I want to expand on yesterday’s brief comment about the design issues with the Whitestone Bridge, because it illustrates not just a moment in the history info suspension bridges, but a never-ending tension in structural engineering.
The picture above, from the 1991 HAER survey of the bridge, gives a good sense of its appearance. The fact that Othmar Ammann was involved is very much visible in the tower design, which is pretty much the tower used at every suspension bridge he worked on. This picture obviously shows it after the 1943 side trusses were added for additional stiffness, and before they were removed in 2003 in favor of wind fairings. The trusses are fastened above the original plate girder stiffeners, with the girders serving as the bottom chord for the trusses. Both the girders and the trusses are discontinuous at the towers, as they are not meant as primary supports, but rather to stiffen the deck. Given the talent and experience that went into the original the design, the big question is: what went wrong?
The short version is that there are many different possible analyses for a given structure, but they are not all equally easy to perform and they are not all equally useful. The most basic is stress under gravity load, because if that isn’t okay, the structure likely collapses. Stress under wind load is of similar importance. Stress under seismic load is something that we now consider to be as important as wind (in areas where any appreciable level of seismicity exists) but was not realistic until the middle third of the twentieth century. Static deflection – how much the structure will move – under the various loads is important if you want the structure to be usable and to prevent damage from excessive movement. And questions of dynamic deflection – how fast the structure will move, at what frequency it will vibrate, and so on – are much more recent. I’ve listed these issues in (roughly) increasing order of difficulty and decreasing amount of use. Building structures never suffered from a lack of dynamic analysis until the 1960s, because the buildings were so heavy and relatively so rigid that motion was quickly damped out. As glass and lightweight-panel curtain walls replaced masonry, and floor spans became longer in the 1960s, buildings began to suffer from dynamic problems. There’s a whole class of 1960s and 70s high-rises that had to be stiffened or otherwise modified because of dynamic sway issues.
Suspension bridges are pretty much entirely structure, with not enough mass to provide much damping. This has been known since the beginning of modern (metal) suspension bridges in the nineteenth century: vibrations and rhythmic sway from wind could bring a bridge down. The problem was that, using the analytic tools available at that time, there was no way to be sure what was needed. Adding mass to the deck helped, but was not necessarily enough. Stiffening trusses in the same plane as the cables helped by distributing local loads across several suspenders: if a portion of deck started to move excessively, the trusses forced it to drag the neighboring areas along for the ride. John Roebling pushed the idea of using diagonal stays, effectively making his bridges hybrid suspension/stayed bridges, and it worked quite well. But the stays never really caught on with other engineers. Plate girders began to be used instead of trusses in the 1900s, because steel material was getting relatively less expensive, so the big plates required became a reasonable alternate to trusses. The girders were as strong as the trusses they replaced but not as stiff; their solid webs also caught the wind better. The more advanced analytic tools of the 1930s were still forms of static analysis, so the flaw here was not obvious in modeling. It became obvious in reality: the first generation of bridges with plate-girder stiffeners were known to be more “live” than their predecessors.
The use of computer modeling allows for math-heavy analysis to be realistically performed on complicated structures in a way that had not previously been possible. But even before computer use, there were ways to address the problem: following the 1940 collapse (from wind-induced dynamic movement) of the Tacoma Narrows Bridge, deck stiffening once again took center stage. The George Washington Bridge was stiffened by having a second deck and large trusses added, the Golden Gate had its trusses connected below the deck to improve torsional stiffness, and the Whitestone got those big trusses.
The old tools and old methods worked for bridges with main spans of maybe 2000 feet and deep stiffening trusses. It turned out that they did not for spans of 3000 feet and shallow plate-girder stiffeners. Ditto for sideway in high-rises with 25-foot girder spans, terra cotta interior partitions, and masonry exterior walls compared to those with 40-foot girders spans, gypsum-board interior partitions, and glass exterior walls. In other words, the efficacy of an analytic tool or a design method is situational, and has to be re-examined every so often. How often depends on how fast designers are pushing the boundaries.
