Subway Capacity Thread

[bulk88]

3 hours ago, RR503 said:

the availability of flex capacity and competence in incident response play large roles too. The latter things are very much functions of operation and signal design, and not so much ones of the initial failure. 

Competence. BIE (this case was MOW equip on ROW). T/O sits in cab until RTO radios "train at B4-XXX identify yourself", only then triage/track occupancy protocol happens. OT baby.

[RR503]

34 minutes ago, bulk88 said:

Competence. BIE (this case was MOW equip on ROW). T/O sits in cab until RTO radios "train at B4-XXX identify yourself", only then triage/track occupancy protocol happens. OT baby.

That sort of event progression is extremely rare. The vast majority of ops radio when something is wrong. It’s in the management of the ensuing chaos that things go south — dispatchers not thinking creatively, not short turning, us not having gap trains, etc. 

[Jsunflyguy]

5 hours ago, bulk88 said:

Competence. BIE (this case was MOW equip on ROW). T/O sits in cab until RTO radios "train at B4-XXX identify yourself", only then triage/track occupancy protocol happens. OT baby.

Yeah, all 20 minutes in most cases, what a score, most T/Os groan when they hear "we need you to do another half".

 

 

4 hours ago, RR503 said:

That sort of event progression is extremely rare. The vast majority of ops radio when something is wrong. It’s in the management of the ensuing chaos that things go south — dispatchers not thinking creatively, not short turning, us not having gap trains, etc. 

It should go without saying, but monitoring a model board at a Master Tower with a model board of dozens of unidentified trains is much different than manning one interlocking with a dozen lever Model 14, eyes and ears closer to the situation. More to the point the prevailing strategy is containment from what I can tell, better to have 1 problem than 6; you have any examples of lack of creative dispatching on hand?

[RR503]

19 minutes ago, Jsunflyguy said:

It should go without saying, but monitoring a model board at a Master Tower with a model board of dozens of unidentified trains is much different than manning one interlocking with a dozen lever Model 14, eyes and ears closer to the situation. More to the point the prevailing strategy is containment from what I can tell, better to have 1 problem than 6; you have any examples of lack of creative dispatching on hand?

True, though the master towers usually get staffing proportional to their burden.

Yes, actually. Night of March 18/19. This was during 8th Ave/53 FasTrack. A bunch of work trains bound for 53 plugged the northbound local track on 6th south of/in 47-50. Now, the  and  were scheduled up that track into Queens, but north of 34 they could have followed the  in crossing to the northbound express to circumvent the disruption...but they didn’t. So what we got were some truly terrifying runtimes (including some 10ish minute dwells at 34th) as the blockage’s delay effect cascaded down the line. I was tracking the nights events, and saved a stringline partway through the mess. On mobile so can’t format to display in the page, but you can view it here. 

There are, of course, thousands of other events like this. Tonight’s massive gap in  service could have been remediated with pulling a  or  (of which there were plenty) down Culver. Some of the last two nights’ service gaps on Concourse could have been partially mitigated by running the Concourse  layups up there in service (which has, FWIW, been done before). The half-hour long gap in n/b  service that evolved this afternoon seemingly from a T/ABD could have been eased with a local . 

On the more general scale, there are things like emphasizing reverse peak skips to reduce rider impact of service-making, allowing for short turning, forcing RCC to communicate better with itself/towers, would all do massive things for service quality. 

[RR503]

So remember how I said:

On 4/2/2019 at 9:27 AM, RR503 said:

No tail tracks, and the crossover is set back a bit from the station. It's probably possible to squeeze another 2 or so out with faster turnarounds though...

It'd be exceedingly easy to create one, provided you know how to use illustrator. All the tph data is public, and train lengths are common knowledge. I may give it a shot...

Well...

Or go here for a PDF, if you prefer. 

Right now, the width of a given line is proportional only to capacity. Cars/hour will have to be a different iteration. 

As always, I'm happy to share my work. Please reuse/distribute as you wish. If you want my data or more info, PM me! Corrections/suggestions/comments are more than welcome, too. 

Edited April 15, 2019 by RR503

[RR503]

In case you would like a more minimal version of the map:

https://drive.google.com/open?id=19D5BbbUTEeryaCtQe6LadQsDs39Us3Jx

[Bay Ridge Express]

2 hours ago, RR503 said:

So remember how I said:

Well...

Or go here for a PDF, if you prefer. 

Right now, the width of a given line is proportional only to capacity. Cars/hour will have to be a different iteration. 

