What Would Really Raise Terminal Capacity?

[CenSin]

On most lines, there’s always the frustrating delay for trains entering terminals. Inbound trains are often held outside the terminals (sometimes several stations behind) to wait for outbound trains to leave the terminal as scheduled. That means if an outbound train is scheduled to leave every 6 minutes and there’s a conga line of 4 trains from West 8 Street to Sheepshead Bay, that train at Sheepsheead Bay is going to experience an extra 18~24 minutes of delay. The , similarly, might be held at Kings Highway, on the express track between stations, and on the tracks approaching Brighton Beach. I’ve heard similar problems on the and such as well. It’s not how trains are supposed to be run for best service, but it is a byproduct of physical constraints and/or tight operational budget.

Often, I hear the solution is to just increase the frequency of train service (a.k.a. just overclock the damn system), but that adds an ongoing (operational) cost which may be feasible in the best fiscal situation but not perpetually.

The other solution is just to have more terminal capacity, which is a one-time cost and (mostly) not subject to the ebbs and flows of operational dollars. From what I’ve observed on the /, the conga lines form often on the Flushing end despite 3 terminal tracks. I’m attributing this to the large footprint of the switches outside the station and the lack of tail tracks. In Queens, the does have 6 tracks it can turn trains on, but 2 tracks must pass through to dump passengers before turning. I don’t know how smooth that is in practice.

I could have sworn seeing early plans for the 34 street–Hudson Yards station include 3 tracks and 2 platforms. I cannot find the diagrams, but how much would that have alleviated the problem of trains waiting to enter the terminal station? There’s also the tentatively triple-tracked 125 Street terminal station planned for 2 Avenue’s . How much would that help with terminal operations? I can imagine the frustration of missing connections to trains since that is going to be a moderately busy transfer complex.

[Lawrence St]

Forest Hills. Please, PLEASE, can we extend the and to 179th St during rush hours, I can not stand the delays that occur from laying up at Forest Hills.

[Deucey]

3 hours ago, CenSin said:

Often, I hear the solution is to just increase the frequency of train service (a.k.a. just overclock the damn system), but that adds an ongoing (operational) cost which may be feasible in the best fiscal situation but not perpetually.

One of the things I realized playing Jan Bochmann's BAHN years ago is that you can reduce the wait if you turn trains back faster - if it's 4 minute service, have the train on the occupying platform 15-20 seconds before the next train arrives (even if it's a bumper blocks at the station situation instead of tailback tracks). 

But that's a "I programmed this in my spare time" simulation. Real life is different because with all the technical things that could be done, human emotions/life factors in. Train staff could be late because of low morale or bathroom runs or someone talking to them. So unless the same crew that brought the train in is the same one that takes it out, that "break between runs" thing is going to always make terminal delays a regular thing - even if the departure time is set up to be before the next train arrives, or the speed across switches to/from the entering track is sped up, etc.

You technically could stop delays by running 1 or 2 fewer trains on the line (since if most terminals are like  or  with no tailbacks, there's two trains platformed when the next train arrives) so there's always an open slot at the terminal, but due to the aforementioned, there's always a chance that the delay isn't operations-based and could be human or mechanical. Plus there may be a gap in service in the run that causes overcrowding that causes another delay.

TL;DR version:

[Trainmaster5]

6 hours ago, Deucey said:

One of the things I realized playing Jan Bochmann's BAHN years ago is that you can reduce the wait if you turn trains back faster - if it's 4 minute service, have the train on the occupying platform 15-20 seconds before the next train arrives (even if it's a bumper blocks at the station situation instead of tailback tracks). 

But that's a "I programmed this in my spare time" simulation. Real life is different because with all the technical things that could be done, human emotions/life factors in. Train staff could be late because of low morale or bathroom runs or someone talking to them. So unless the same crew that brought the train in is the same one that takes it out, that "break between runs" thing is going to always make terminal delays a regular thing - even if the departure time is set up to be before the next train arrives, or the speed across switches to/from the entering track is sped up, etc.

You technically could stop delays by running 1 or 2 fewer trains on the line (since if most terminals are like  or  with no tailbacks, there's two trains platformed when the next train arrives) so there's always an open slot at the terminal, but due to the aforementioned, there's always a chance that the delay isn't operations-based and could be human or mechanical. Plus there may be a gap in service in the run that causes overcrowding that causes another delay.

