Applying physics to traffic
Oct. 2nd, 2003 06:11 pm![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
jenkFrom Sharon Begley's science column in The WSJ:
Now There's Proof --
You're in a Traffic Jam
For No Good Reason
You've gotten it down to a science. Although the radio is in full panic mode -- reporting a jackknifed tractor-trailer on the highway, extra volume at the bridge, a protest blocking traffic two blocks from your destination -- you are sure you can arrive on time by outfoxing the fools who are wedded to their usual routes and finding the road less traveled.
Scientists have some news for you: Don't bother.
Theoretical physicists began delving into traffic in the late 1990s, sicking equations of nonlinear dynamics and models of cellular automata (yes, they are as opaque as they sound) on such mysteries as why stop-and-go traffic can, for no evident cause, start zipping along again. Although the research has yet to produce a practical way to eliminate congestion, it has shown that there are empirical causes for such miseries as traffic slowing to a crawl even when there is no accident or rough pavement in sight.
It isn't that the highway gods have it in for you. Physics does.
Many of the successes applying theoretical physics to traffic mysteries spring from the fact that cars behave much as molecules in a gas. The equations of gas dynamics, therefore, do a pretty decent job of modeling and predicting traffic.
In a flowing gas, a random fluctuation in molecular movement -- analogous to a car slowing down briefly, perhaps while the driver tunes the radio -- flips smooth flow into a clogged mess. A compression wave, which is simply a zone where the molecules/cars have gotten very packed, propagates backward from the bottleneck.
So far, so intuitive. But the compression wave can persist for hours. That leaves drivers who never saw the first car decelerate totally mystified about why they have slowed to a crawl. By one estimate, three-quarters of traffic jams have no visible culprit. The cause came and went hours ago, but its effects linger.
A defining trait of nonlinear phenomena, such as moving cars and moving gas molecules, is that minuscule fluctuations can have effects disproportionate to their size, like the proverbial butterfly flapping its wings in Beijing creating a hurricane across the ocean.
One such fluctuation is a brief burst of cars from an on-ramp. It makes sense that if you suddenly get, say, a 20-car funeral cortege entering a road, cars behind it will likely have to slow down. But even after the cortege is at the cemetery and the on-ramp traffic has dropped below normal, a bottleneck can persist at the merge. Not trusting the memory of drivers, German physicists confirmed this effect with data from sensors on European highways. Next time you have to slow down as you pass an on-ramp that's empty, blame the butterfly effect.
In the most recent research, three physicists (including one in Rome, where they know a thing or two about driving in maddening tie-ups) modeled urban traffic. The goal was to see whether drivers have any chance of avoiding jams. In other words, if you know that on Avenue A traffic often flows like refrigerated honey, but on Avenue B it zips along, can you find a fast route?
When the city is relatively empty, says physicist Andrea De Martino of the University of Rome, drivers who choose a way to avoid jams manage to find fast routes. If you listen to traffic reports at 5 a.m., you'll know that you should head for Avenue B.
But as the streets fill up, the density of cars reaches a tipping point. They become equally distributed over the grid, as in midtown Manhattan at rush hour.
In such a traffic-dense city, the savvy driver has a problem. "Finding a fast route becomes extremely hard," says Prof. De Martino, who led the research described in a paper submitted to the journal Europhysics Letters. "Drivers start changing their route every day. This leads to a dramatic increase in traffic fluctuations."
Fluctuations mean that some arteries are clogged on Monday, others on Tuesday. You have no better chance of finding the fast road than of choosing the fastest line at the grocery checkout.
Overall, these smart drivers have an even lower chance of avoiding snarls than do drivers who choose their route randomly. According to the model, when traffic is heavy throughout the urban grid, the laid-back drivers actually fare better than drivers intently seeking out the fastest route. "Being traffic-smart is good in an uncrowded situation, but it's extremely inefficient in a congested one," says Prof. De Martino.
Memo to road warriors whose brilliant alternative routes don't spare them snarls: Give up.
Theoretical physicists haven't exactly endeared themselves to the civil engineers, who long had traffic to themselves. Their approach to congestion has been to build more roads, time vehicle entry and eliminate bottlenecks, such as bad merges. Unfortunately, the power of tiny, chance fluctuations to cause gargantuan snarls means that such rational measures are destined to fail often. Stuff happens.
As engineers validate more of the physicists' traffic models, theorists are prepared to refine their work. As Prof. De Martino admits, "Whether the models are a good description of human drivers, I do not know. I assure you that there are no good models of Rome's drivers."
