Now theres a suppose to depart at 8:07 AM, but it has to hold because the has to leave first, adding an extra minute or two to the run time. See what I'm saying?

[RR503]

You're both right here. 34 St is a shitty place to merge trains in part because it's surrounded by stations with high dwell times, making the overall fluidity in the area even lower. 57 St theoretically would be better not because of switch speeds (I now think I may have been mistaken on switch speeds, pending a front window ride) but b/c at least in one direction, you're immediately hemmed in by a station. I personally think that, given the affinity Astoria has for 49 St, we're better off merging at 34 if we're going to have a merge -- but that's if we're going to have a merge; I'd obviously much rather we didn't. 

[Jemorie]

@Around the Horn, a little help. Do you think what @Lawrence St is saying is true? I pinged you since you live on the Brooklyn end of the  and have a better knowledge on how the line along with the other lines it interacts with along its run works in terms of actual running times, crowding, all that. I’ll be back later since I’m busy for the moment.

@RR503, thanks.

Edited November 1, 2019 by Jemorie

[Around the Horn]

It does not matter where you make the  switch from the local to the express, the simple fact that it has to switch at all is what disrupts everything. 

The  and  already run obscenely close together along 60th street in rush hour and all you need is an  and a  arriving at 57th Street southbound at the same time to completely nerf the whole trunk line. conversely with the merge at 34th Street, I (the hypothetical dispatcher that has to deal with this madness) can at least have that  train sit in Times Square for a little bit rather than in a curved tunnel in between two stations. (5th Avenue also doesn't have a southbound punch box to request that move BTW) 

I will always advocate for full deinterlining of Broadway, especially after last semester when my schedule lined up with the  trips from 96th Street. Not having to deal with the merge at all and simply flying down Broadway express was amazing and such a tantalizing taste of what Broadway and the  train can be like full time. Also helps that the operators on those trips haul ass.

[Jemorie]

Thank you. That’s all I was pointing out from the beginning. Trains that arrive at merge points at the same time, and everything is delayed 2 minutes each one by one.

[T to Dyre Avenue]

On 11/1/2019 at 5:38 PM, Around the Horn said:

It does not matter where you make the  switch from the local to the express, the simple fact that it has to switch at all is what disrupts everything. 

The  and  already run obscenely close together along 60th street in rush hour and all you need is an  and a  arriving at 57th Street southbound at the same time to completely nerf the whole trunk line. conversely with the merge at 34th Street, I (the hypothetical dispatcher that has to deal with this madness) can at least have that  train sit in Times Square for a little bit rather than in a curved tunnel in between two stations. (5th Avenue also doesn't have a southbound punch box to request that move BTW) 

I will always advocate for full deinterlining of Broadway, especially after last semester when my schedule lined up with the  trips from 96th Street. Not having to deal with the merge at all and simply flying down Broadway express was amazing and such a tantalizing taste of what Broadway and the  train can be like full time. Also helps that the operators on those trips haul ass.

I’ve taken the  up and down the full Broadway express tracks. It is pretty amazing. If  upper management would just break out of their bureaucratic inertia, then it would be the same with the . And with more trains per hour.

[Trainmaster5]

My experience with bottlenecks was different from most riders because of the types of jobs I had. I also had the advantage of experienced supervisors at key locations in Brooklyn and Manhattan. I worked PM rush hour jobs on the  for 20 years or so. In my experience with bottlenecks the delays at the Junction, while  irritating to the riders, didn’t automatically make a n/b  or  late. RR503 and some other posters are realizing that the schedule is the determining factor in lateness reports. I had a job that started at East 180th St and ran light to Bowling Green and looped for n/b service . If there was an earlier delay s/b that affected the Lexington Avenue trains heading to/from Brooklyn the Grand Central dispatcher would coordinate with the Nevins dispatcher and depending on the situation I would loop and go in service 10 minutes early. Other times the GC dispatcher would loop me at Brooklyn Bridge an I would go in service from there heading n/b . There were two instances where I was looped at the Bridge and ran light to Union Square for service. Both times that happened he would instruct my C/R and I to proceed northbound when we determined that we had a sufficient load. In my experience with the Lexington Avenue line the pm slowdown was timeframe from 4:45-5:25 pm from Borough Hall to Brooklyn Bridge and continued up to 149th St-Grand Concourse. I don’t know how many times I’ve “raced “ my friend on the  from the Bridge n/b and we would enter Grand Central together . Sometimes I would be entering East 180th St and he would call me on the radio as he was passing Whitlock. Point is that when I entered Dyre I was on time. The magic of scheduling 👌. Just my experience. Carry on.

