As subway service has gone from good to bad to worse over the past few years, signal timers have come up frequently as a topic of conversation. The timers, as I explored in March, arose out of a 1995 subway crash that was ultimately determined to be due to human error, and the MTA has spent over two decades slowing down the trains because, well, who knows. No one could explain why by the time year 23 of the signal timers arrived that many were being installed. Subway service has been slowed because the MTA opted to slow the trains, and now Andy Byford is starting to do something about it.
Emma Fitzsimmons of The Times broke the news in an article on Monday in which you will find a few comments by me. The essence is this: By attacking faulty signal timers and raising speed limits on others, Byford believes he can alleviate the system-wide slowdown caused by the MTA’s own internal decision-making processes without sacrificing safety. As anyone who has experienced a slow crawl of, say, a 2 or 3 train north of 86th St. or an R train ride up or down 4th Ave. in Brooklyn will attest to, eliminating these slow spots is a move that can’t happen soon enough.
The MTA announced the changes in a press release on Monday afternoon as part of the awkwardly-named “Save Safe Seconds” campaign:
This past weekend, several months of careful testing and study have led to the safe increasing of five speed limits between 36 St and 59 St on the nr line in Brooklyn, with 15 mile-per-hour zones being increased to 20 or 30 miles per hour. Twenty-nine more increases throughout the system have also been approved by a safety committee and will be rolled out in coming weeks, with Transit officials estimating speed limits to be safely increased at more than 100 locations throughout the system by the springtime. The speed limit changes already approved increase speeds generally in the 10 to 20 mile per hour range to speeds that reach the 40s.
The same team doing this work is also testing and fixing speed regulating signals called “time signals” or “timer signals,” with 95 percent of some 2,000 such signals tested since the initiative began in late August. Approximately 267 faulty timer signals have been discovered and approximately 30 of them have been fixed so far in what amounts to very labor-intensive work to inspect, diagnose and repair or replace numerous possible pieces of equipment during times of exclusive track access for workers such as weekends or nights.
“Safety is always our top priority, and we’re working hard to maximize our subway’s potential within the boundaries of stringent safety standards,” said NYC Transit President Andy Byford. “Subway cars have come a long way in safety and performance since the system’s speed limits were first put in place up to a century ago, and some speed-regulating signals have become miscalibrated over time, forcing trains to go slower than they need to. We’re taking a fresh look, with no compromise to safety, at how to reduce delays and get people to their destinations sooner.”
This announcement though seems modest in comparison with the findings in Fitzsimmons’ article. According to her reporting, the MTA has installed around 2000 signal timers throughout the system — or the equivalent of around 3 for every mile of track. Even though we can point to multiple causes for the declining reliability of subway service, it’s hard to understate just how costly these signal timers have been as trains have slowed to a crawl lately. The initial effort to remove them is a modest too. Here’s Fitzsimmons:
Over the summer, Mr. Byford created a new “speed unit” — a three-person team that traveled every mile of track on the system in an empty train to find areas where trains could safely move faster. The team identified 130 locations where the speed limit should be increased. So far, a safety committee at the transit agency has approved 34 locations for speed increases…
About 30 signals have been repaired in Brooklyn, from the DeKalb Avenue station to the 36th Street station, on the B, Q, D, N and R lines, and near the 9th Avenue station on the D line. Mr. Byford wants to eventually fix all of the faulty signals, though he cautioned that the work is complex and could take awhile.
I’m guardedly optimistic that this move is the start to a solution for our speed woes. As Fitzsimmons notes, NYC subways are among the slowest in the world, and as I keep saying, that’s largely in part due to the MTA’s own choices that slowed down speeds to an unacceptable level in response to discrete incidents caused more by human error than faulty signals. Unfortunately, 30-40 timer fixes every few months won’t do much to fix speeds, and any observant rider can reel off a handful of spots where timers have become more noticeable in recent years (the Franklin-Atlantic run on the 4/5, Grand Army Plaza to Bergen St. on the 2/3, the Q/N heading north into Union Square, the 6 between 51st St. and Grand Central, etc., etc., etc.). So for this to pay real dividends, Byford will have to push for a faster pace, and a faster pace is what frustrated NYC subway riders deserve. For this one, the MTA has only itself to blame, and only the MTA can fix it.
31 comments
Where can we actually find the report? I looked at the MTA website and it’s not obvious (to me) which committee this was reported too yesterday (Monday) . The NYT article (which I read before your article) doesn’t provide a link either …
Also it’s good that a union rep has been to observe the work of the speed team and aren’t just blocking it.
(as a side note I really wish they didn’t publish the committee papers as a single document rather than each report individually)
The time savings pale in comparison to merging conflicts that are built into the schedules.
Let me add the schedule’s phantom dwell times that are designed to improve the terminal on time performance metric.
What do you mean by “phantom dwell times”? *curious*
According to current GTFS schedule, 8444 trips will be scheduled tomorrow. Of these 789 or 9.3% of the trips will have one or more stops with a scheduled dwell time of 5 minutes or more each at one or more stops. Most of these are the stops just before the terminal. Terminal OTP is percent of trains that arrive more than 5 minutes late, according to the schedule. Pad an extra 5 minutes just before the terminal and – OTP improves.
Also, tomorrow, 1627 or 19.3% of the trips will have an accumulated scheduled dwell time of 5 or more minutes during the trip.
On looking at the GTFS-RT feed, the dwell times exist for the most part. Trains are more than 5 minutes late arriving at the 5 minute dwell time stops. The 5 minute dwell time translates into getting many of these trips on time by suddenly increasing their scheduled run time.
Fascinating. Their own publicly released reports show abysmal on-time performance rates, even with their padding and cheating taken into account. The numbers are even worse than we thought.