As always, I'm happy to share my work. Please reuse/distribute as you wish. If you want my data or more info, PM me! Corrections/suggestions/comments are more than welcome, too. 

You are literally the best

[RR503]

10 minutes ago, Bay Ridge Express said:

You are literally the best

It’s really my pleasure! I love making a good map...

[Lance]

Very nice work there @RR503. I'm glad you were able to put this together as it consolidates all of these capacity constraints conversations (try saying that three times fast) into one very easy to decipher map. If I may make a suggestion: for services like the and , I'd recommend using eastbound / westbound instead of north / south, though that's just a personal preference.

[ibroketheprinter]

Quick beginner question here. What actually limits train capacity to headways like 8-20 minutes? What could be done to bring the TPH up?

[Lex]

18 minutes ago, ibroketheprinter said:

Quick beginner question here. What actually limits train capacity to headways like 8-20 minutes? What could be done to bring the TPH up?

That depends on what you're referring to. Some areas have lower ridership potential than others, as do certain times. Some things that hamper throughput are inherent design flaws, likely due to a failure to future-proof (I can think of a couple of areas off the top of my head).

There are others, but because there are many factors, I won't even try to get into every one of them.

[RR503]

2 hours ago, ibroketheprinter said:

Quick beginner question here. What actually limits train capacity to headways like 8-20 minutes? What could be done to bring the TPH up?

The most common limit is NYCT’s myopic service planning. The agency plans its system’s capacity piecewise — when making decisions about frequency and routings, it almost without exception neglects to account for dynamics between lines and between service and ridership development. What you end up with are disasters like the CPW corridor, where you have a line that once ran 64tph into the core running 30 all the while paralleling lines that face serious issues (think Jerome-Lex and IRT Broadway), or areas where lines that are at capacity incite development along them because they have decent service, while others with poor service (but spare capacity) remain largely stagnant (think  versus ).

Then there are the crowding guidelines that the agency uses themselves. A full train per NYCT is a train with three square feet per standee — a crowding level that is essentially inoperable, what with the dwell issues such loads cause. Simply raising those standards would have a significant impact on the image of lines that are deemed to be at capacity. 

But service planning isn’t, of course, the only issue. The reality of our situation is that NYCT is pretty objectively bad at running its railroad. We lose 4tph on Lex express because we don’t know how to control dwell, 6tph over the Williamsburg Bridge because competently designed signal systems seem to be beyond our reach, 10tph of Manhattan-Queens capacity because of poor terminal operation at Forest Hills, 15tph Manh-Queens because of poor routing choices through the Queens tunnels, 16 or 17tph Manh-Brooklyn because running sensical service patterns through Rogers Jct is just beyond us, and 20tph Manh-Brooklyn because the agency just *had* to reinvent the wheel when designing train identification systems at Dekalb Avenue interlocking. These are just the highlights of what would be an encyclopedia-length dive into how 40 or so years of operational amnesia and institutional myopia have systematically eroded our subway’s capability — erosion which absolutely plays a role in the low throughputs you see on the map. 

The final category is infrastructural limitation. The Ditmars Boulevard terminal is a great case study here — it reduces potential throughput by 15tph through its inefficiency. South Ferry  and the system’s various flat junctions make up most of the rest of this category.  

This is of course just an overview of the capacity problems we face. I’m happy to go into more detail, and there’s already a lot that I and others have written here and elsewhere on these issues. 

[ibroketheprinter]

7 minutes ago, RR503 said:

The most common limit is NYCT’s myopic service planning. The agency plans its system’s capacity piecewise — when making decisions about frequency and routings, it almost without exception neglects to account for dynamics between lines and between service and ridership development. What you end up with are disasters like the CPW corridor, where you have a line that once ran 64tph into the core running 30 all the while paralleling lines that face serious issues (think Jerome-Lex and IRT Broadway), or areas where lines that are at capacity incite development along them because they have decent service, while others with poor service (but spare capacity) remain largely stagnant (think  versus ).

Then there are the crowding guidelines that the agency uses themselves. A full train per NYCT is a train with three square feet per standee — a crowding level that is essentially inoperable, what with the dwell issues such loads cause. Simply raising those standards would have a significant impact on the image of lines that are deemed to be at capacity. 

