I-35W bridge collapse – What happened on August 1st and after
August 29, 2007 2 Comments
One of the interesting scientific questions that emerges from the tragedy of the I-35W Bridge Collapse is how traffic responds. There are several time horizons for looking at this.
Most immediately are those who are on the link leading up to the bridge. MnDOT’s traffic cameras show the cars turning around on the freeway within seconds of the bridge collapsing, before the dust clears literally. “Video footage of the collapse from Mn/DOT traffic camera 628. 6:05 p.m., Aug. 1, shows an edited two-minute clip from a traffic camera at the south end of the bridge. Initially, the camera is pointed to the south away from the bridge. When traffic comes to a stop, the camera pans to the north where the bridge has just collapsed. (wv file)”. This is a rational response on the part of drivers who don’t know what else may collapse. As my wife says, there are two types of people “those who run towards the meteorite and those who run from it”. Survivors are those who ran from it.
Over the next few minutes and hours, word of the bridge collapse spread. My student Shanjiang Zhu has organized MnDOT’s loop detector data into a movie that shows the 15 minute traffic counts on all the loop detectors in the Twin Cities, comparing that number with the previous Wednesday’s count at the same time of day. Blue indicates lower volumes, red higher volumes. Clearly after the collapse, people heard quickly through various sources (cell phone, variable message signs, radio, etc.), and avoided large swaths of I-35W in the vicinity (which turns blue) and complementary feeder links, while competititve substitute links (Mn 100, I 35E, parts of I-94) saw an increase. We still have to compute how overall traffic volume and Vehicle Kilometers Traveled changed.
Once people were informed, on subsequent days people searched for alternatives. The alternative the first day for some was to avoid driving, but that quickly changed, and different routes became natural substitutes. A second movie compares the counts on the 15 days after the collapse with the average of the previous 8 weeks same day of week (so a Thursday is compared with the eight pre-collapse Thursdays). This illustrates the changes network wide. The
movie is available.
Finally, there may be some longer term adaptations, but we don’t have enough information only one month into the changed situation to know about this yet. With colleagues Henry Liu and Kathleen Harder, we have obtained a National Science Foundation Small Grant for Exploratory Research to look at all of these issues in some more depth.