[RR503]

11 hours ago, Trainmaster5 said:

My experience with bottlenecks was different from most riders because of the types of jobs I had. I also had the advantage of experienced supervisors at key locations in Brooklyn and Manhattan. I worked PM rush hour jobs on the  for 20 years or so. In my experience with bottlenecks the delays at the Junction, while  irritating to the riders, didn’t automatically make a n/b  or  late. RR503 and some other posters are realizing that the schedule is the determining factor in lateness reports. I had a job that started at East 180th St and ran light to Bowling Green and looped for n/b service . If there was an earlier delay s/b that affected the Lexington Avenue trains heading to/from Brooklyn the Grand Central dispatcher would coordinate with the Nevins dispatcher and depending on the situation I would loop and go in service 10 minutes early. Other times the GC dispatcher would loop me at Brooklyn Bridge an I would go in service from there heading n/b . There were two instances where I was looped at the Bridge and ran light to Union Square for service. Both times that happened he would instruct my C/R and I to proceed northbound when we determined that we had a sufficient load. In my experience with the Lexington Avenue line the pm slowdown was timeframe from 4:45-5:25 pm from Borough Hall to Brooklyn Bridge and continued up to 149th St-Grand Concourse. I don’t know how many times I’ve “raced “ my friend on the  from the Bridge n/b and we would enter Grand Central together . Sometimes I would be entering East 180th St and he would call me on the radio as he was passing Whitlock. Point is that when I entered Dyre I was on time. The magic of scheduling 👌. Just my experience. Carry on.

Via Twitter, I present this chart to illustrate just how distorting schedules can be. This is the nb  in late October:

Look at the purple line. The median  train is 6 mins behind at Supthin; if OTP was measured there, it'd be <50%. But end-of-line holds at Parsons and 169 make it so that the median  hits 179 early, so the stats work out. Whether this is good or not is a matter of perspective: on-time arrivals make managing crews that much earlier, but the distortions applied to achieve them here make terminal OTP-based performance stats all but meaningless (and this is, ofc, before we account for trip editor, flexing, people straight up entering terminal arrivals fraudulently, etc). This is why I advocate so heavily for runtime-based metrics. People can't fudge train speeds!

What I find especially galling about the above is the placement of the padding. If you're going to lengthen schedule runtimes, you may as well do it in a way that ensures your merges are timed correctly -- so put a little south of Church, a little south of Bway Laff, a little south of 36 St. Don't just put it all at the end; merge coordination is a legit operational benefit of having properly written schedules!

Edited November 5, 2019 by RR503

[Caelestor]

As many have pointed out, the  needs to stay on the express tracks and run up 2 Ave. Merging onto the local tracks, whether at 34 St or 57 St, is undeniably a theoretical capacity decrease for the entire Broadway Line because one train effectively occupies both tracks as it uses the switch. Also, if the  isn't on time, it would only delay the  if it stayed on the express tracks vs the as well. That said, changing the  also requires changing the , which is currently scheduled as a branch of the . Basically the  trains have two different northern terminals at 96 St and Astoria, two southern terminals at Whitehall and 86 St / CI, and even two different routes through Manhattan (Broadway local vs express).

What really should happen is that the  should be scheduled together, as they would share the same trunk line (Broadway Express tracks) and terminals, albeit along a different branch in Brooklyn. Then the  from Forest Hills and the  from Astoria could be scheduled together down the Broadway local tracks, with varying southern terminals at Whitehall, Bay Ridge, or Gravesend during peak hours if the other 2 terminals can't turn 25 tph. If DeKalb deinterlining becomes a reality, then the  could even share the 4 Ave express tracks while the  get Brighton to themselves, and route consolidation can propagate throughout the entire B Division. 

Long story short, deinterlining reduces the effective number of services that conflict with each other and potential delays, benefiting operators and riders alike.