Which requires another question: What is “accumulated scheduled dwell time”? It sounds like it’s the total of all scheduled dwell time for a given train, but that would greatly exceed 5 minutes on almost any line in New York.
Here is the analogous concept in commuter rail. The accepted standard in North America is that a train is considered late if it is more than 5 minutes and 59 seconds late at the last stop. So what do you think they do? They schedule the time between the next-to-the last and the last stop to be the required normal travel time plus substantial padding, say 10 minutes (sometimes the padding is not concentrated between the last two stops, but spread out between the last few stops; frequently this is accomplished by notes in the schedule stating that the train may depart the last few stops before the published departure time). On a good day when everything runs on time and the train arrives at the last stop early, nobody complains. If the train is late within a reason the padding allows them to claim that it was actually on time because it still arrived at the last stop before the published arrival plus 6 minutes. A train will show as late only if it is grossly late at the last stop.
Basically the definition that they use for late is bad and allows for gaming. A better definition is: A train is late if it is more than 6 minutes late at any stop including the departure stop regardless of whether it makes up time en-route (this used to be common in the old days, now with PTC and all the electronics it is impossible for the engineers to push the limits and make up any time). That way if there is delays anywhere en route they are counted as late. Of course, the probability that any commuter railroad or the subway switching to this definition is zero because it will show how bad the service really is.
For most people, knowing the train can consistently make it to the terminal in 35 minutes is much more useful than knowing it could do it in 29 in the dead of night when there are no passengers or other trains gumming up the works.
I think most people would prefer trains operating reliably with a 29 rather than a 35 minute trip duration. The extending schedules ploy makes it difficult to discover that problems exist. Ignorance isn’t bliss, when it comes to solving problems.
The presence of other trains gumming up the works is a clue that the schedules have built in merging conflicts. There are plenty of those, even after schedules are extended.
Take the other trains away the people using them would be a bit miffed. It’s just awful the way these passenger trains get screwed up by letting people use them.
Subway vs. commuter rail service levels make this a bad choice for gaming the system. One example is the Jamaica Center bound E train having a scheduled 5 minute dwell time at Sutphin, when the headway is 4 minutes.
Yes, there are few things worse than having to travel to a terminal and always having to sit at the previous stop for 5 minutes because both terminal tracks are occupied.
…..so is this new higher speeds or a return to prior limits
if the latter, isn’t great that our great achievement is with 21st century technology that we are able to what we did in the past?
Has a system every been built like a continuous loop with stations evenly spaced allowing trains to run much like a marching band?
In my mind, instead of expanding the #7 to the hudson yards, I would build a one directional continuous monorail with with 4 stops in the loop. The loop would have been from/at times square or the bus terminal to penn station to the convention center to the intrepid and back to the bus terminal. Such a loop could have run in an automated mode and without fare collection as most people would have been transferring from other modes of transportation including Hudson River ferries. (Obviously the furthest that anyone would ride would be 3 stops and in all likelihood most would ride for 1 or 2.) A continuous platform with access to the street at every block would have also allowed for a highline type experience.
And it would ultimately be a more significant waste of money, especially since it has no reach of its own.
Sounds like Miami’s Metromover.
Ben, I’m pretty sure many if not all of the locations listed here don’t actually have timer signals. These are just locations where NYCT is raising the speed limit. For example, the southbound 2/3 between Bergen St and Grand Army Plaza is all uphill and has no timers currently.
Yes. I should clarify: The speed limit adjustments are first. Tackling timers is second.
Aha–I guess that explains why none of my most-hated timers are on the list.
I hope in the next round they get rid of all those red signals on the northbound Q between 42nd and 57th. 57th is no longer a terminal, and there is no regular switching between tracks here anymore, so there is no need for all the red signals. I also despise the mysterious slowdown of northbound 2/3 trains passing Christopher St.
Trains are indeed capable of going through there with notable speed, but if someone has to slow down so much, they’re probably not operating properly.
For what it’s worth, I’m no fan of it, but that’s considerably lower on my list. After all, Franklin Avenue forces local trains to crawl in for no good reason (and this is a bidirectional issue).
The 4 and 5 come into Franklin ave southbound the fastest of any station in Brooklyn. Maybe Boro Hall beats it.
On that line southbound I mean.
I’m well aware of that, but that wasn’t my focus.
Yes, that’s one of the spots on my most-hated list. I just rode through there yesterday–it’s perfectly straight track, and the operators are constantly braking and accelerating. It’s literally nauseating.
I hope they address the SB G train through the tunnel under Newtown Creek. That train just crawls as it enters Greenpoint Avenue Station, seems unnecessary.
surprised not to see something about the manhattan bridge lines heading into brooklyn. thats some of the perplexingly slowest moving traffic in the system
Something about the structure of the bridge that causes trains on it to list and roll back and forth like a ship at sea just before it rolls completely over.
Even at the current low speed it’s slightly unnerving. No idea if speeding up would exacerbate the rolling motion, but I wouldn’t want to be riding the train on a day when it rolled all the way over onto its side.
I once saw a film clip on TV of the Manhattan Bridge at a close up angle and in fast-motion. Seeing how it twists as trains pass over it is very, very unnerving! Unfortunately the bridge has an inherent design flaw where the tracks are located on the outside edges of the deck vs. being placed in the center. This why the bridge went through a decade-long rehab in the 80’s/90’s.
The RPA released a study showing how much heavier than normal NYC subway trains are. They focused on the energy wasted moving the excess weight (as well as the added heat output), but added wear and tear on infrastructure is certainly another effect of it.
http://library.rpa.org/pdf/RPA.....ubways.pdf
I wish this addressed where the R train really crawls: between Atlantic Avenue and DeKalb Avenue, especially northbound.