But service planning isn’t, of course, the only issue. The reality of our situation is that NYCT is pretty objectively bad at running its railroad. We lose 4tph on Lex express because we don’t know how to control dwell, 6tph over the Williamsburg Bridge because competently designed signal systems seem to be beyond our reach, 10tph of Manhattan-Queens capacity because of poor terminal operation at Forest Hills, 15tph Manh-Queens because of poor routing choices through the Queens tunnels, 16 or 17tph Manh-Brooklyn because running sensical service patterns through Rogers Jct is just beyond us, and 20tph Manh-Brooklyn because the agency just *had* to reinvent the wheel when designing train identification systems at Dekalb Avenue interlocking. These are just the highlights of what would be an encyclopedia-length dive into how 40 or so years of operational amnesia and institutional myopia have systematically eroded our subway’s capability — erosion which absolutely plays a role in the low throughputs you see on the map. 

The final category is infrastructural limitation. The Ditmars Boulevard terminal is a great case study here — it reduces potential throughput by 15tph through its inefficiency. South Ferry  and the system’s various flat junctions make up most of the rest of this category.  

This is of course just an overview of the capacity problems we face. I’m happy to go into more detail, and there’s already a lot that I and others have written here and elsewhere on these issues. 

Thank you for your explanation!

 

A few follow-up questions:

 

¬ What more specific changes could potentially made to solve the issues?

¬ How could terminals be redesigned to increase capacity?

¬ On the subject of physical limitations, could current infrastructure be altered to reduce bottlenecks?

¬ Are the majority of the issues managerial or the aforementioned infrastructural limitations?

 

Just my own take:

 

I am by no means knowledgeable about the field of rapid transit but as someone who has crawled into Dekalb from Atlantic on the Q in the mornings, sat at a red signal for 2-3 minutes before moving once again, and has watched as their local C overtook the A on the express track and even managed to beat the A from 86 (where the A train passed on the express as our train was leaving) to 59th and again to 34th, something is amiss. Common sense would dictate that taking express trains would provide a faster ride than a local but more often than not I find myself choosing local over express. Not a totally relevant example but nonetheless: The 6:51 from Centre Av to Penn usually takes about 40-45 minutes to get from the previously mentioned stations. However, the 6:24 to Atlantic which there is a direct connection to Penn  takes 35-40. Most people would then assume the 624 is an express but in fact it calls at Lynbrook, Valley, Rosedale, Laurelton, Locust, Jamaica, (connect express to Penn) (Rosedale, Laureton, and Locust Manor usually take longer to board because of the amount of people boarding) Whereas the 651 calls at only Lynbrook, Valley, Jamaica, Penn (Faster boarding times than 624, too, because fewer passengers)

[RR503]

2 minutes ago, ibroketheprinter said:

 ¬ What more specific changes could potentially made to solve the issues?

 ¬ How could terminals be redesigned to increase capacity?

¬ On the subject of physical limitations, could current infrastructure be altered to reduce bottlenecks?

¬ Are the majority of the issues managerial or the aforementioned infrastructural limitations?

A lot of the change that needs to happen is managerial. Things like revising service guidelines to account for inter-corridor dynamics, creating a unified development planning process where MTA and NYC sit at a table and identify not just the areas where demand warrants upzoning, but also areas where excess subway cap allows it and changing train routings to simplify merge patterns and eliminate 'shadow capacity' could all help on the more broad level. On the nitty-gritty operations side, my 'greatest hits' would be (and this is in no way exhaustive) changing operations procedures to facilitate reliable high capacity operation (so, for example, encouraging use of station time cutbacks, of local recycle, of a train's full acceleration and braking potential), changing terminal procedures to reduce terminal congestion (think eliminating fumigation and enforcing the 'board your outbound train on the inbound platform' rule at relay terminals), revising maintenance procedures to minimize track time and service disruption, systematically reviewing signal system design to find places where capacity/speed could safely and simply be increased, and reviewing operations at all major junctions.

The latter point, for whatever it's worth, provides a great example of how bad ops can kill -- one that I probably overuse, but feels relevant given your mention of issues at Dekalb. The junction there is congested by the fact that they use cameras placed at the home signals entering the interlocking to identify trains, forcing them to stop mid-tunnel and thus reducing capacity -- all of this despite the fact that they have punchboxes at entrance stations, whose information they could use to assign routings as they do everywhere else in the B division. And just like that, we go from 60tph over the Bridge to 40. 

There also should be a priority infrastructure review. Resignallings cost money; terminal and interlocking rebuilds even more. Generally, we need to learn how to spend money properly in this city, but in the realm of subway infrastructure the need is especially pressing. The overwhelming build mentality that's come to define our region blinds us to all the smaller, more boring (yet more impactful) things we could do to infrastructure to make our system work better. So not to say we shouldn't build SAS 2, but if even a fraction of 5.5 billion dollars went to things like a reconfig of the Astoria terminal along the lines of the PCAC's (excellent) plan, or grade separations of Myrtle and 142, or a resignalling of the approaches to the Williamsburg bridge, or a rebuild of Marcy into a 3 track/2 island platform station over the bus terminal, I can almost guarantee we'd see more benefit per dollar spent than through pure expansion. Again, we have a lot of capacity lying around; let's learn to use it. 