[Jemorie]

1 hour ago, Caelestor said:

As many have pointed out, the  needs to stay on the express tracks and run up 2 Ave. Merging onto the local tracks, whether at 34 St or 57 St, is undeniably a theoretical capacity decrease for the entire Broadway Line because one train effectively occupies both tracks as it uses the switch. Also, if the  isn't on time, it would only delay the  if it stayed on the express tracks vs the as well. That said, changing the  also requires changing the , which is currently scheduled as a branch of the . Basically the  trains have two different northern terminals at 96 St and Astoria, two southern terminals at Whitehall and 86 St / CI, and even two different routes through Manhattan (Broadway local vs express).

What really should happen is that the  should be scheduled together, as they would share the same trunk line (Broadway Express tracks) and terminals, albeit along a different branch in Brooklyn. Then the  from Forest Hills and the  from Astoria could be scheduled together down the Broadway local tracks, with varying southern terminals at Whitehall, Bay Ridge, or Gravesend during peak hours if the other 2 terminals can't turn 25 tph. If DeKalb deinterlining becomes a reality, then the  could even share the 4 Ave express tracks while the  get Brighton to themselves, and route consolidation can propagate throughout the entire B Division. 

Long story short, deinterlining reduces the effective number of services that conflict with each other and potential delays, benefiting operators and riders alike.

Basically, under this scenario (all Broadway Express trains operating between SAS and Coney Island via Manhattan Bridge South Side / 4 Av Express, with one of them serving Sea Beach while the other serves West End), any unplanned service changes during rush hours means only Broadway-related lines are affected, instead of spreading them over to the 6 Av corridor.

For example, any issue going on the 4 Av express tracks, the Manhattan Bridge south side, and/or the Broadway express tracks, you send the  and  local (including via Montague) in either one or in both directions. Basically, the issue is minimized. Yes, there will be delays, but at least the whole system is not affected. Just because Broadway is affected does not mean everything else in the system should go down as well. Only the  are going down, leave everyone and everything else in the system alone. That is what you call a true minimization.

Otherwise, the  will experience more regular service, shorter headways, and trains arriving at their terminals on time for their next scheduled departure trips heading back the opposite direction. It’s pretty ridiculous to have the South Brooklyn BMT branches consist of two routes traveling to/from two respectively different Manhattan trunk lines (Broadway and 6 Av) when it should be one.

Edited November 5, 2019 by Jemorie

[RR503]

Broadway deinterlining is a bit like whack-a-mole. The most obvious and quick solution to the 34 St issue is to do  to 96 and increase  frequencies. That quickly leads to a capacity problem at City Hall, however. The curve is only capable of supporting 21tph of service; 15tph of  and 10tph of  is 25. The obvious solution to that? Canal St. But now you're playing relay terminal at a high ridership core station with significant loads on  trains running through during both peaks -- loads whose closest alternative is frequently the , lines to which you *really* don't want to be adding passengers. The other potential solution here (and what I'd do) is to run both the  and  to Astoria at 20tph, but we need better switches at Ditmars to do that so...🤷‍♂️. There's also, ya know, the (thorny, important, potentially transformative) question of how to replace the  on QB. 

When CBTC comes online on Bway, this whole issue will go away (provided they fix the platform sharing rule), but CBTC is likely 8-10 years away for the corridor. So we're stuck playing this game for now. 

Even if you can solve the relay terminal issue, you've neglected to address another major source of delay on the  -- one which generally overshadows 34 St in the southbound direction -- which is the Whitehall St terminal operation (see the charts at the bottom of this post for detail; the oddly shaped things are violin plots which show the distribution of data (in this case runtimes) relative to y axis, the lines are 25, 50, 75 percentile runtimes, and the bars are the difference btwn 25 and 75 percentile, which is what you should pay attention to here). Solving Whitehall essentially entails moving the terminal operation elsewhere -- there really isn't a good way to run a single pocket. Canal St isn't an option unless we get _really_ good at relaying trains, and moreover cutting  to Canal would significantly cut service on an important/under-capacity subway route through the US's third largest CBD. So Brooklyn, here we come!

But where? Cheapest would be 9th Avenue, but...ugh. Merge. On a slow-ass switch, no less. Though I will say that 9th Avenue, provided one or two signal mods, is an objectively excellent terminal, as you can hold a full length train waiting to get into the middle there outside of the line of through  traffic, and you can also achieve super easy access to 38 St yard for layups, if there's ever capacity to do so. A tradeoff worth considering.

