Greater Greater Washington

Transit


I tracked every Metro trip I made for two years, and here's what I found

When you keep track, it's funny what patterns appear in Metro trips. I've been doing it for 2 years. During that time, I have ridden 75% of the WMATA fleet, and been delayed about 2% of the time, but more so far in 2014.


Photo by the author.

In February 2012, I decided to start keeping track of a few attributes of my trips on Metro. The primary motivation was to track the cars I'd ridden on, but I also log delays, hotcars, and other information about every trip. It's important to point out, though, that this is anecdotal information. It's not a statistical sample, but rather just my experience.

My commute regularly takes me on the Green and Red lines. On a normal day, I ride 4 trains, two in the morning and two in the evening. Of course, I also make non-commute trips, to go downtown for dinner or out on the weekends. However, I don't actually keep track of "trips," per se. I log rides. So my normal commute involves 2 trips, which I log as 4 rides.

On average, I ride Metro 18.25 days each month. December (15.5 days/month) is always the lowest, since I spend time in Georgia during the holidays. August is the highest, at 21.5 days per month. On average, I ride 3.99 trains each day.

Over the past 24 months, I've ridden a Metro train 1,758 times.

Delays

I log any delay in excess of 3 minutes. On average, generally less than 2% of my rides were delayed.

In terms of counting delays, if I'm aboard a train that stops mid-journey, the clock starts immediately, but I only log the delay if it exceeds 3 minutes. When waiting for a train, I start the clock as soon as the scheduled headway has elapsed. For example, during rush hours, the Green Line is supposed to come every 6 minutes, so I start counting delay after 9 minutes waiting.

In 2012 (March through December), I took 758 rides and experienced 15 delays, which means that 1.98% of my rides were delayed. Since most of my trips comprise 2 rides, that's roughly equivalent to having 3.96% of my trips delayed, though that's not an exact number, since I don't record "trips."

In 2013, I took 866 rides and experienced 15 delays, which means that 1.73% of my rides were delayed. In 2014, so far, I've taken 134 rides and experienced 11 delays, which means that 8.21% of my rides were delayed. That's a significant increase.

Most of the delays I encounter are relatively minor. 75.6% of the delays I've experienced since February 27, 2012 are less than 12 minutes. Delays of 12 to 19 minutes make up 12.2% of my delays. Only 12.2% are 20 minutes or longer.

Coverage

Since I started recording car numbers, my commute patterns have not changed. I ride between Greenbelt and Silver Spring, changing from the Green Line to the Red Line at Fort Totten.

In overall numbers, 48.9% of my rides were on the Red Line. 43.0% were on the Green, and 7.6% were on the Yellow. I have rarely ridden the Orange or Blue lines. Those numbers don't move much between years.

However, we can see a difference if we divide the data set into before and after June 18, 2012, the date that Rush Plus started. With Rush Plus, three Yellow Line trains per hour continued north from Mount Vernon Square to Greenbelt during rush hour, in addition to the existing 12 Green Line trains per hour. As a result, my commute used to be almost exclusively on the Green and Red lines, and now there's a better chance of getting a Yellow Line train.

Before Rush Plus, my rides were almost evenly split between the Red and Green lines, with 48.7% of rides on the Red and 49.5% on the Green. The Yellow was at a paltry 1.5%. After Rush Plus started, the numbers have changed a bit. The Red Line still makes up about the same amount at 48.9%. But the Green has dropped to 41.7% and the Yellow has risen to 8.8%.

Cars
Since i started logging car numbers, I've ridden 74.1% of the WMATA rail fleet. I've ridden 91.3% of the 6000 series cars, 82% of the 4000 series, 78.7% of the 3000 series, 74.5% of the 5000 series, 69.7% of the 2000 series, and 56.1% of the 1000 series.