TL;DR version:

I thought you almost hit the nail on the head. You had the right idea.  The best terminals, in the IRT as I was taught, were two of the oldest ones. Same crew brings the train in and departs on it. Brooklyn Bridge on the and old South Ferry on the   and the   back when it used the outer loop. I worked "road" jobs on the and the and many times the East or West Side would fall apart for one reason or another and we'd be turned back north at South Ferry, Bowling Green or Brooklyn Bridge. The average rider wouldn't know the circumstances ( and wouldn't care) but many times we would arrive outside the terminal and have to wait. Sometimes it might be 8 or 10 minutes before we would enter the station and I'd hear the complaints. Sometimes people would understand, sometimes they wouldn't. Fact of the matter was that we weren't even supposed to be there at that time. Maybe the leaving crew was on lunch or a bathroom break (comfort relief). Many times when I worked the we would actually have to clean the train out and I'd have to lay it up north of the station or relay it south of the station so that the scheduled train and crew could enter the terminal. Imagine the backups after a major delay when things clear up and there's a conga line of trains outside the terminal. Finishing crews, starting crews, and crews scheduled to make a trip in the opposite direction that haven't arrived yet converging on the terminal's doorstep.. My C/R and I had a good rapport with the supervision at Dyre / East 180/ Utica so it usually worked out for us but I've seen situations where things got rather testy between supervision and the hourlies. The average rider doesn't know or care about those things but it does impact service to a large extent. Just my observations. Carry on.

[JeremiahC99]

On 3/6/2020 at 2:24 PM, Lawrence St said:

Forest Hills. Please, PLEASE, can we extend the and to 179th St during rush hours, I can not stand the delays that occur from laying up at Forest Hills.

This one can be resolved not be extending them, but rather doing some policy change when it comes to fumigation at the Forest Hills terminal. Once we exhaust that, then we can extend the lines to 179th Street.
 

I do have an extension of the local lines to 179 planned as part of an extension of the Queens Blvd express tracks to Springfield Blvd, but that may not come to fruition.

[Italianstallion]

On 3/6/2020 at 11:27 AM, CenSin said:

On most lines, there’s always the frustrating delay for trains entering terminals. Inbound trains are often held outside the terminals (sometimes several stations behind) to wait for outbound trains to leave the terminal as scheduled. That means if an outbound train is scheduled to leave every 6 minutes and there’s a conga line of 4 trains from West 8 Street to Sheepshead Bay, that train at Sheepsheead Bay is going to experience an extra 18~24 minutes of delay. The , similarly, might be held at Kings Highway, on the express track between stations, and on the tracks approaching Brighton Beach. I’ve heard similar problems on the and such as well. It’s not how trains are supposed to be run for best service, but it is a byproduct of physical constraints and/or tight operational budget.

Often, I hear the solution is to just increase the frequency of train service (a.k.a. just overclock the damn system), but that adds an ongoing (operational) cost which may be feasible in the best fiscal situation but not perpetually.

The other solution is just to have more terminal capacity, which is a one-time cost and (mostly) not subject to the ebbs and flows of operational dollars. From what I’ve observed on the /, the conga lines form often on the Flushing end despite 3 terminal tracks. I’m attributing this to the large footprint of the switches outside the station and the lack of tail tracks. In Queens, the does have 6 tracks it can turn trains on, but 2 tracks must pass through to dump passengers before turning. I don’t know how smooth that is in practice.

I could have sworn seeing early plans for the 34 street–Hudson Yards station include 3 tracks and 2 platforms. I cannot find the diagrams, but how much would that have alleviated the problem of trains waiting to enter the terminal station? There’s also the tentatively triple-tracked 125 Street terminal station planned for 2 Avenue’s . How much would that help with terminal operations? I can imagine the frustration of missing connections to trains since that is going to be a moderately busy transfer complex.

I don't believe the proposed 2nd Ave. terminal at 125th is going to have 3 tracks any longer. That plan was dropped.

[darkstar8983]

Extending the  and  to 179 St would make fumigation worse because now you have to fumigate  trains (and the rush hour on the express track) and the  on the local and adding unnecessary mileage because the train will be carrying air past 71 Av

[Eric B]

Being caught behind this many times, I've said they should extend only the ones actually laying up (not simply relaying back in service). And they could run down the express to Parsons where they are not in the way of anything else, then get everyone off, and then use the switches to head back to the yard or the layup positions. (though I think they've gone back to running the 179 's down the express, and I thought of this at a time when they all were dropping out at Kew Gardens. Still, it would be much less disruptive than the local track at Continental).

