For a long time, New Yorkers have been waiting for the debut of communications-based train control. Long promised for the L line, the automated control of trains should speed up travel and subway efficiency while increasing the number of trains the MTA can run per hour on one of the system’s most popular lines.
While the project has faced numerous obstacles — including the objections of union leaders who don’t want to see an automated system replace their union members — the MTA had been running live overnight tests of the system along the L line since February. Recently, the agency had increased the test runs to a 12-hour period starting at 7 p.m., but all has not been wine and roses though for Transit’s test case.
As Heather Haddon reports today, a few technical glitches have led trains to miss station platforms by a few feet. When a train does that, Transit regulations require that it move on to the next station, and some straphangers have missed their stops. She reports:
Running the L line on autopilot at night is causing trains to shoot past platforms, forcing straphangers to miss their stops, motormen and union officials said.
Because of the software fluke, drivers have to travel to the next station to let passengers off, according to the officials.
One Brooklyn mailroom worker, who didn’t want to be identified, said he was late for work repeatedly for several weeks after the L train missed his stop in Bushwick.
“It’s not perfected yet. It’s not working. And it’s definitely not cost-effective,” Keith Harrington, union vice chairman for train operators, said of the $326 million system.
This isn’t the first time technical problems have popped up in regards to the CBTC program. Last month, I wrote about jerking motions and breaking problems aboard the CBTC trains. What is interesting this month, however, is Haddon’s sourcing.
In this article, the complaints about the CBTC program come from “motormen and union officials.” These are the same people who stand to lose their jobs if and when CBTC is deemed a success. As the MTA plans to start running it in Queens soon as well, train drivers may have a reason to fear for their employment future.
For Transit’s part, spokesman Charles Seaton said the problem, according to Haddon, “does not impair passenger safety” and will be solved soon. It should be. CBTC, after all, is a technology whose time has come.
16 comments
With all of the hundreds of thousands spent on this technology on the L line, it should be made clear to the public that it will only allow the TA to run ONE more train during the rush than before. It is always being talked about including by Ben in this article as adding more trains which isn’t defenitely the case. And this project is millions over budget too.
People keep harping on how CBTC isn’t going to significantly increase the capacity of the L line. That may be true. But it wasn’t really installed on this line to beef up capacity. It was installed on the L first because it’s the one line, other than the shuttles, that doesn’t share tracks with other routes. It made a lot more sense to work out the the kinks in the system on an isolated line instead of installing it first on, say, the East Side IRT.
(If you want to increase capacity on the L line, lengthen the platforms to 10 cars already.)
Slightly related: In DC they are apparently having a problem stopping their trains in the right location. Seems to be more human error though:
http://www.washingtonpost.com/.....02099.html
I saw that yesterday on Greater Greater Washington and meant to include it in this post. Things could be worse; we could have conductors opening the doors way too early. I enjoyed that WMATA story.
CBTC–Communication-Based Train Control–is a system that affects the timing and spacing of trains. ATO–Automated Train Operation–is another system undergoing trials on the L train. ATO seems to be the system that’s screwing up.
Kid Twist: CBTC and ATO work hand-and-hand (in fact CBTC is a considered a type of ATO). What CBTC is a positioning system for the train that tells the computer where the train is and how far it is from other trains, which causes the computer train operator (ATO) to speed up or slow down. If the train is stopping at the wrong location, I’d say it could be a problem with either systems.
I believe the projection was that CBTC would improve capacity by 18%, not just one more train during rush hour. Lengthening the platforms is an utter non-starter. In addition to the environmental issues, it makes CBTC look cheap.
Stopping trains at the right spot is actually not that easy for an automated system! Usually, they use a somewhat conservative approach where the train coasts the last few feet to the stopping point before applying full brake, which loses a few seconds over manual stopping and makes for a slightly less comfortable ride for passengers. My idle speculation is that NYCT has been trying a less conservative stopping algorithm in hopes of winning back a bit more capacity, but it still needs tweaking. Incidentally, I don’t think CBTC is really needed for capacity given that the East Side IRT operated 35 trains per hour in the peaks back in the 1950s, and those trains were faster too. The Moscow Metro, one of the world’s busiest rapid transit systems, has 39 tph on the busiest lines, and uses a conventional ATC system based on coded track circuits with color light signals and train stops as a backup, and all trains are driven manually. Top speed is 50 mph, (and is generally reached between most stations) and average speed is around 25 mph.
