In May 2013, a bridge spanning the Skagit River along Interstate 5 in Washington state catastrophically collapsed, after an oversized trailer clipped one of the bridge's cross beams. A new analysis by engineers at the University of Illinois at Urbana Champaign confirms the many factors that contributed to the collapse, and offers recommendations for how to prevent similar failures in the future.
They published their analysis in the Journal of Performance of Constructed Facilities. "The bridge repair costs exceeded $US15 ($20) million, and that doesn't account for the economic losses that the area felt because they and visitors no longer had access to the interstate," co-author Tim Stark said in a statement. "Even though this accident occurred three years ago, it's still very important because many bridges have this same design, not only in Washington but in other states."
An entire section of the bridge fell into the water, taking two other cars with it. Fortunately there were no fatalities, although three people suffered minor injuries. Built in 1955, the bridge was a crucial link between Vancouver, British Columbia, and Seattle, with more than 71,000 vehicles crossing every day. After the collapse, people naturally scrambled to explain how it could have happened, including Mike Lindblom, a reporter from the Seattle Times:
A preliminary report by the National Transportation Safety Board attributed the failure to the bridge's so-called "fracture-critical" design, whereby a small crack in just one essential part can trigger a chain reaction of even more failures. That's what happened in 2007, with the collapse of the similarly designed I-35W Mississippi River bridge in Minneapolis. There are more than 10,000 ageing through-truss bridges in the U.S.
The Illinois team suggested ways to reinforce such bridges to reduce the risk of catastrophic failure in the future. "[T]he initial damage of where the truck hit was not a primary support, it was a cross beam, and the damage cascaded, causing the entire collapse," co-author Jim LaFave said. "We can selectively add supports so there are ways to redistribute the impact load, so the structure can remain stable and stay standing even if there's damage to a particular area."
And then there was human error. The truck with the oversized load was erroneously driving in the outside lane, where the clearance was lower, in part to give space to another passing truck.. The state's Department of Transportation rubber-stamped an oversize-load permit to the trucking company without verifying clearance limitations along the proposed route. And the driver underestimated the height of his load by a critical two inches.
Stark and his colleagues attribute much of this to poor record-keeping. "The key issue in this case is the variable vertical bridge clearance," he said. "Many bridges have a square opening, so the clearance is the same across all lanes. The problem with this bridge was that it curved down over the edge lanes. The oversized trailer was 4.57m 23cm tall. The database said the bridge was 5.18m 8cm, which was in the center — almost two feet higher than the edges, which is where the oversized trailer was travelling."
The curved opening of the Skagit River bridge was a key factor, because the posted height was the maximum in the center. (Image: Tim Stark)
The authors recommend changing policy to report the lowest vertical clearance for bridges, rather than the highest — and periodically verifying that data with LIDAR measurements.
There was a pilot car travelling ahead of the truck, outfitted with an antenna to detect any possibility of too-low clearance. But the driver of the pilot car admitted to using a cell phone (albeit with a hands-free device) while crossing the bridge. Even if the pilot car had alerted the truck driver, there would not have been time to avoid a collision. The driver was following the pilot car too closely: more than half the recommended distance for adequate response (121.92m, or a five-second response time, instead of the recommended 263.65m, allowing for a 10-second response time).
"In this case, the pilot car either did not impact the bridge or the driver didn't hear the impact. They never called the truck, so that part of the safety mechanism failed," Stark said. "One solution we suggest is a sensor at the top of the pole that automatically contacts the oversized vehicle if it hits an object. This eliminates the drivers having to communicate quickly so the oversized vehicle can change course. Also, the pilot car antenna was not straight, so it was not accurately measuring the full height."
It will never be possible to eliminate human error entirely. We are fallible creatures, and sometimes all those tiny lapses in judgement — typically harmless on their own — combine in disastrous ways. But the lessons learned from the 2013 Skagit River bridge collapse could make catastrophic failures just a bit less likely in the future.