I must be penny-wise and pound-foolish.
After all, somehow, the small gripes that cost me a few seconds always end up turning into massive research projects. My article in February on Salt Lake City International Airport’s slow-moving walkways is a classic example for which I’m still made fun of — a minute lost per airport trip turned into a multiweek investigative journey.
This week’s data column is in that vein. I recently found myself frustrated by having to wait about three seconds longer than I usually did at one particular intersection, the one at Interstate 80 and 700 East. You see, previously, the intersection was religiously smooth sailing heading northbound: one green light led into another, meaning I didn’t have to stop at the second light.
My expectations were shattered in late July. All of a sudden, the green light at the next signal didn’t come, forcing me to come to a halting stop ... before the green came on seconds later.
Like I said, it’s a small gripe.
Since then, I’ve spoken to Utah Department of Transportation signal engineers, researched the system they use, and learned about how all this works. Truth be told, I’m pretty impressed. I also learned of the best way to report problems to UDOT — so that if you have a traffic-light beef, you too can take action.
As you’re probably aware, most intersections have a host of technology associated with them.
There’s the traffic signal controller itself, usually found in a metal box at one corner. Most intersections have sensors just before the intersection lines, which detect if cars are waiting at a light or passing through. Many also have cameras poised atop the traffic lights to allow UDOT workers to monitor those intersections with their eyes.
That’s not all. Many intersections also have radar detectors or other sensors, located about 300 to 400 feet before an intersection, to count traffic approaching a light, as well as to learn about the speed at which cars they are traveling. The sensors generally track information 10 times per second, meaning that just the sensor data alone ends up being about 19 megabytes per day for each intersection. Some traffic light controllers upload that data to UDOT’s servers in real time; in others, they upload it every 10 or 15 minutes.
That’s a lot of data when you multiply it by the 2,000 or so intersections UDOT tracks. It ends up being about one terabyte per month across the state. But it’s useful information, which can be used to optimize traffic-light timing in all of Utah.
How does UDOT keep track of it all? In 2011, the the department developed the Automated Traffic Signal Performance Measures system, with help from Purdue University and the Indiana Department of Transportation. ATSPM is an open-source software that allows signal engineers to monitor individual stoplights and the whole system to see where things can be improved. Federal transportation managers have since encouraged the system to be adopted by various transportation departments across the country.
And because they were after my own heart, they put an impressive amount of transparency into the system. In particular, the entire dataset is available online for anyone to see — not just UDOT traffic engineers, not just consultants, not just academics who have petitioned UDOT for access, not just journalists. Anyone. It’s all at UDOTtraffic.utah.gov/ATSPM.
“Our goal in optimizing mobility, improving safety, preserving infrastructure and strengthening the economy,” the ATSPM’s Frequently Asked Questions page says, “means that all who have a desire to use the data should have access to the data without restrictions.”
I love this so much.
Using that traffic data
So what can you do with this data? Well, you can measure how efficiently traffic lights are operating. There are number of different goals signal engineers can examine — and optimize — in the ATSPM:
• Arrival on green: What percentage of vehicles arrive at an intersection where the light heading in that direction is already green? To maximize this, engineers frequently put lights along one street on the same cycle length — say, 90 seconds — so that lights in one direction can sync up.
• Approach delay per vehicle: When approaching an intersection — or waiting at one — how much is each car delayed, on average?
• Split failure: How many times does a red light come before all of the traffic waiting at a light is able to get through?
• Pedestrian phases: Pedestrian traffic wildly differs. Most intersections assume pedestrians will cross streets at about four feet per second, or 2.7 mph. Some intersections, near schools or elderly homes, assume slower walking times. In general, how efficiently does an intersection manage pedestrian traffic alongside its vehicle duties?
This can all be calculated for any stretch of time. So, for example, I can look at the intersection of 4500 South and State Street, and see which directions saw green at which times between 5 and 6 p.m. on Thursday.
Or take my small gripe intersection, at I-80 and 700 East, where I had to wait a few seconds longer than usual. The arrival on green percentage for cars hitting northbound took a sudden tumble, just as I experienced:
With it came longer average delay at the intersection as well.
Again: ATSPM is online for anyone, so you can pull charts like these, and many others for the intersections you use on a regular basis. The biggest difference is that, on the system’s back end, UDOT signal engineers can adjust the plans the actual lights use to try to minimize traffic.
Reporting necessary fixes
How do the signal engineers know when things aren’t working well at a particular intersection?
First, there’s a fairly regular recalibration. UDOT goes to each of its intersections every few years (more often in places with extreme growth, like the southwest end of the Salt Lake Valley) to recalculate average traffic counts and rebuild light timing from there.
What about for more urgent problems? In some cases, the system can raise red flags. For example, if a sensor at a particular stoplight shows no cars triggering it for an unusual amount of time, it probably means that the sensor is broken. The system notifies engineers, who then notify technicians to go out to the signal to investigate.
Sometimes, that’s a quick fix that can happen in a couple of days. Other times, parts have to be shipped out, and the road temporarily closed, to repair the troubled sensor. In either case, the signal engineers will find a way to patch the light’s timing — by, for example, telling a light to always give one direction a green light every cycle, even if no car is detected in that direction due to the broken sensor.
In a large number of cases, problems require human detection. That can come from either the signal engineers watching the system or from the general public. UDOT has a number of ways you can reach out, but the preferred method is through the “Click ‘n Fix” App, available for both Apple and Android devices. There, you can send reports of things not working smoothly — traffic lights, yes, but also things like potholes and broken cameras.
In the case of the intersection at I-80 and 700 East, the slight delay in light timing actually was causing a traffic problem at peak times. Cars waiting for the light to change were sometimes blocking the turn lanes for traffic coming off the freeway. When that happened, engineers investigated the problem. UDOT suspected that a power surge had caused two connected lights to become disconnected, resulting in the mismatched green-light timing.
Coincidentally, the problem was fixed a day before I called UDOT. Arrival on green percentages are back to normal — and so is my commute.
Now, how do I get my seconds back?
Andy Larsen is a data columnist for The Salt Lake Tribune. You can reach him at email@example.com.