Now There's Proof --
You're in a Traffic Jam
For No Good Reason
You've gotten it down to a science. Although the radio is in full panic mode -- reporting a jackknifed tractor-trailer on the highway, extra volume at the bridge, a protest blocking traffic two blocks from your destination -- you are sure you can arrive on time by outfoxing the fools who are wedded to their usual routes and finding the road less traveled.
Scientists have some news for you: Don't bother.
Theoretical physicists began delving into traffic in the late 1990s, sicking equations of nonlinear dynamics and models of cellular automata (yes, they are as opaque as they sound) on such mysteries as why stop-and-go traffic can, for no evident cause, start zipping along again. Although the research has yet to produce a practical way to eliminate congestion, it has shown that there are empirical causes for such miseries as traffic slowing to a crawl even when there is no accident or rough pavement in sight.
It isn't that the highway gods have it in for you. Physics does.
Many of the successes applying theoretical physics to traffic mysteries spring from the fact that cars behave much as molecules in a gas. The equations of gas dynamics, therefore, do a pretty decent job of modeling and predicting traffic.
In a flowing gas, a random fluctuation in molecular movement -- analogous to a car slowing down briefly, perhaps while the driver tunes the radio -- flips smooth flow into a clogged mess. A compression wave, which is simply a zone where the molecules/cars have gotten very packed, propagates backward from the bottleneck.
So far, so intuitive. But the compression wave can persist for hours. That leaves drivers who never saw the first car decelerate totally mystified about why they have slowed to a crawl. By one estimate, three-quarters of traffic jams have no visible culprit. The cause came and went hours ago, but its effects linger.
A defining trait of nonlinear phenomena, such as moving cars and moving gas molecules, is that minuscule fluctuations can have effects disproportionate to their size, like the proverbial butterfly flapping its wings in Beijing creating a hurricane across the ocean.
One such fluctuation is a brief burst of cars from an on-ramp. It makes sense that if you suddenly get, say, a 20-car funeral cortege entering a road, cars behind it will likely have to slow down. But even after the cortege is at the cemetery and the on-ramp traffic has dropped below normal, a bottleneck can persist at the merge. Not trusting the memory of drivers, German physicists confirmed this effect with data from sensors on European highways. Next time you have to slow down as you pass an on-ramp that's empty, blame the butterfly effect.
In the most recent research, three physicists (including one in Rome, where they know a thing or two about driving in maddening tie-ups) modeled urban traffic. The goal was to see whether drivers have any chance of avoiding jams. In other words, if you know that on Avenue A traffic often flows like refrigerated honey, but on Avenue B it zips along, can you find a fast route?
When the city is relatively empty, says physicist Andrea De Martino of the University of Rome, drivers who choose a way to avoid jams manage to find fast routes. If you listen to traffic reports at 5 a.m., you'll know that you should head for Avenue B.
But as the streets fill up, the density of cars reaches a tipping point. They become equally distributed over the grid, as in midtown Manhattan at rush hour.
In such a traffic-dense city, the savvy driver has a problem. "Finding a fast route becomes extremely hard," says Prof. De Martino, who led the research described in a paper submitted to the journal Europhysics Letters. "Drivers start changing their route every day. This leads to a dramatic increase in traffic fluctuations."
Fluctuations mean that some arteries are clogged on Monday, others on Tuesday. You have no better chance of finding the fast road than of choosing the fastest line at the grocery checkout.
Overall, these smart drivers have an even lower chance of avoiding snarls than do drivers who choose their route randomly. According to the model, when traffic is heavy throughout the urban grid, the laid-back drivers actually fare better than drivers intently seeking out the fastest route. "Being traffic-smart is good in an uncrowded situation, but it's extremely inefficient in a congested one," says Prof. De Martino.
Memo to road warriors whose brilliant alternative routes don't spare them snarls: Give up.
Theoretical physicists haven't exactly endeared themselves to the civil engineers, who long had traffic to themselves. Their approach to congestion has been to build more roads, time vehicle entry and eliminate bottlenecks, such as bad merges. Unfortunately, the power of tiny, chance fluctuations to cause gargantuan snarls means that such rational measures are destined to fail often. Stuff happens.
As engineers validate more of the physicists' traffic models, theorists are prepared to refine their work. As Prof. De Martino admits, "Whether the models are a good description of human drivers, I do not know. I assure you that there are no good models of Rome's drivers."
