Bridge into Troubled Waters

The prediction could be cast as a bad news-good news joke. Bad news: in or around the year 2000, somewhere in the world a major cable-stayed bridge will fail catastrophically. Good news: it almost certainly won't be Alaska's own Sitka Harbor bridge.

This prediction comes courtesy of Henry Petroski, who is chairman of the department of civil and environmental engineering at Duke University and an uncommonly graceful writer on engineering-related topics. The prediction, which appeared in Petroski's regular column in American Scientist, concerns psychology almost as much as it does engineering.

Petroski begins with a British student's research. According to the dissertation of Paul Sibley with the University of London, a discernible pattern attends the successes and failures of bridge-building technology over the past 150 years. Sibley concentrated on four notable failures.

The earliest of these was the cast-iron bridge over the Dee River in England. In 1847, three 108-foot spans of the twelve-span railroad bridge collapsed under the weight of a passenger train. The bridge's designer, Robert Stephenson, was so horrified by the accident that he set himself to devising a new, stronger mode of bridge construction. The Dee bridge had been built of flanged girders trussed with tie rods; Stephenson's new bridges were pure trusses, far better able to withstand the lateral torsion that had wrenched apart his earlier bridge. Other engineers took up the design, and truss bridges crossed waterways everywhere.

In 1879, the truss bridge crossing the Firth of Tay dropped another train into the water. A court of inquiry found the bridge inadequate to the wind stresses imposed by its location. The railroad company halted work on another truss bridge then under construction, and commissioned a different design. The engineers came up with the cantilever bridge, which successfully spanned another Scottish firth.

But eventually that design proved fallible too. The struts buckled on a bridge over the St. Lawrence River in Quebec while it was yet under construction. The 1907 disaster killed 74 construction workers on the bridge when it fell.

Realizing that the cantilever design had severe shortcomings, engineers turned to suspension bridges. These had been working well at least since the Brooklyn Bridge was completed in 1883, but it turned out that this design also had it shortcomings. In 1940, the Tacoma Narrows bridge tore itself to pieces in a windstorm, earning itself the nickname of Galloping Gertie and casting a pall over the safety of suspension bridge design.

So, according to Sibley, a bridge collapses about every 30 years because engineers have asked too much of a given design. They've taken something that worked, made it lighter, less expensive, more elegant, until they've turned it into something that doesn't work.

Petroski takes the pattern through one more permutation. After suspension bridges fell out of favor, so to speak, box girder bridges next looked like the conservative designer's answer to spanning wide bodies of water. And so they seemed to be---until 1970, when both the Milford Haven (Wales) and the Yarra River (Australia) bridges suffered badly buckling plates. Petroski sees the same problems at work; engineers push the envelope, and eventually the envelope tears.

Which brings us up to the design of the moment, the cable-stayed bridge. Sitka's handsome harbor-spanning bridge, completed in 1971, was the first of its kind in the United States. According to Sibley's and Petroski's analysis of the pattern, that means it is likely to be conservatively designed and possibly even overbuilt. It should be capable of withstanding whatever humans and nature throw at it.

Not so its yet unbuilt younger siblings. By the time the century turns, engineers will probably design a cable-stayed bridge using cheaper, lighter, more elegant parts that ends with a splash landing. It's an expression of a very human problem, and certainly not one limited to engineers. Familiarity can breed stupidity.