A bright idea for easing traffic

Tweaking traffic-light timing to ease urban congestion and cut engine idling is one of the oldest, cheapest fuel-saving tricks in the book – a great idea for a nation laboring with $4 a gallon gasoline.

The problem is that the United States is not very good at it. Last year, a national “traffic signal report” gave the nation a “D” for the timing of its 250,000 intersection stop lights.

But Byungkyu Park wants to overhaul the system, using 21st-century timing methods that could save the US more than a billion gallons of gas a year. By applying raw computer horsepower and clever new algorithms to the old problem of “traffic signal optimization,” Dr. Park and his University of Virginia engineering team may have found a way to save more gas than all of the hybrids on the road today.

Most computer models today treat traffic flow as if it had the relatively uniform dynamics of a stream of water – all cars moving at about the same rate from intersection to intersection. But Park explains that traffic lights can be synchronized more precisely if traffic flow is calibrated on a “micro” rather than “macro” level – that is, on the number of individual cars and trucks on the road at a given time. That makes for a huge computational challenge, but also opens the door to more accurate computers models that mirror real-world traffic flows to a far greater degree.

“To solve this problem, we realized we should look into the individual vehicle level [and] understand individual human driving behavior,” Park says. “We are using a high-fidelity computer model to deal with individual driving behavior to create a microscopic traffic simulation program that models basically every single vehicle in the system.”

Park’s model allows for different driving styles that impede or change the flow of traffic. They include aggressive drivers that zoom ahead of the pack, lane changers, and slow-rolling truckers. His goal: to optimize light timing so erratic traffic ends up moving uniformly “like a train.”

To do that, however, would require taking special note of that bane of traffic regulators – the left-hand turn. Lights that hold left-hand turn lanes at a standstill too long can lead to long backups that impede traffic in other lanes, a common problem. His system would vary the length of left-turn lane signals to suit traffic conditions, allowing enough time for enough motorists to get through to prevent backups.

Even when signals are set to let drivers get where they are going as quickly as possible, that often does not save on gas and emissions, Park says. More tightly optimized signals, he found, often produced a trade-off that might delay travel “a few seconds,” but yield greater fuel savings and emissions reductions.

A major problem: speeders. Park found that his model would not conserve fuel well without some basic traffic enforcement to mute extreme driver behavior. For instance, drivers that accelerate to 55 miles per hour in a 45-m.p.h. zone create too wide a “distribution of desired speed” for effective fuel savings. “If you can assume a higher enforcement rate, or a better compliance rate, it may be possible to achieve both the fewest delays and the best fuel consumption,” Park says.

If traffic-timing systems were updated using conventional methods, the US could cut fuel consumption by up to 10 percent – about 17 billion gallons a year, the National Transportation Operations Coalition found in its 2007 traffic-signal report card. (Emissions also would be by cut more by than 20 percent.) If Park’s hyper-efficient micro-model traffic-timing method were widely applied, another 1 billion to 2.5 billion gallons could be saved.

“With the increase in computer horsepower and improvements in sensors [embedded at intersections], we could take a lot of ideas and put them into the much more doable range for city staff and budgets,” says Timothy Lomax, a research engineer at Texas A&M University’s Texas Transportation Institute and lead author of its national mobility report.

Last year’s mobility report found congestion across 437 urban areas wasting 2.9 billion gallons of fuel. Signal timing isn’t the only cause of congestion, but it’s one of the “top five” reasons, he says, and may account for as much as half of the congestion.

For about $250 million a year – about $1,000 per intersection – the entire US could have well-timed lights, far less congestion, and far higher fuel savings, says Philip Tarnoff, director of the Center for Advanced Transportation Technology at the University of Maryland.

The payoff in fuel savings alone, calculated before this year’s run-up in gas prices, was a return on investment of at least 40 to 1 and perhaps more.

But politicians tend instead to spend billions on building or widening roads – a far more costly approach to yield similar results, Dr. Tarnoff says. “You would think traffic-signal optimization would be a no-brainer,” he says. “But it’s hard to get the federal government to fund traffic light timing. It’s kind of dull stuff that’s just not going to get votes.”

Some cities, however, have taken the initiative and are reaping rewards. After Plano, Texas, received a “C” on its 2005 traffic-signal report card, the city got to work timing its lights and won a “B” grade last year. It also saved about 850,000 gallons of gas, eliminated 36 million unnecessary stops, and cut 745,000 hours of delays, it estimates.