Yesterday it took me over two hours to get from Framingham to Haverhill, a trip of about 60 miles almost entirely on Interstates. This isn’t unusual at rush hour on Friday, and it gave me lots of time to think about traffic congestion. Why does traffic slow down periodically and then suddenly speed up again with no apparent causes? I started forming a hypothesis that involved considering the events as shock waves, and today I searched for “traffic shock wave” to see what thoughts experts have had on the idea.
Sure enough, it turned up quite a few relevant hits. The idea is fairly obvious: When traffic is dense enough, any slight disruption in the flow of traffic can propagate backwards, forcing many drivers to slow down. People can speed up again at the head of the jam, since the slowdown has created a relative vacuum ahead of them, but others are caught in the tail. By the time the congestion dissipates, the driver who caused it is long gone. This has been confirmed in simulations and on a test track.
A refinement to this idea which I haven’t seen, maybe just because I don’t know the keywords to look for, is that once the effect starts, it should be prone to positive feedback. If a driver changes lanes in dense traffic, others must slow down to maintain distance. As gaps get smaller, people will have to brake more when someone cuts in in front of them. This makes the traffic still denser, so the next lane change has a stronger effect. If you’ve ever seen an idiot cut across three lanes in heavy traffic to get to an exit, you know just how strong it can be.
Each lane change may only cost a couple of MPH, but the cumulative effect can bring traffic to a standstill.
If this hypothesis is valid, you’d expect such jams to be especially common in areas where tailgating is the rule and drivers rush into any opening in a lane. Los Angeles, at least by reputation, confirms what I expect.
If anyone knows of any work that’s been done on this notion, I’d like to hear about it.