This once again feels lacking in detail. Please ask more questions if you have!

[Bay Ridge Express]

5 hours ago, RR503 said:

The latter point, for whatever it's worth, provides a great example of how bad ops can kill -- one that I probably overuse, but feels relevant given your mention of issues at Dekalb. The junction there is congested by the fact that they use cameras placed at the home signals entering the interlocking to identify trains, forcing them to stop mid-tunnel and thus reducing capacity -- all of this despite the fact that they have punchboxes at entrance stations, whose information they could use to assign routings as they do everywhere else in the B division. And just like that, we go from 60tph over the Bridge to 40. 

I don't understand why they thought this was a good idea. What's wrong with placing punch boxes there? Furthermore, it interrupts (what was supposed to be) an artistic display that riders could enjoy on the NB tracks at the original Myrtle Av station. Now, it's used as a stopping point for trains, and you can barely even notice the artwork there.

[Union Tpke]

@RR503 Do you know anything about this project or have any thoughts on it?

S33931
LIFE CYCLE REPLACEMENT OF SPEED ENFORCEMENT SYSTEMS**
$10M - $50M

This project will install overspeed protection equipment at five locations system-wide.

All existing speed sensor heads shall be replaced.

A time control aspect shall be installed on every signal of the Fallback Grade Time system, as required.

Replace cables and related equipment, as required.

Furnish speed control signs, as required.

Duration of Contract 49 Months

[RR503]

7 hours ago, Union Tpke said:

@RR503 Do you know anything about this project or have any thoughts on it?

S33931
LIFE CYCLE REPLACEMENT OF SPEED ENFORCEMENT SYSTEMS**
$10M - $50M

This project will install overspeed protection equipment at five locations system-wide.

All existing speed sensor heads shall be replaced.

A time control aspect shall be installed on every signal of the Fallback Grade Time system, as required.

Replace cables and related equipment, as required.

Furnish speed control signs, as required.

Duration of Contract 49 Months

This seems to be wheel detector replacement...lemme ask around. 

[Union Tpke]

50 minutes ago, RR503 said:

This seems to be wheel detector replacement...lemme ask around. 

Thanks. I thought they were getting away from using wheel detectors.

[Around the Horn]

Is there a way to calculate how much theoretical capacity we're losing by using gap fillers at Union Square on the ?

[Deucey]

1 hour ago, Around the Horn said:

Is there a way to calculate how much theoretical capacity we're losing by using gap fillers at Union Square on the ?

Isn’t USQ a 15mph track and a 30-40 second stop because of the gap fillers?

[RR503]

Gap fillers aren't actually what limit Lex -- it's GC and now the succession of lower Manhattan stations too. 

For the theoretical calculation, though, simply find the throughput difference between your base headway (jn seconds) and your base headway +10; that'll get you a relatively decent approximation of the situation. 

Edited May 7, 2019 by RR503

[Lex]

1 hour ago, RR503 said:

Gap fillers aren't actually what limit Lex -- it's GC and now the succession of lower Manhattan stations too. 

For the theoretical calculation, though, simply find the throughput difference between your base headway (jn seconds) and your base headway +10; that'll get you a relatively decent approximation of the situation. 

In fairness, they definitely slow operations, especially when they fail.

[RR503]

2 minutes ago, Lex said:

In fairness, they definitely slow operations, especially when they fail.

Absolutely, but there’s a difference between being slow and being the limiting factor on route capacity, which is what I was trying to get at above. 

[Deucey]

1 hour ago, RR503 said:

Gap fillers aren't actually what limit Lex -- it's GC and now the succession of lower Manhattan stations too. 

For the theoretical calculation, though, simply find the throughput difference between your base headway (jn seconds) and your base headway +10; that'll get you a relatively decent approximation of the situation. 

How does GC limit? The S curve doesn’t seem that severe - what is it, 20 mph?

[RR503]

7 minutes ago, Deucey said:

How does GC limit? The S curve doesn’t seem that severe - what is it, 20 mph?

It’s the dwell times and to a lesser extent the reluctance to use ST that kills there. In Lower Manhattan, the latter issue becomes the primary issue, and you can throw in Bowling Green congestion in too.