Network News

X My Profile
View More Activity
Share Stories  |  Traffic  |  Columns  |  Q&A     |  Get Gridlock:    Twitter |    Facebook  |     RSS   |  phone Alerts

Metrorail's on-time decline

Metrorail's on-time performance dropped sharply after the trains went on manual control following the June 22 Red Line crash, according to a Metro staff report.

The staff review, developed for presentation Thursday to Metro board members who requested it, shows a decline of about 6 percentage points in on-time operations from June to July 2009. The trains, designed to be operated automatically, have remained under the control of their operators since then, and are likely to continue in manual mode for a year.

For the six months ending in December, which is as far as the report's statistics go, on time performance was off an average of 5 percentage points from the previous six months when they trains were under automatic control.

The statistics confirm a deterioration in service that was evident to many riders. In letters, e-mails and comments here on the blog, they attribute the deterioration to a cutback in the number and length of trains. Metro officials have said repeatedly that this has not occurred.

The report links the on-time decline to the end of automatic train control. It shows a very similar decline in performance occurred a decade ago when the trains went into an extended period of manual control, also as a safety measure.

The staff notes several reasons for the decline:
-- Train operators need additional time to manually stop at the stations and work the doors.
- Controllers in the operations center must closely monitor the system to ensure sufficient space between trains.

Board members asked for this report because they know from their experience and from the complaints they hear from riders that train arrivals are erratic and both the platforms and trains seem more crowded than ever, despite a decline in ridership that Metro links to the weak economy.

The report notes some contrasts with conditions a decade ago:
-- While weekday ridership has declined slightly since the recession, it is 35 percent higher than in 1999.
-- The rail fleet is 50 percent larger, and the trains are either six cars or eight cars, rather than four cars and six cars. (The report doesn't say this, but I'll suggest that while more passengers can be accommodated, the operational problems increase with a larger fleet.)
-- Metro imposed additional safety restrictions that included reducing speeds under certain conditions and stopping all trains at the fronts of the platforms. The latter move increases the time trains spend unloading and loading passengers.
-- The oldest cars were surrounded with newer cars on all trains. Trains made up of the same type of car travel 17 percent farther between breakdowns.
-- The trains weren't built to be driven. Under manual control, the distance between delays and failures decreases as the number of problems related to propulsion, brake, pneumatic and door systems increases.

To riders who believe the gaps between trains are widening or who have to wait for a train or two to pass by before them can cram aboard the next one, the main surprise here may be that the on-time statistics are so high. In the six months before the crash, the report says, the trains' on-time performance averaged 93.5 percent. In the six months afterward, the average was 88.8 percent.

One of the most interesting number was that 93.5 percent pre-crash performance under automatic train controls. In the six months before automatic control was suspended in March 1999, the on-time rate was 93.1 percent, suggesting that Metro had not significantly improved its on-time performance over the decade, even under normal conditions.

By Robert Thomson  |  March 9, 2010; 8:00 AM ET
Categories:  Metro  | Tags: Dr. Gridlock, Metrorail, Red Line crash  
Save & Share:  Send E-mail   Facebook   Twitter   Digg   Yahoo Buzz   StumbleUpon   Technorati   Google Buzz   Previous: Foggy Bottom escalators out
Next: Petition seeks more money for Metro


They ought to define what on-time means, in their context. To airlines, it's departure or arrival within 15 minutes. What's Metro's standard?

And 93.5%-88.8% is not equal to 5.0%

Posted by: Chris737 | March 9, 2010 9:28 AM | Report abuse

In the airline context, an on-time departure does not refer to what time the plane leaves the ground--it refers to when they push back from the gate. That's why you see some carriers (Northwest were known for this) pushing back, then having the aircraft sit for a long time out on the tarmac.

Thankfully, Metro doesn't really have anything comparable to that.....I think!

Posted by: 1995hoo | March 9, 2010 10:15 AM | Report abuse

"Trains made up of the same type of car travel 17 percent farther between breakdowns."

So, in addition to the jerky stops attributed to having the old cars sandwiched between newer ones, this configuration also makes trains more likely to break down?

