Late Trains Now 2 Minutes

[BuddR32]

Under a provision in the newly adopted state budget, the LIRR can no longer report a train as on time if it arrives at its final station later than two minutes after scheduled time. This is a third of the current 5:59 allowance.

Expect OTP to plummet.

[Head-end View]

I think this is overkill. I always thought the 5:59 rule was reasonable for commuter trains. Anything less than that is really not a significant delay.

[Absolute-Limited Advance-Approach]

I think this is overkill. I always thought the 5:59 rule was reasonable for commuter trains. Anything less than that is really not a significant delay.

Almost nothing in daily is timed down to 2 minute precision. 2 Minutes is basically catching 2 15 codes (based on a 12 car train). It's just politics maneuvering to tank LIRR's image. I guess Passengers can get ready for Conductors to close down in peoples faces.

[Commuter X]

My prediction is that the LIRR will pad the schedules (just like the airlines do) to maintain their stellar record.

[DutchRailnut]

the 5 minute and 59 seconds is a nation wide standard set by APTA. main reason for it is accounting and timekeeping functions.
its a lot easier to go by 1/10 of hour segments, that if by doing it in 1/30 increments.

[EuroStar]

The problem with the new and old rules is that they both allow the railroad to pad the time between the last two stations in order to show on-time performance. For example, the way this works is on NJT is by padding the time between say Secaucus and NYP by 5-6 minutes during the morning rush hour. Nobody complains if their trains arrives at NYP early, so the padding helps the on-time performance regardless where the delay occurred. Padding between intermediate stations occurs too, but it tends to be much less because if the train actually makes it to the station without needing the padding then it ends up waiting at the station because if it departs early the customers who miss it will complain loudly to the powers to be (this is more relevant on inbound runs; on outbound runs the railroads frequently disclose that some trains may leave before the scheduled time). Padding between the last two stations does not risk this type of wrath from the customers.

I would have thought that a more customer friendly approach would have been to call a train late if it arrived 6 minutes or more after its scheduled time at ANY station along its route, not only the last one. Doing so guarantees that the train is keeping pace with its schedule along the whole route and is much harder to game. This is especially relevant for lines which see demand for trips to multiple stations along a route, not just inbound in the morning and outbound in the evening. Metro-North's New Haven Line is an example of such a line where we should not just care about the arrival time at GCT, but also at least for the arrival time at Stamford (and some other stations with lots of jobs around the stations).

[Nasadowsk]

the 5 minute and 59 seconds is a nation wide standard set by APTA. main reason for it is accounting and timekeeping functions.
its a lot easier to go by 1/10 of hour segments, that if by doing it in 1/30 increments.

Pretty much every OS since Unix deals with time to at least millisecond increments, and 64 bit time is accurate to the second (or probably better) or so over the life of the universe, so it's even good for accurate accounting of Amtrak trains.

It's not 1955 anymore. Then again, APTA's probbly in a panic their AOL disks won't work on Windows 95 after the year 2000, so YMMV....

[MattW]

Why shouldn't we strive for perfection? The Japanese time their trains down to the second.

[DutchRailnut]

not really but nice rumor

[Absolute-Limited Advance-Approach]

Why shouldn't we strive for perfection? The Japanese time their trains down to the second.

When we have Japanese level networking, Japanese level budgeting, Japanese level passenger cooperation and Japanese level labour abuses we can dream of that. Outside of Tokyo and the high priority Shinkansen lines the Japanese rail system, Subway and main line functions closer to other nations (still better than many others, but not at the mythical pinpoint timing).

OTP will always be dubious as trains on the LIRR spend a substantially higher amount of time at restricted speed compared to other counties. A substantial number at Penn, Atlantic Ave is 5mph, so is LBH, etc. We'll need to address that for starters.

The problem with the new and old rules is that they both allow the railroad to pad the time between the last two stations in order to show on-time performance. For example, the way this works is on NJT is by padding the time between say Secaucus and NYP by 5-6 minutes during the morning rush hour.

As far as padding, that can only go so far before increasing the amount of equipment required to the point that it induces its own delay/financial limitation, I don't think there will be an exceptional increase, though there will be one.

I just pulled up the North East Corridor NJT Timetable, every train after 3806 (5am) for the entire day until 3292 (0058am) shows a steady 13-15 minute time between Secaucus and NYP, where do you show evidence of a pad?

Also keep in mind that during rush hour travel times will be longer as trains are run on speed control so if trains are in the block ahead, the MAS may be 90 but a train ahead will force you to 60 or 45, of course that needs to be reflected in the schedule as increased running time. Just like you can get down the Van Wyck in 15 minutes at 2am, that doesn't mean it's reasonable to expect the same result at 8am.

