Ads are being blocked

For us to continue writing great stories, we need to display ads.

Un-block Learn more


Please select the extension that is blocking ads.

Ad Block Plus Ghostery uBlock Other Blockers

Please follow the steps below

The Case for Driverless Trains, By the Numbers

In 8 revealing facts and figures, plus 3 charts.

The Washington Metro red line has resumed automated operation for eight-car rush-hour trains. (John / Flickr)

Automatic train control returned to the Red Line on Washington, D.C.'s Metrorail system today, following years of infrastructure upgrades that occurred in the wake of a deadly 2009 crash. For now, the change will be limited to eight-car rush-hour trains, which will once again start and stop via computer control. Human drivers will remain on board to open and close doors, and other parts of the system won't be automated for a while to come, but the news was a win for an agency that sorely needed one.

CityLab has made the case for driverless trains before, but some new evidence from an impressive survey of 23 rail-based transit systems around the world that operate (or have plans to operate) highly automated lines gives us several new facts and figures to bolster that position. The work comes courtesy of a group of researchers at the Railway and Transport Strategy Centre of Imperial College London, who presented it at a conference earlier this year. Let's dive in.

There are 4 main types of automated trains.

Type 1 trains have automated elements but are manually driven. Type 2's have automatic train operation (ATO) but a driver in the cab to operate doors and get the train started. Type 3 trains have an attendant in a passenger car to operate doors. And Type 4 ATOs, the highest, are capable of going totally driverless (though occasionally these, too, have an official on board for other purposes).

Automated trains only operate on 6% of the world's total rail-transit line length.

At least, that was the case as of 2013, according to the Imperial College London research team. Some notable metros with automated service include Paris (Line 1), Copenhagen, and Vancouver. London has announced plans to go driverless in the future, and Honolulu is working toward the first fully automated major system in the United States.

Honolulu plans to build the first fully automated major metro system in the U.S. (HART)

Staff savings can be as much as 70%.

Or, to be more precise, the ratio of staff per asset (trains plus stations) was found to be 70 percent less in unattended automated systems compared with "fully staffed" systems (see the chart below). On automated systems where stations but not trains are staffed, the staff savings were still 30 percent. That's a huge reduction in labor costs, saving more money for actual operations.

Cohen et al (TRB, 2015)

The highest-frequency metros (42 trains per hour) are fully automated.

Speaking of saving money for actual operations, the Imperial College researchers point out that the two highest-frequency metro lines in their study—running 42 trains an hour—are both Type 4 fully automated systems (below). A majority of the existing and planned automated lines they came across ran more than 30 trains an hour. Automation isn't the only factor in greater frequency, but the two certainly appear to be closely related.

Cohen et al (TRB, 2015)

The rate of return for automation is estimated at 10-15%.

Obviously, cost savings will vary from system to system, and automation requires a significant up-front investment from transit agencies. But the operational savings should be significant (reportedly 30 percent in the case of the Paris Metro, compared with conventional service). And two transit systems in the Imperial College report estimated their rate of return at 10 to 15 percent.

Driverless trains have 4-6% more room for passengers.

Removing the driver's cab saves space on the train. The Paris Metro reportedly added 6 percent to its capacity, and one system in the Americas estimated a capacity savings of 4 to 5 percent, according to the Imperial College study. (Sadly, systems in this study were kept anonymous as part of the terms of the data exchange.) One European system also reported a savings of 1 percent on the cost of trains that came without cabs.

Line 1 of the Paris Metro has seen big benefits from embracing driverless trains. (Dave A / Flickr)

Automated lines were in the top third for reliability.

Six ATOs provided data on reliability, measured as distance between incidents that caused more than a 5-minute delay. Compared with whole transit networks in the study, these lines were in the top third in terms of reliability, with one fully automated line improving reliability by 56 percent (below). Even upgrading from a Type 1 to a Type 2 ATO provided up to a 33 percent reduction in 5-minute-delay incidents, according to other research.

Cohen et al (TRB, 2015)

Headway regularity—i.e. space between trains—is better than 99% on automated systems.

One system reported that headway regularity on their automated trains is upwards of 99.8 percent. That means less train bunching, and thus a better experience for riders waiting on the platform. More anecdotally, in discussing the upgrades to D.C.'s Metro, Deputy General Manager Rob Troup pointed to headway regularity and the general smoothness of the ride as huge benefits to the computer-driven system:

“As all the trains start to move to ATO, we’ll have consistent rides, consistent headways,” Troup said, referring to the spacing of trains. “We won’t have trains bunching up. We won’t have trains lagging. And we’ll have smoother acceleration, smoother stops.”

So there you have it: a small army of facts and figures pointing to the operational benefits of automated trains, in addition to the well-known safety benefits of reducing human error. Obviously there are trade-offs—trains with more advanced automated capability will cost more and could mean fewer jobs, but they'll also potentially reap greater ridership rewards—and no train is completely immune from accidents. But the case in favor of automation is a strong one.

About the Author