After the crash, there was strong public support for shuffling the old cars to the middle of trains because of the myth of increased safety. But now that we have to deal with the inconvenience of being thrown around every time the train comes to a stop (because of the mismatched cars), AND the fact that these mismatched trains break down more frequently, doesn't that outweigh the marginal safety benefit of never placing the old cars at the front of trains? How long will it take to undo this reconfiguration?

I know the old cars aren't as crash-safe as the newer ones, but since we're stuck with them, maybe Metro should focus on preventing malfunctions like the one that caused the 2009 crash rather than slapping a band-aid on the problem by hiding the old cars between newer ones.

Posted by: VDouglass | March 9, 2010 10:20 AM | Report abuse

Umm Chris...that is 5 percentage points. Just saying

Posted by: EricS2 | March 9, 2010 11:00 AM | Report abuse

True, that is 5 percentange points by division. But the line right above that in the report is 93.1-88.4=4.7. So, one of them's wrong. Take your pick. Just saying.

(typos like this in a public and relatively significant presentation don't inspire much confidence - not that we've had an abundance of that lately)

Posted by: Chris737 | March 9, 2010 11:20 AM | Report abuse

No doubt "on-time" figures also don't apply to the stealth service reductions. Trains that are never dispatched cannot be late.

Posted by: member5 | March 9, 2010 11:34 AM | Report abuse

From Dr. Gridlock: VDouglass, yes, Metro says the train configuration with the old ones sandwiched in the middle makes the trains more likely to break down. The jerky motion we all experience is probably a combination of the mismatched cars and the manual control. As I say, the trains weren't meant to be driven.

You raise the very interesting and very basic question of whether we're better off in manual control or automatic control. Metro board members have been wondering whether that issue should be reviewed.

I think a return to automatic is unlikely in the next year. Dave Kubicek, the Metrorail boss, is not recommending it until Metro is sure that no train will disappear from the monitoring system.

I also think the old 1000 Series cars are probably going to stay in the middle of the trains till they're replaced by new cars a couple of years from now. Once Metro did that as a safety move, who's going to step in and say, We don't really need it?

Posted by: rtthomson1 | March 9, 2010 11:37 AM | Report abuse

After all the lies, I rarely believe anything metro says.

"Nothing to see here, move along...nothing to see here..."

Posted by: 123cartoon | March 9, 2010 12:55 PM | Report abuse

What is on-time performance? Does Metro have an operational definition? Is it when the train stops at the platform? That would seem reasonable. Yesterday, I travelled from Shaw to New Carrolton. The Yellow Line train was 2 minutes early; I was walking on to the platform as the train arrived, and the Orange Line train was 3 minutes late (all according to Metro's Trip Planner). So my question is what is the tolerance for on-time performance? Is it plus or minus 1, 2, or 5 minutes? In addition, a train that is early can be just as bad as a train that is late, especially at a transfer station or if arriving by bus.

While Metro runs in manual mode, it may make sense to look at the train schedules and make adjustments to account for all the new variables as a result of manual operations.

Posted by: mrdcmetro | March 9, 2010 12:56 PM | Report abuse

"In letters, e-mails and comments here on the blog, they attribute the deterioration to a cutback in the number and length of trains. Metro officials have said repeatedly that this has not occurred."
I seem to remember Metro saying it was going to increase the time between rush hour trains from an average of 5 to 7 minutes. That would consititute a cut back in the number of trains (more time means fewer trains). Also, just from riding practically every day, I rarely see any 8 car trains announced on the boards. Certainly the trains I usually get on are only 6 cars. Me thinks Metro isn't telling the whole truth in this regard.

At any rate, I don't mind paying more, but I want to start seeing a little bang for my buck. Improvements need to start happening, regardless of how the trains are operated.

Posted by: akchild | March 9, 2010 1:04 PM | Report abuse

Mixed consists lead to a rougher ride as a result of the trains needing to "interpret" the signals being sent from the trains of a different type. That short delay in communication can cause a rougher trip.