[Commuter X]

What is the official definition of "arrived" ? Lets use Penn Station as an example

Is it when the train passes an arbitrary marker in the East River tunnel?
Is it when the train pulls into the platform -- but is still moving?

The definition I use is that the train has stopped at a platform, and the doors open allowing the passengers to exit.

[Absolute-Limited Advance-Approach]

What is the official definition of "arrived" ? Lets use Penn Station as an example

Is it when the train passes an arbitrary marker in the East River tunnel?
Is it when the train pulls into the platform -- but is still moving?

The definition I use is that the train has stopped at a platform, and the doors open allowing the passengers to exit.

The timing system captures the arrival when the train occupies the track circuit at the station platform, should that not be captured for some reason (following in on a restricting etc) its the time you take the home signal at C/JO + 1 minute.

There is no way to capture when the doors open short of sending someone to station level.

[CTG]

I just pulled up the North East Corridor NJT Timetable, every train after 3806 (5am) for the entire day until 3292 (0058am) shows a steady 13-15 minute time between Secaucus and NYP, where do you show evidence of a pad?

The evidence of a pad is when you compare the times in the opposite direction. As you note, times from Secaucus to NYP are consistently 13 to 15 minutes. Yet times from NYP to Secaucus are either 9 or 10 minutes. The same thing happens at the other end of the line. Schedule time from Hamilton to Trenton is anywhere from 9 to 13 minutes. From Trenton to Hamilton, however, is nearly always 6 minutes. This type of time/distance warp happens across nearly every NJT schedule.

While NJT is notorious for schedule padding, the LIRR is not. Lindenhurst to Babylon is 6 minutes, Babylon to Lindenhurst is 5 minutes. Plandome to Port Washington is 6 minutes, Port Washington to Plandome is 5 minutes. So maybe a one minute pad in there.

In order for the LIRR to acheive the same "on-tme performance" under a two minute threshold as they had under the six minute threshold, they would simply add four minutes running time between the next-to-last and last stop on each train. No change in operations required. No change in "on-time" performance.

[EuroStar]

The evidence of a pad is when you compare the times in the opposite direction. As you note, times from Secaucus to NYP are consistently 13 to 15 minutes. Yet times from NYP to Secaucus are either 9 or 10 minutes. The same thing happens at the other end of the line. Schedule time from Hamilton to Trenton is anywhere from 9 to 13 minutes. From Trenton to Hamilton, however, is nearly always 6 minutes. This type of time/distance warp happens across nearly every NJT schedule.

Thank you! That is exactly what I had in mind.

[Erie-Lackawanna]

I just pulled up the North East Corridor NJT Timetable, every train after 3806 (5am) for the entire day until 3292 (0058am) shows a steady 13-15 minute time between Secaucus and NYP, where do you show evidence of a pad?

The evidence of a pad is when you compare the times in the opposite direction. As you note, times from Secaucus to NYP are consistently 13 to 15 minutes. Yet times from NYP to Secaucus are either 9 or 10 minutes. The same thing happens at the other end of the line. Schedule time from Hamilton to Trenton is anywhere from 9 to 13 minutes. From Trenton to Hamilton, however, is nearly always 6 minutes. This type of time/distance warp happens across nearly every NJT schedule.

While NJT is notorious for schedule padding, the LIRR is not. Lindenhurst to Babylon is 6 minutes, Babylon to Lindenhurst is 5 minutes. Plandome to Port Washington is 6 minutes, Port Washington to Plandome is 5 minutes. So maybe a one minute pad in there.

In order for the LIRR to acheive the same "on-tme performance" under a two minute threshold as they had under the six minute threshold, they would simply add four minutes running time between the next-to-last and last stop on each train. No change in operations required. No change in "on-time" performance.

The primary rationale for the specific examples you note is not schedule pad (although that’s a significant, but likely not a major part of it), but to reflect the operating realities of approaching a terminal complex in cab signal/ATC territory. When approaching Penn, cab signals and speed control gradually bring you down to restricted speed far in advance of where you might otherwise brake without speed control. This adds significantly to running time. Departing Penn, you don’t have an equivalent gradual acceleration; once you get a train length west of A’s interlocking, you wind it out and off you go.

The gradual deceleration creates congestion at Penn (and other major terminals), which slows the approach even more. For departing trains, the train ahead will clear quickly and you’ll get less restrictive signals much more quickly.

The LIRR and Metro-North, which operate under the same conditions, schedule the same way, although the physics of the terminals at Babylon and Harmon, for example, result in somewhat less additional running time for arriving trains than is evident at Penn, Brooklyn, and GCT.

Jim