[Jova42R]

In terms of QBL, why not run the  to 179 via local and the  to Jamaica via Express, with rush hour s express to Jamaica? Couldn’t that work?

For 168th, I know that this is controversial, but why not run some s to 181 or 207 on the  and cut some s to 168

Thoughts @CenSin @Eric B

[Jsunflyguy]

9 hours ago, Jova42R said:

In terms of QBL, why not run the  to 179 via local and the  to Jamaica via Express, with rush hour s express to Jamaica? Couldn’t that work?

For 168th, I know that this is controversial, but why not run some s to 181 or 207 on the  and cut some s to 168

Thoughts @CenSin @Eric B

Jamaica, being built as a short-line location and not a terminal cannot support  and , in fact during the Rush it can't even support all the  service, partly the reason for the Hillside  . Also New Yorkers have a well documented history of the express train-dive, all the Hillside people forced onto the  will simply run over to the  and pack those trains making them even more crowded. There is also the matter of putting all of the East Queens Riders onto the local will cause a ridership shift on the Buses which may not be absorbable by the current surface network and is generally an election loser for anyone within a 3 vote radius of the idea.

Things that raise capacity:

1) Loops
2) Adequate crew facilities

3) Cooperative ridership

4) Low-conflict routes for trains entering/leaving service.

5) Reliable enroute service (no brainer) 

6) Resilient route structure, for example the QBL should have a crossover for Exp-Local & Local-Exp so that one intransigent passenger doesn't nuke service. There are enough a**holes in New York that this is a persistent vulnerability. 

[N6 Limited]

At Jamaica Center, why can't they move trains out of the way onto the stub tracks, to alleviate terminal congestion?

[P3F]

The switch layout at 33rd Street PATH (a three track terminal) is much more compact than the one in Flushing, which in theory should increase capacity by reducing the time trains need to traverse the interlocking. However, I do not ride PATH often, so I don't know if it works well in practice.

[NewFlyer 230]

15 hours ago, N6 Limited said:

At Jamaica Center, why can't they move trains out of the way onto the stub tracks, to alleviate terminal congestion?

The crossover in between Jamaica Center and Sutphin Blvd was built in the middle of the tunnel section so it cuts down capacity with how long it takes for the train to cross.  trains usually have to wait at Sutphin Blvd for a train to leave Jamaica Center because of that. Plus past Jamaica Center, the stub tracks don’t have a crossover so they simply couldn’t cross over to alleviate delays. 
That’s why they have those extra  trains go to 179th while the  travels via Hillside. 
 

In the original plan the  and  local trains were suppose to go to Jamaica Center and it would have been interesting to see how that would have played out if the MTA went along with that plan. People would have definitely bailed out at Kew Gardens, and I could see a bit of a merging issue at Briarwood with the local and express trains.

 Can you imagine a  and  going to Jamaica Center? I can tell you that they most likely would have the same problems as the  currently has with capacity. 

[T to Dyre Avenue]

They wouldn’t be able to send both the  and  there. One of them, but not both. They already can’t operate all rush hour  service out of JC, so never mind sending all locals there. If the  and  were still the QB locals, they would probably have the same issue. Including bailing for the  at Union Turnpike, especially since the full time local would have been the , which then as now, doesn’t go to Manhattan. Too bad they can’t seem to figure out to fit another crossover switch in a location closer to the platform, so they would then be able to run more  service.

Edited March 11, 2020 by T to Dyre Avenue

[N6 Limited]

3 hours ago, NewFlyer 230 said:

The crossover in between Jamaica Center and Sutphin Blvd was built in the middle of the tunnel section so it cuts down capacity with how long it takes for the train to cross.  trains usually have to wait at Sutphin Blvd for a train to leave Jamaica Center because of that. Plus past Jamaica Center, the stub tracks don’t have a crossover so they simply couldn’t cross over to alleviate delays. 

What I mean is like if there is a conga line into Jamaica Center, instead of holding at Sutphin, Jamaica-Van Wyick, etc, waiting for a train to leave Jamaica Center, couldn't they send a train into each stub, so that another train can enter the station?

[Lex]

23 minutes ago, N6 Limited said:

What I mean is like if there is a conga line into Jamaica Center, instead of holding at Sutphin, Jamaica-Van Wyick, etc, waiting for a train to leave Jamaica Center, couldn't they send a train into each stub, so that another train can enter the station?