Other options here would be 62nd St or Bay Parkway on West End (both ugly single pocket relay terminal ops, though they're close(r) to CI/could potentially work at low frequencies), Kings Highway on Sea Beach (ditto), or Bay Ridge. Bay Ridge is by far the most operationally workable, though also is pretty pricey, and would require either an acceptance of significant non-revenue miles for trains going to/from CI, the use of 9th Ave as a part time terminal for layups (soft discharge at 36 s/b, hard discharge at 9th ave, go to yard from there) or investment in yard capacity at 38 St. Pick your poison, I say!

These sort of annoying nuances are important to understand when presenting deinterlining alts. Of the above options, I would strongly advocate for investment in Astoria/38 St and the creation of a -like route from Astoria to Bay Ridge. But that means $$$ and time, so buckle up. 

Edited November 5, 2019 by RR503
added the promised graphs

[Around the Horn]

Doing that second graph for the train to Bay Ridge in the PM rush would look like a crime scene LMAO

 

I still stand by taking the off of Broadway and Queens Blvd, focusing on 4th Avenue and then going up Nassau, as a long term solution for Broadway with the and a new picking up the slack on Broadway and Queens Blvd respectively. The many merges on the current line are just simply not sustainable anymore.

[RR503]

17 minutes ago, Around the Horn said:

I still stand by taking the off of Broadway and Queens Blvd, focusing on 4th Avenue and then going up Nassau, as a long term solution for Broadway with the and a new picking up the slack on Broadway and Queens Blvd respectively. The many merges on the current line are just simply not sustainable anymore.

I second this strategy.

17 minutes ago, Around the Horn said:

Doing that second graph for the train to Bay Ridge in the PM rush would look like a crime scene LMAO

Believe the technical term is YEEEEEEK. You love to see how the beginning of  service in the AM just wrecks runtimes and variability in the area (see second graph, which measures runtimes from Cortlandt to Whitehall through the day). 

 

 

[Jemorie]

@RR503, do you, personally, like the current  or do you feel they should send all West Side express trains to one Bronx/Brooklyn branch and all East Side express trains to the other Bronx/Brooklyn branch? What’s your take on the matter?

EDIT: for me, personally, I have nothing against the current  despite the 142nd Street Junction, Rogers Junction, and Flatbush Avenue, as well as the southbound 14th Street-Union Square  gap fillers and the entire southbound platform not being equal to the northbound one without gap fillers.

Thanks.

Edited November 6, 2019 by Jemorie

[RR503]

8 minutes ago, Jemorie said:

@RR503, do you, personally, like the current  or do you feel they should send all West Side express trains to one Bronx/Brooklyn branch and all East Side express trains to the other Bronx/Brooklyn branch? What’s your take on the matter?

If you said to me that I could fix just one thing in the system, this would be it. That junction has been studied over and over and over again, and the recommendation is always the same: fix it. This is the sort of thing that won't get easier with CBTC, and is the sort of thing that's infinitely easier to do now than it will be later. I do not overstate the issue when I say that this is *the* capacity constraint on the A division. 

I'd personally do  Utica/New Lots and  Flatbush. 

[P3F]

5 hours ago, Jemorie said:

It’s pretty ridiculous to have the South Brooklyn BMT branches consist of two routes traveling to/from two respectively different Manhattan trunk lines (Broadway and 6 Av) when it should be one.

It's not really that ridiculous, because it's simply the most convenient setup for riders on the BMT branches based on ridership patterns.

[Jemorie]

4 minutes ago, P3F said:

It's not really that ridiculous, because it's simply the most convenient setup for riders on the BMT branches based on ridership patterns.

My bad. It's just the merging issues they deal with on a daily basis along their routes ( ). So any unplanned service changes that happen, all of them are affected.

[RR503]

24 minutes ago, P3F said:

It's not really that ridiculous, because it's simply the most convenient setup for riders on the BMT branches based on ridership patterns.

That was a decision made pre-Bleecker St, pre-signal mods and pre-service performance data in an era when job distributions looked quite different than they do today. I would not be at all surprised if the rider mins equation favored deinterlining today; Broadway and 6th run within a block of each other north of 14 and Dekalb is...quite the dumpster fire. 

Edited November 6, 2019 by RR503