During the Monday morning rush, many Red Line riders got their first notice about potential delays when they saw an electronic alert issued by Metro. The Twitter version, at 8:26 a.m., said: “Red Line passengers may experience delays in both directions – trains single tracking between Medical Center & Grosvenor.” The problem turned out to be a cracked rail, and the trains had to share one track in that area through the peak period.
Early Wednesday morning, a crash on the Capital Beltway’s outer loop near Georgia Avenue delayed thousands of drivers in an area that doesn’t need any help to be congested. Some of those drivers might have benefited if they had known that the Intercounty Connector, East-West Highway or University Boulevard could serve as alternatives for westbound travel.
The information available to most travelers in the D.C. area can tell them what’s happening on their routes at that moment, but it doesn’t play out the consequences for them when they take to the roads or rails. Even the currently available information is useless if travelers don’t look for it before they leave home or office — and many of them just go. It’s only when they run into trouble that they check signs, smartphones or radio reports.
A better way: Calculate the impact of this moment’s problem on an individual commute coming up in, say, a half hour or an hour. Push that information out to the traveler, along with advise on rerouting the commute to get around the problem.
Matt Cole, executive vice president with Cubic Transportation Systems, is among those who says he can achieve that, and in the not too distant future. Better use of data to predict commuting problems works both for the commuters and for the transportation departments that build and operate the travel network, Cole says.
In a phone interview this week, we discussed commuter challenges and what it would take to get us to the world he and others involved in travel data systems can imagine.
“People self-optimize,” he says. When commuters follow one route for months, or years, they figure out when they need to wake up and when they need to get out the door to be on time for work. For many, the commuting time equals wasted time, so they may not leave much of a margin for error on their trips. If there’s a problem on their route to work, there may be consequences with the boss. “What that means, though, is that when something goes wrong, people haven’t planned for it to go wrong.”
(Cole speaks about commuting from a national and international perspective. In the heavily congested D.C. region, as well as in other urban areas, many commuters build in “buffer time,” to account for the unpredictability of long commutes, which may involve several modes of travel. I think his vision regarding the usefulness of predictive data can work for both types of commuters, the ones who think they know the route so well that they leave little margin for error, as well as the ones who think they know the route so well they build in lots of buffer time.)
The key element in his concept is that real-time information isn’t good enough anymore. Commuters need predictive data, and they can have it.
Fast forward a few years and play out that same morning commute on the Red Line, with a traveler from western Montgomery, or Frederick County. The bedside alarm is set for the normal wake-up time that would allow the commuter to get to the Shady Grove station for a ride to an office near Metro Center. The commuter uses a smartphone app that monitors all elements of the trip and detects problems. Maybe there was a crash on Interstate 270 near Shady Grove, or maybe it’s a repeat of the cracked rail near Metro’s Grosvenor station.
Here’s the great part: The monitoring system doesn’t just know what’s going on right now, while the commuter is still asleep. It uses data to predict what the commute will be like an hour from now, when the commuter is ready to walk out the door. If that means the commuter needs to get up early, the smartphone can set off an alarm.
But that doesn’t complete the mission: In this future world, the traveler may be set up to pay for the Metrorail ride via smartphone. The app could alert the traveler that a better alternative on this day is to go a MARC station or a commuter bus stop. The app already would have arranged for fare payment by one of those alternative routes.
The app does all the planning and worrying for you, because you’ve let it in on your travel habits. “You could store your regular journeys through a trend of the transactions that you create,” Cole says. You wouldn’t need to fret about whether you’ve left enough time for the trip. If the program decides that you haven’t, because of some unexpected problem, it’s going to alert you in time to do something about that.
Cole thinks the technology to do this could be available in five years. But several key things need to happen to make such a system work. The flow of data into the monitoring system needs to be complete and constant. Transportation agencies regionwide need to participate in sharing travel data and fare systems. The program needs to understand how traffic and transit planners will respond to disruptions. “The playbook would be programmed in,” Cole says.
Let’s look at that last point, based on Monday’s Red Line situation. The program would need to know much more than the riders did at first to make its predictions. In fact, it would need to know more than Metrorail controllers did at first.
In an e-mail exchange, Metro spokesman Dan Stessel described today’s electronic alert system:
The goal is to give a first alert — and then, if it is severe, to follow it with additional messages, as you saw us do on Sept. 30 (alert went out saying “consider alternate travel options”).
But here’s why it’s not possible to give that information in the first alert, which we issue at about the 10-minute mark: Oftentimes, the first people on the scene of an issue are not the same people who are qualified to assess its severity. Take today [Monday], for example. The signal system detected the cracked rail, as it should, but the initial response is for a signal problem (that’s what it looks like in the control center) — so, ATC [automatic train control] personnel are dispatched.
When ATC discovers it’s actually not a signal problem but rather a break in the rail, they are not qualified to assess it. We need track department professionals on the scene to say whether it is safe to continue running at restricted speed, or whether the repair can be temporary (short duration single-tracking) or needs a permanent fix right then and there. We need the Emergency Response Team members (track department experts) to arrive for that. And in rush hour traffic, that can take time.
A predictive program isn’t going to know that before the transit staff on the scene does, unless the track technology gets more sophisticated. But even at that point, the program needs to understand how the Metrorail controllers will respond to the problem of moving riders around a problem at rush hour.
Faced with the need to have trains on both directions share the one good track, the controllers used a greater-good theory. They gave priority to the inbound trains, the ones carrying the most passengers during the morning rush. Riders on the outbound trains heading toward Shady Grove were most affected by this pattern. Their trains skipped the stops at Medical Center and Grosvenor to clear room for the inbound trains.
The predictive program would have to know that Metro controllers will follow this procedure for a problem near Grosvenor. And the program will have to know that if the single-tracking occurred at Metro Center, the controllers would not react the same way. In the heart of downtown during the morning rush, their is no priority direction for commuters.
But outside the core, Metro officials like the way this priority system works. Stessel said that during the problem hours of 8 to 10 a.m. Monday, the extra wait time for the inbound Red Line trains was no more than two minutes.
A predictive program would need to understand not only what the train controllers will do, but also what the results are likely to be.
There’s little reason to doubt we’ll get there. The potential benefits are too great for both the commuters and the transportation agencies. Cole cites a comment made during the Fix My Commute forum by Di-Ann Eisnor, an executive with the Waze travel navigation service. She had noted that transportation systems don’t have a capacity problem as much as they have a distribution problem. Too many people want to use the same route at the same time. If there’s a way to help them spread out — both by time and distance — it eases stress on the individual commuters and on the transportation network.
This works on transit as well as on highways. “If you have a bus and rail line on the same corridor and the bus is under capacity, you can create some multi-modal shift,” Cole says. “The bus will have better utilization of capacity.”
- YouTube video shows Cubic Transportion’s NextCity strategy in a London Transport Museum Exhibit.
- How can we plan for commuting’s Futurama?