Of the 818 unique cars I've ridden, I've ridden 323 only once. The remaining 495 I've ridden more than once. I've ridden one car, #6058, 9 times. Two more, #4005 and #4086, I've ridden 8 times.

One question I've been asked several times is whether the cars move between lines very often. Surprisingly, they seem to. Of the 495 cars I've ridden more than once, 60.4% have been on different lines.

However, since the Yellow and Green lines share a rail yard at Greenbelt, it wouldn't be surprising to see those cars on Green one day and Yellow the next (in fact, sometimes a Green Line train from Branch Avenue becomes a Yellow Line train to Huntington when it leaves Greenbelt). So, I looked at the numbers counting the Green and Yellow as one line. Even counting them the same, I've ridden 54.4% of cars on more than one line. I've ridden three cars on 3 lines, the Red, the Green/Yellow, and one other line.

I'll continue to keep track of my Metro trips. I've found that having the data available makes it easier to note trends. For example, so far in 2014, I've found myself much more frustrated with Metro. Since I actually record my delays, I can go back and look. That's how I can say for certain that my delay rate has quadrupled.

But it's also really interesting to know that I've ridden on just under 75% of the cars Metro owns. Since the 1000 series is going to be retired starting in the next few years, it will be interesting to see whether I'll manage to ride them all before they disappear.

Matt Johnson has lived in the Washington area since 2007. He has a Master's in Planning from the University of Maryland and a BS in Public Policy from Georgia Tech. He lives in Greenbelt. Hes a member of the American Institute of Certified Planners. He is a contract employee of the Montgomery County Planning Department. His views are his own and do not represent the opinion of his employer. 

Comments

Add a comment »

This is cool and interesting.

by nativedc on Mar 3, 2014 1:42 pm • linkreport

All-around fascinating.

What I find particularly interesting is the cars moving between lines. The red line exchanging cars with other lines is especially interesting, because if I recall correctly, the only way to do that is to use one of two non-revenue single-track connections: red and yellow/green near Fort Totten and red and blue/orange near Metro Center.

by Tim on Mar 3, 2014 1:53 pm • linkreport

Having a base rate on delay that isn't based on Metro data is very handy. It's less often than most people would predict but despite being only non-random a sampling of all Metro trips, it's probably a good estimate and a lot less than many people would guess.

by Rich on Mar 3, 2014 2:05 pm • linkreport

The larger slice for delays on the 2014 pie chart seems a little bit deceptive, next to the other two, considering it's only measuring two months. However, it is a bit fascinating to be a bigger chunk compared to larger measurements from the other two year. Makes me wonder how much the amount and length of delays varies by time of year.

by Ben Schumin on Mar 3, 2014 2:29 pm • linkreport

Very cool.

I'm a little surprised Metro doesn't try to restrict certain cars or car types to certain lines like NYC. I know they're "bellying" the 1000 series, but there are more than enough of the other series to make 6000 or 4000-only trains.

I guess the only question I have is where the crossover is between the Red Line and the other lines. How can a car on the Orange Line get to a Red Line yard?

by David Edmondson on Mar 3, 2014 2:33 pm • linkreport

Very cool! I'm impressed with your data collection - I can't imagine having the wherewithal to track everything in a crowded train.

I wonder if you've ridden less of the 1000 series cars because your commute/habits make it such that you're more likely to board the first two or last two cars of a train.

by DC Transit Nerd on Mar 3, 2014 2:35 pm • linkreport

@ David Edmondson

There are two track connections between the Red Line and the rest of the system. There is a non-revenue track connection between the Red and Green/Yellow south and east of Fort Totten (which the late-90s Green Line Commuter Shortcut used before the Mid-City Line was completed in 1999). There is also a non-revenue connection between Red and Orange/Blue connecting near Farragut Square (going outbound from the Red Line south of Farragut North to outbound on the Blue/Orange east of Farragut West). You can sometimes see "money trains" use the track during late nights.

by Reza on Mar 3, 2014 2:54 pm • linkreport

@DC Transit Nerd:
That's an excellent point, and one I probably should have made in the post.