Posted by: FHMetro | March 9, 2010 1:10 PM | Report abuse

Part of the problem is the difference in definition between "service cuts" by Metro and the passengers. Metro measures service in terms of number of trains running on the line at rush hour. The public measures service in terms of "how often do the trains come".

If Metro dispatches the same number of trains as they always have (to them, no service cut), but they travel slower on their runs, the trains are going to come to each individual station less often. If, for example, it takes trains 1 hour to make a run, and there are 10 trains on the line, a train will pass any individual point on the line (station) once every 6 minutes. Increase that running time to 2 hours, and they will pass any point on the line every 12 minutes.

Sure, that example is a huge exaggeration...the time to run trains hasn't doubled. But even a little delay can have a huge impact if the delays cascade down the line, and when trains are running at rush hour levels to serve rush hour demand, every fraction of a minute means more people to cram onto each train.

So you wonder...if the same number of trains run slower and therefore they come less often during rush hour, where do those "missing trips" go? Simple...the trains are still running on their rush hour runs after rush hour is over. In otherwords, if an operator is sent out and told to make 3 round trips, and he goes slower, he will finish his runs later. That doesn't serve the public, because the public isn't spreading out their trips later...they are trying to cram onto the fewer trains that actually arrive during the rush hour.

Dr. G.: I have two requests for you.
1) Can you see if you can get some data from Metro about how late their trains are running? Its one thing to say X% of trains run late, but I think it would be useful to know if the average delay (average number of minutes late) has increased since the trains went to manual control. That is what really has a huge impact on "how often trains visit a station" during the rush hour.

2) Can you find out if Metro has updated its travel times in its station PID's? For example, if in the past, it took 5 minutes for a Red Line train to get from Union Station to Gallery Place...but now it takes 7 minutes, do the PID's still tell people at Gallery Place that a train is 5 minutes away when there is a train at Union Station? Or has Metro updated them to reflect that it now takes longer? One thing I've noticed from travelling on Metro is that the PID's seem to underestimate how long a train will take to arrive...sometimes severely. If the PID says the train is 5 minutes away, I think most people expect that it will count down the minutes in roughly 60 second intervals. In some downtown stations, I've counted up to 90 seconds before the PID would increment down to the next minute. Bring a stopwatch down to one of the downtown stations at rush hour and try it...

Posted by: thetan | March 9, 2010 1:25 PM | Report abuse

"Trains have what Metro spokeswoman Lisa Farbstein called a "grace period," which she said for peak-service trains can be two to four minutes late but still count as "on time." Delays at other times can be longer."

Posted by: Razor04 | March 9, 2010 1:45 PM | Report abuse

I'd like to point out one thing regarding the sandwiching of the 1000 series cars. It's pointless... Metro admitted after the recent NTSB hearing that it was nothing more than a PR move. On top of that we saw that moving them to the middle made no difference during the recent railyard crash. Those suckers still telescoped in the middle of the train! Even worse the middle cars tend to be the most crowded cars.

Why put them somewhere where they are no safer and just cause maintenance and operational issues? If metro could show that there was some advantage to it I'd support it but I'd rather have a smoother ride with fewer "mechanical difficulties" otherwise.

Posted by: Razor04 | March 9, 2010 1:51 PM | Report abuse understanding is that the PID's update based off of where a train currently is. That is why the numbers adjust up and down and why they tend to be wildly inaccurate when the trains are further apart.

Regardless though the things are broken. Just last night I watched a train to Shady Grove arrive into Metro Center even though the PIDS said it was still 3 minutes away.

Posted by: Razor04 | March 9, 2010 1:53 PM | Report abuse

Will the "PIDS" at the ends of the lines ever work?

Posted by: member5 | March 9, 2010 3:13 PM | Report abuse

From Dr. Gridlock: thetan, I'm particularly interested in your stopwatch idea. In your comment, you articulated something that me me was just a vague feeling -- that a Metro minute on a PID lasts a lot longer than 60 seconds. I'd apply that same theory of relativity to Next Bus arrival times.

If I could pull off the experiment -- it's really just a matter of time -- I think I'd find that the calculators behind the PIDS, Next Bus and Trip Planner all need to be reset. The trains and buses are not following those schedules.