Again, there's no crossover there.

In other words, once that space is filled, anything sitting there is stuck until traffic dies down. The only way to address that is to install one while leaving enough space to slot some trains.

[KingsbridgeKid2532]

51 minutes ago, Lex said:

Again, there's no crossover there.

In other words, once that space is filled, anything sitting there is stuck until traffic dies down. The only way to address that is to install one while leaving enough space to slot some trains.

When I lived in Norwood the D would relay into the tail tracks and for the most part that worked. But there where times the D's would have to wait in between BPB and 205th or just be held at BPB.

[RR503]

On 3/9/2020 at 8:04 PM, Jsunflyguy said:

1) Loops

Agree with everything here except for this. Loops are good for terminal fluidity, but they wreck your ability to mitigate pipeline issues -- your terminal becomes GIGO. Would add tail tracks and well designed signals entering the terminal to the list as well -- that latter issue plays a role at JC. 

Edited March 12, 2020 by RR503

[Deucey]

6 hours ago, RR503 said:

Agree with everything here except for this. Loops are good for terminal fluidity, but they wreck your ability to mitigate pipeline issues -- your terminal becomes GIGO. Would add tail tracks and well designed signals entering the terminal to the list as well -- that latter issue plays a role at JC. 

I don’t necessarily think they’re bad, but I do think that the only “good” one in NY is the City Hall Loop since it’s long enough to hold several trains because of City Hall station.

Contrast that with South Ferry with the gap fillers and the 2-3 minute wait at Rector St to move people to the first 5 cars and let the  ahead GIGO.

The lines terminating in Chicago’s loop work well even though the trains keep moving to do the return trip and level cross active tracks - but CTA has some savvy dispatchers.

Only loop I can really think of that should be a model for  if they do loops is LA’s Transit Mall/Downtown Long Beach Station, but it’s a two-track terminal island station that holds two trains with the third holding roughly a minute while one departs - at least during rush hour a few years ago (lord knows how that’ll work when there’s more trains after the Regional Connector opens in the next two years).

I digress to say that loops may not do the job in NY that they could because dispatching and them being single tracks. If South Ferry Loop had an island platform, reliable gap fillers (or could’ve been straightened), I think it could’ve turned as many trains as New South Ferry does - even without CTA-level dispatching. (edit) But it would either need to do no crew change or that new crew would have to be on the platform on arrival to do pre-checks and whatnot.

Edited March 13, 2020 by Deucey

[Trainmaster5]

Again I have to admire the way you think. You might have been taught by my old instructors 😀. The old KISS way of thinking. RR503 did point out the capacity vs the velocity difference so I'll just add my interpretation to that. With a perfect loop configuration there's no need for a crew change and the standard brake tests that must be done as per rule. Train arrives, doors open, passengers exit, enter, doors close, train moves on. Limited dwell time. Typical 2 track station means signal and switch lineup into the station, train dumps air and doors open. Train across the platform closes down while switch and signals align and proceeds. All's well and good if everything works as designed. Let there be a signal or switch problem, especially across the switches, and you have now delayed 2 trains at the terminal. Look back at the recent service advisories and it appears that signal, followed by switch, delays predominate. Two of my instructors were line guys from the old school. It was those gentlemen who instilled the KISS mentality in our heads. When I worked on the work trains and I noticed the diamond crossovers being removed throughout the IRT I asked one of them the reasoning behind it.  You guessed it, KISS. The fewer variables in the equation meant better service in the long run. Just my take. Carry on.

[RR503]

10 hours ago, Trainmaster5 said:

Again I have to admire the way you think. You might have been taught by my old instructors 😀. The old KISS way of thinking. RR503 did point out the capacity vs the velocity difference so I'll just add my interpretation to that. With a perfect loop configuration there's no need for a crew change and the standard brake tests that must be done as per rule. Train arrives, doors open, passengers exit, enter, doors close, train moves on. Limited dwell time. Typical 2 track station means signal and switch lineup into the station, train dumps air and doors open. Train across the platform closes down while switch and signals align and proceeds. All's well and good if everything works as designed. Let there be a signal or switch problem, especially across the switches, and you have now delayed 2 trains at the terminal. Look back at the recent service advisories and it appears that signal, followed by switch, delays predominate. Two of my instructors were line guys from the old school. It was those gentlemen who instilled the KISS mentality in our heads. When I worked on the work trains and I noticed the diamond crossovers being removed throughout the IRT I asked one of them the reasoning behind it.  You guessed it, KISS. The fewer variables in the equation meant better service in the long run. Just my take. Carry on.