For non-commute trips, all bets are off, but for commute trips, you are correct tbat I do "prewalk" to a certain position to be in ideal position for transferring/exiting.

My morning trip on the Green Line, I ride in the last car (6th or 8th), which is never a 1000 series car. After transferring to the Red Line, I ride in the 2nd car, which is also never a 1000 series car.

In the afternoon, I ride in the 6th car of a Red Line train, which is sometimes a 1000 series car if the train is 8 cars long. Then I change to the Green Line, where I again ride in the 6th car, which is sometimes a 1000 series car if the train is 8 cars long.

by Matt' Johnson on Mar 3, 2014 3:28 pm • linkreport

What is the shortest contiguous route (measured in stations passed with minimal duplicates) to pass thru the entire system? How long would I be spending on the Metro on a weekday to do that?

by Bill Smith on Mar 3, 2014 3:36 pm • linkreport

@Ben Schumin:
It's not deceptive. The 2012 and 2013 (and 2014) charts are all delays as percentages of trips.

In 2012 and 2013 each, I experienced 15 delays on Metro. The whole year.

So far, despite taking about the same number of trips per month as in 2012 and 2013, I've experienced 11 delays.

In 2012 and 2013, the worst month (both years) was April, with 4 delays each. This year, I had 4 delays in January and 7 delays in February.

So, in order for things to even out, I'll have to go the next 10 months and only experience 4 more delays. If not, the 2014 rate will be higher than the 2012 and 2013 rates.

by Matt' Johnson on Mar 3, 2014 3:38 pm • linkreport

Cool post, Matt.

So, not quite 1 out of 20 rides have a delay of more than 3 minutes. For your average commuter who rides 10x a week on a simple round trip, that would mean they encounter a delay once every two weeks.

I guess over my 2.5 year experience with metro, that sounds roughly accurate. I rode the NOVA portion of the orange line into dc for 1.5 years, and the red line from takoma downtown for 1 year.

Every once in a while, you'd hit a week where it seemed like every time you got on there was a delay. But I also have memories of many perfectly timed, fast trips.

I wonder if your low % of delays, Matt, is slightly biased by the fact that you ride the red line in the non-peak direction?

by Nick on Mar 3, 2014 3:40 pm • linkreport

I wonder how the stats would look for someone who takes metro into the heart of downtown DC. I don't take metro often enough to track this. Anyone want to volunteer??

by Gerald F on Mar 3, 2014 4:06 pm • linkreport

@David Edmondson: I believe the New York subway has some differences in particular lines' infrastructures. So some cars are build specifically for certain lines.

I'm more familiar with Boston, which has vastly different infrastructure. For example, most lines use third rails, but the green line and the above-ground portion of the blue line use catenary. Different lines also have different platform heights, tunnel widths, etc.

Here in DC, it's all the same, so there's no reason to restrict trains to lines.

by Tim on Mar 3, 2014 4:40 pm • linkreport

What times of day do you use the rail and how many/what days per week ?

With the 2% delays I assume you do not work on weekends?

by kk on Mar 3, 2014 7:26 pm • linkreport

"I believe the New York subway has some differences in particular lines' infrastructures."

Broadly speaking, the New York City subway has two different sizes of cars -- one, slightly smaller, for the numbered services (1 through 7) and another for the lettered services (A/C/E, B/D/F, etc.) The difference dates back to the time when the NYC subway was built by private companies.

by Andrew on Mar 3, 2014 8:50 pm • linkreport

Brilliant work, very interesting. Thanks for sharing it.

by Ryan S on Mar 3, 2014 9:03 pm • linkreport

@David Edmondson

It's not surprising really when you consider the 1000 series comprises the largest fleet. As you know, they are restricted to "belly" service, so there's no such thing as a 1000-only train. In order to achieve maximum fleet utilization, it becomes necessary to place at least one pair of them in every consist. This renders single-series trains of any other vintage almost nonexistent.