Looking back at the comment from akchild, I think there's a reference there to the Metro proposal to increase the gaps between the Red Line trains. That proposal came before the Metro board in January, but was put off. The idea was a bit of a hybrid. It was presented partly as a money-saving measure, and partly as an acknowledgment of reality: Red Line trains are not making the trip between Shady Grove and Glenmont as quickly as they used to.

Posted by: rtthomson1 | March 9, 2010 3:40 PM | Report abuse

Dr. G and thetan, in reference to the timing of PID's I always thought with NextBus (well, and with Metro Rail too) that the time is an estimate based on where the bus or train is located. So if a bus is traveling on East-West Highway and it is 1/2 a mile away, it should take 3 minutes to reach your destination (based on the usual travel time/flow of traffic/etc.). But in reality there are many factors that could increase this time (traffic, an accident blocking lanes, getting stuck at a traffic light) or even decrease this time (bus caught all green lights, bus is speeding, bus didn't stop to pick up passengers). Same thing with trains. Sometimes trains stop in tunnels or at platforms and hold for awhile. No idea why they do this but it happens and would certainly result in the PID time not counting down. While I agree that the PID times probably need to be reset for the trains to account for the longer travel times due to manual control, I think it will still always be an estimate and therefore never completely accurate.

Posted by: UMDTerpsGirl | March 9, 2010 3:57 PM | Report abuse

From the article:

"The trains weren't built to be driven."

Doh! We forgot Driving!

Posted by: ghokee | March 9, 2010 4:17 PM | Report abuse

So does anyone know what's up with the new announcement about avoiding impending death? I know they're trying to get their point across but it is a little over the top and insulting considering all the people metro has killed over the past few years.

Posted by: Razor04 | March 9, 2010 4:44 PM | Report abuse

PID time is not updated continuously. I am guessing it updates at 1-minute to 2-minute intervals. I have seen a 5 minute PID time for a train arrival with the actual arrival at 7 minutes. So what happens, I believe, is the PID time sticks at 5 minutes as the PID recalcs the arrival time. I have seen the opposite as well: a 2 minute PID display with train approaching the platform and the edge lights flashing moments later. My guess is Metro probably has a pretty course update interval to keep the PID time from jumping up and down too often.

Posted by: mrdcmetro | March 9, 2010 5:02 PM | Report abuse

When I read the comments posted here I see a lot of misinformation being presented as fact:
The trains are not as on time as before the Red line accident: since the accident the trains have been operated manually. This requires coordination between the train operator and the operations control center. When in full automation mode computers make the decisions and are much faster than people.

The jerking of the trains during start/stops are caused by the mismatch of cars: there have always been a 'mismatch' of cars. Metro rail cars have to be able to fuction with every other type of rail car. The jerking is caused by the manual operation of the train.

A PIDS minute is longer than 60 secconds: the information on the PIDS system is performed in real time with estimates of times between stations. If a train stays at a station shorter or longer than 'normal' then the new arrival time is calculated and displayed on the PIDS system. If a train is stopped in a tunnel the PIDS system isn't updated until the train starts moving again.

Posted by: Jimof1913 | March 9, 2010 6:13 PM | Report abuse

As others have said, I too assume that the PIDS calculate the time based on the location of the train. So if I'm standing on the platform at Judiciary Square, and there is a train at Union Station, and
"traditionally" it takes 2 minutes to get from Union Station to Judiciary Square, then the PIDS will say "2 minutes".

But if, due to slower operation of the trains, it now takes trains 3 minutes to travel from Union Station to Judiciary Square...the PIDS will still say the train is "2 minutes" away unless Metro goes in and recalculates the actual travel time between Union Station and Judiciary Square.

Has Metro acknowledged that trains are operating slower and re-calculated the time/distance that goes into their PIDS? Or have they not? If they have, then they are saying "We estimate the next train is 2 minutes away", which is useful information. If they have not, then they are saying, "Based on where the train is, it used to take 2 minutes to get to you, but it's going to take longer now"...which is useless information to me as a customer.