KISS is generally a good principle by which one may run the railroad, but I do think it’s worth recognizing the ways loops hamper dispatchers’ ability to mitigate service problems. If something happens on the southbound  before the PM rush, there’s a near 100% chance that the problem will ricochet back north thanks to the fact that you have little to no terminal recovery time. This in and of itself is a tradeoff — recovery time is an operating cost — but that sort of operating margin really saves the railroad from having to be constantly skipping/holding/pissing off riders. I again don’t think this is an issue that’s clearly best one way or another, just again, tradeoffs....

[Union Tpke]

13 hours ago, Trainmaster5 said:

Again I have to admire the way you think. You might have been taught by my old instructors 😀. The old KISS way of thinking. RR503 did point out the capacity vs the velocity difference so I'll just add my interpretation to that. With a perfect loop configuration there's no need for a crew change and the standard brake tests that must be done as per rule. Train arrives, doors open, passengers exit, enter, doors close, train moves on. Limited dwell time. Typical 2 track station means signal and switch lineup into the station, train dumps air and doors open. Train across the platform closes down while switch and signals align and proceeds. All's well and good if everything works as designed. Let there be a signal or switch problem, especially across the switches, and you have now delayed 2 trains at the terminal. Look back at the recent service advisories and it appears that signal, followed by switch, delays predominate. Two of my instructors were line guys from the old school. It was those gentlemen who instilled the KISS mentality in our heads. When I worked on the work trains and I noticed the diamond crossovers being removed throughout the IRT I asked one of them the reasoning behind it.  You guessed it, KISS. The fewer variables in the equation meant better service in the long run. Just my take. Carry on.

Which specific crossover removals are you talking about?

[CenSin]

So I gave the question some more thought, and it seems that we have three different problems which are being conflated:

1. Terminal capacity: the number of trains that can hold in the terminal simultaneously. Terminal capacity can be best thought of as analogous to battery capacity (watt-hours). A battery with high capacity can sustain a system for longer before it runs out. It can also absorb a greater surge of power.
2. Terminal bandwidth: the number of trains that can be turned back per unit of time. Terminal bandwidth can be best thought of as analogous to battery capability (amp-hours). A battery may have any capacity, but it will only be able to accept/discharge a certain amount of power per unit of time.
3. Scheduled frequency: the number of trains requested to enter/leave the terminal.

The three are different problems, but related in that they both contribute to delays. Being able to accept more trains in the terminal solve the problem of delays caused by surges of trains during rush hour. Being able to turn back train faster keeps the terminal tracks open to accept more trains. And scheduling the number of trains to take advantage of the terminal bandwidth keeps trains rotating in and out of the terminal. “Procedures” should be self-explanatory.

Most importantly, improvements in these metrics also improve the one thing that matters to the passengers: how long is the train going to take to leave my station and get to the station I want to be at? And the key seems to be separating inbound traffic from outbound traffic as close to the terminal as possible. Loop stations have a particular advantage in that inbound = outbound just like any other intermediate station along any line, so in some cases another train can pull into the terminal before the previous one has even pulled out completely if the signals are spaced closely enough.

[Trainmaster5]

3 hours ago, Union Tpke said:

Which specific crossover removals are you talking about?

The big ones were 168th St on the and 149th St-Concourse at the north end on the and line. From my understanding the diamond crossovers were deemed too costly to maintain. I was a work train guy but I believe it was a track and signal decision. I know that RTO had no say so.  My recollections.  Carry on. 

[bobtehpanda]

17 hours ago, Deucey said:

I don’t necessarily think they’re bad, but I do think that the only “good” one in NY is the City Hall Loop since it’s long enough to hold several trains because of City Hall station.

There are a couple of reasons why we don't build loops anymore (and even Paris does not, and their Metro historically uses lots of loops)

• Cost & space: they take up more space, which cost more.
• Constructability; particularly with cut and cover loops require lots of curves, which usually require at least some takings, increasing the cost
• They are hard to extend. This is a big thing; the Paris Metro stopped constructing loops, particularly loops that had single-platform stations alongside the single track portion, because it's a massive PITA to have to build two tunnels going different paths than two or one tunnel in the same path. It has also complicated slightly the construction of the Northern Line extension to Battersea in London.