There seems to be a widespread misconception that there is some disadvantage to running mixed-series trains. In fact there's no evidence to support this. There have been some issues with the intercom and PA systems, but this is a result of a system-wide retrofit to upgrade all cars with digital communication systems. These problems would have come to light either way, because at any given time there may be a train with cars all from the same series but only one or two of them have the new digital comm system. All other car borne systems operate completely independent of one another and don't "know" what type of car they're coupled to.

Interesting fact: the 4000 series is the only fleet 100% in service.

by dcmike on Mar 3, 2014 9:52 pm • linkreport

@dcmike:
That's not true.

The 2000s (76 cars), 4000s (100 cars), and 6000s (184 cars) are all fully in service.

The 1000s, 3000s, and 5000s have cars out of service due to crashes.

by Matt' Johnson on Mar 3, 2014 10:06 pm • linkreport

It took me 90 minutes to get from Capitol South to Bethesda the Saturday before last.

by Capt. Hilts on Mar 3, 2014 10:25 pm • linkreport

@Matt Johnson, 6050-6051 was involved in the New Carrollton Yard derailment in January 2013 and sustained major body damage and has been out of service since. I'm quite sure there's at least one set of 2000s out too but I don't know the number off the top of my head. I can check later.

by dcmike on Mar 4, 2014 6:08 am • linkreport

@dcmike

I seem to recall a report from somewhere that the drives on different series cars aren't quite in sync, so they speed up and slow down at slightly different rates, leading to jerkiness and potentially greater wear. You're saying that's inaccurate?

I also believe that the internal information display system in the more advanced cars doesn't work right if there are intervening cars from earlier series... so a 6000 at the front of the train may display the name of the next station and the side on which the doors will open, while one at the back may just say "Red." Is that incorrect?

by Dizzy on Mar 4, 2014 7:59 am • linkreport

The propulsion systems (or drives) on different series cars are in sync. It's fundamental to their operation. When a train is being driven, the operator (or the ATO system) selects an acceleration (or braking) mode from a fixed number of rates. It's not like an automobile with an infinitely variable throttle. In WMATA's case, there are 3 acceleration rates and 6 braking rates. The propulsion system responds to the rate request exactly the same irrespective of the car series. They are all engineered to operate within the same range of rates, and there are discrete feedback loops to ensure each car accurately maintains the requested rate (via motor current, voltage, rotational speed, and load weight [the car measures the number of passengers by sampling the pressure in the air springs]).

You are absolutely correct in that the 1000 and 4000 series cars inhibit the operation of the "Next Station" displays when placed in a mixed consist. I suppose I considered that more of an inconvenience to passengers rather than an operational issue. Even though these car series are slated for replacement, WMATA is actually look at addressing this. These cars are currently in the testing phase of being retrofitted with an FRA event recorder system, which incidentally would lend itself to supporting the means with which to relay the sign data to trailing cars.

by dcmike on Mar 4, 2014 9:46 am • linkreport

Mixed consists hurt reliability.

The different cars might be designed to work together and ensure the same rates of acceleration and whatnot, but the reality of operation isn't always so kind.

For more, read this: http://greatergreaterwashington.org/post/5186/time-to-reaassemble-railcars-into-single-series-trains/

The mixed consists have a 17% worse breakdown rate than trains of one car type.

by Alex B. on Mar 4, 2014 10:14 am • linkreport

We'll have to agree to disagree.

The linked PDF contains an unsubstantiated claim that does not disclose the source data. It also appears to be making a comparison using more than one variable: uniform consists running in ATO versus mixed consists running in manual operation. So did failures increase simply because consists were mixed, or did they increase because control devices that were not being used previously broke down? Or as a result of operators not skilled in manual operation inadvertently inducing faults? Or some other factor? I wouldn't look to a presentation prepared for a committee on customer service and operations for the answer.

by dcmike on Mar 4, 2014 11:45 am • linkreport

We'll have to agree to disagree.