If they can't get the PIDS to indicate the actual time in minutes, then perhaps they shouldn't display the train arrival times in minutes, and instead say something like "6, Glenmont, at Union Station".

Posted by: thetan | March 9, 2010 6:31 PM | Report abuse

Dr. G and thetan, in reference to the timing of PID's I always thought with NextBus (well, and with Metro Rail too) that the time is an estimate based on where the bus or train is located. So if a bus is traveling on East-West Highway and it is 1/2 a mile away, it should take 3 minutes to reach your destination (based on the usual travel time/flow of traffic/etc.).
UMDTerpsGirl hit the nail right on the head with the, "based on the usual travel time/flow of traffic/etc." If the "usual" is slower than it was a year ago, than the PIDS need to be updated to reflect that. If a speed restriction is put in place, the PIDS should reflect that. There is nothing more annoying to a customer than to see a train is X minutes away, and it takes X+5 minutes to get there.

Posted by: thetan | March 9, 2010 6:43 PM | Report abuse

thetan, I agree the PID's for the trains probably need to be updated to reflect manual control of trains BUT in the long run won't you still have the same issue of trains potentially being held up along their route for whatever reason and thus the PID's still won't be accurate? I think you will still run into the same problems you describe since it's still always going to be an estimate. Hopefully by updating the PIDs there would be more accuracy but I think it's important for people to realize these are estimates and not actuals.

Personally I don't think the time difference between the previous and the current PID's is that big of a difference. Atleast I wouldn't want WMATA to invest any money in this right now as they seem so cash strapped. Now if they could fix this for free...then DO IT! Instant value add. :-)

Posted by: UMDTerpsGirl | March 9, 2010 7:24 PM | Report abuse

@Razor04: it's not "to avoid death," it's, "to avoid theft." It's just not well-enunciated, and is even more poorly broadcast in various acoustic hellholes.

Posted by: EtoilePB | March 9, 2010 7:46 PM | Report abuse

Of course its never going to be able to account for when a train is held up unexpectedly. However if a train is held up regularly (Gallery Place comes to mind), the system should be updated to reflect that. I don't think it would cost hardly anything at all...literally just updating numbers in a table. I'm assuming they can obtain the numbers using archived train location data from Central control.

Posted by: thetan | March 9, 2010 8:33 PM | Report abuse

According to Mr. Thompson the WMATA scheduling department employees went to the board last month and asked for permission to change the scheduled travel times on the red line and maybe other lines too, to bring them into line with actually achievable schedules. The board said no.

So you have to appeal to the board is my guess. It seems unlikely that it would be allowed to let a WMATA scheduling manager unilaterally change scheduling on the red line without getting permission from Montgomery County and DC at the very least.

What's weird is us riders are left with the mythical schedule and what they actually run. The schedule says there's supposed to be a train every 2 or 3 minutes on the red line and in real life it's usually closer to 4 or 5 minutes so I'm guessing the runtime is off by nearly a factor of two. It can't be good for anything to have that level of mythology, it'd be like putting 50 students in classrooms built for 25.

Posted by: annapolisjack | March 10, 2010 6:41 AM | Report abuse

Oh look, a blog post stating facts the riding public already knew.

Posted by: anarcho-liberal-tarian | March 10, 2010 8:46 AM | Report abuse

From Dr. Gridlock: annapolisjack, that Red Line rescheduling proposal came before the Metro board in January. The plan to bring the schedule in line with reality was worth discussing.

One problem was that the proposal was wrapped up in the other proposals to close the $40 million gap in the current budget, and the board decided instead to hold a hearing and seek public comment on all money saving ideas.

Another problem was that the Red Line plan was pretty dramatic. Among other things, it would have ended the train turnbacks at Grosvenor and sent all trains to the end of the line at Shady Grove. The board never makes any big change on the spur of the moment. The plan hasn't come back to the board since January.

If you'd like to review more details of the Red Line plan, check this blog posting:

Posted by: Robert Thomson | March 10, 2010 8:47 AM | Report abuse

The comments to this entry are closed.

RSS Feed
Subscribe to The Post

© 2010 The Washington Post Company