I’d recommend that NYCT install a similar system as it will provide both the speed and capacity benefits of CBTC at a much lower cost and a much lower long-term technology risk (SF Muni’s CBTC system runs on OS/2, and I think is unsupported by the vendor by now). I wonder, is NYCT planning any more CBTC installations anytime soon?
There’s an inverse relationship between train length and the maximum number of trains per hour. NYCT used to run shorter trains, of only 6 or 8 cars rather than 10; that’s how it ran 35 tph. Moscow Metro runs 40 tph in part because its trains are shorter as well, at 6-8 BMT-sized cars.
How does the train length affect line capacity? In terms of time for the rear of the train to clear the station, that’s only a couple seconds’ difference, though I suspect dwell time also increases on a longer train. By the way, Moscow Metro trains on the busies lines are 8-car trains of 63 foot cars, for a total of 504 feet, almost exactly the same as a 10 car IRT train, and longer than an 8-car BMT train. I suspect that the ability to run 35 tph had more to do with pneumatic doors that slammed hard and fast so nobody dared hold them (as is still the case in Moscow), more liberal signalling (fewer grade timers, more and more effective station timers), and better trained operators who knew how to take full advantage of the features of the signal system, including station timers and possibly automatic key-by.
Does Moscow Metro run 35 tph with 8-car trains? My impression was that the busiest line was Koltsevaya, which runs 6-car trains.
At any rate, I think it’s just an issue of clearing the station, rather than dwells. Minimum headway equals safe stopping distance plus length of train. At subway speeds, the stopping distance is often smaller than a full train length, and rarely much larger. For instance, IRT locals seem to start braking only about a quarter or a third of the way into a station. Therefore an increase in train length translates into markedly longer headways (though total capacity still increases).
Nope, the busiest lines are 2, 6, 7, and 9, all of which have 39 tph scheduled headways and 8 car trains. The circle line (5) doesn’t have quite as much service, I think it’s something like 28 tph and six car trains, and because of the peculiarities of the station layouts, the end cars are often fairly uncrowded. On the other lines, they will occasionally try to increase scheduled headways to 40 tph, but it never seems to work out, as there’s not enough room for recovery from minor disruptions.
The principle of CBTC is simply that you can run more trains on a variable block than on a fixed block. If the Moscow Metro installed CBTC, it could run more trains too.
Whether in fact the MTA could run 39tph with current technology is a dubious claim. Anonymous message board posters may say it can, but there is no evidence they’ve ever run a railroad. A few years ago, there was a quack on subtalk who made similar claims, but when he finally posted his “math,” it was full of basic errors.
There are a few components that make up minimum headway. CBTC can only improve one of them: the time between when one train starts and the following train leaves, which incidentally can also be improved by having better accelerating trains. But with a good fixed-block system with time signals or with a conventional ATC system you can get very close to the performance of CBTC, and the real limitations are elsewhere, particularly dwell time, which CBTC can do nothing about. If you want to run more trains, you need to cut dwell times, which requires much less capital expenditure and much more training of conductors and modifying train doors to close faster. By the way, I don’t think MTA can run 39 tph with its current system, nor even with its current technology, but it can and does happen with fixed block ATC.
[…] this week, amNew York’s Heather Haddon reported on some problems plaguing the CBTC tests on the L train. According to union workers who stand to lose their jobs if CBTC is deemed a success, now and then […]
[…] The French city’s experience demonstrates that a conversion process doesn’t have to be intrusive. New York’s attempt at automating its L Train, a process that began in 1997, has been plagued by repeated delays, and the system still doesn’t function properly. […]