Do we? Or should we shift the burden of proof? Prove to me that the mixed consists do not perform worse.

There's tons of anecdotal data and hard data out there that shows how mixed consists don't perform as well. Despite the design interoperability, the slight differences in design between the cars as well as the differences in age, wear, etc. make for different operating characteristics.

by Alex B. on Mar 4, 2014 12:03 pm • linkreport

More generic concerns about bellying the 1000-series and the impacts on reliability:

https://www.wmata.com/about_metro/docs/Vital_Signs_June_2011.pdf

"Railcar Maintenance staff will continue to address the maintenance impact of trains made up of mixed car
types. The restriction to operate the 1000 Series railcars in a belly-only configuration will continue to impact
overall railcar reliability as cars operate better with other cars of the same type. "

by Alex B. on Mar 4, 2014 12:06 pm • linkreport

Matt: have to ask, what are the lowest car numbers of each car series that you've ridden? In my 4 years in DC (using Metro mostly for discretionary trips...almost never for commuting), I managed to ride #1000, #1001, and #1003 (all on the Yellow Line).

by Froggie on Mar 4, 2014 12:15 pm • linkreport

One added nugget of info that will explain why proportionally the 4Ks and 6Ks are Matt's most ridden...

Each series has a home yard that is used for 30-day, 90-day, semi-annual, and annual car inspections. Individual cars clearly make their way around the system, and the below does not prevent a 6K from staying in the yard overnight at Shady Grove, but eventually they all have to come back to home base.

1K: Shady Grove & West Falls Church
2K: Alexandria & West Falls Church
3K: Alexandria & West Falls Church
4K: Shady Grove
5K: New Carrolton
6K: Greenbelt

So a regular commute on the red & green lines will naturally see far more 4Ks and 6Ks since (by my estimate) a good amount car series stay on the line of their home base.

by Rob K on Mar 4, 2014 3:37 pm • linkreport

Are your delayed trips evenly distributed between the lines you ride? Or do you find one line to be less reliable than the other?

by Sam on Mar 4, 2014 6:24 pm • linkreport

I also, ride the Green (Yellow) and Red Lines MOST of the time, and found this MOST INTERESTING.

I think I'm going to start the same "tracking" of cars ridden and any delays.

Thanks for the idea.

Mark - Berwyn Heights MD (Greenbelt / Green Line)

by Mark on Mar 4, 2014 9:58 pm • linkreport

For the initial wait you really need to record the time you wait all the time and then compare to the headway. For a 6 minute headway, ala green line greenbelt in the morning, on average you should wait 3 minutes. If you took 200 trips and totaled all the wait time and divided by 200 you should get 3 minutes. But if it is 4 or 5 minutes you would note a delay.

by Richard on Mar 5, 2014 8:15 am • linkreport

I'm impressed by your level of commitment and attention to detail. Can you do escalators next?

by Arnold on Mar 5, 2014 12:42 pm • linkreport

@MJ-this is great.

by Tina on Mar 6, 2014 9:01 am • linkreport

Add a Comment

Name: (will be displayed on the comments page)

Email: (must be your real address, but will be kept private)

URL: (optional, will be displayed)

Your comment:

By submitting a comment, you agree to abide by our comment policy.
Notify me of followup comments via email. (You can also subscribe without commenting.)
Save my name and email address on this computer so I don't have to enter it next time, and so I don't have to answer the anti-spam map challenge question in the future.

or

Support Us

How can our region be greater?

DC Maryland Virginia Arlington Alexandria Montgomery Prince George's Fairfax Charles Prince William Loudoun Howard Anne Arundel Frederick Tysons Corner Baltimore Falls Church Fairfax City
CC BY-NC