The 1934 concrete-truss McMillin Bridge in Washington state.
I’ve written a number of posts on the topic of incompatibilities between different aspects of buildings and they’ve mostly been about nuisances. This one is different and was triggered by my reading the OSHA report on the collapse of the FIU footbridge in Miami during its construction. This collapse, on March 15 last year, killed six people and injured a number of others.
The report is thorough, well-written, and on-topic, so I see no need to discuss its findings. Read it yourself. As is true in the majority of structural failures, there was more than one cause that led to the final result. What caught my eye, in addition to the discussion of how the failure occurred, was the topic I started on above: a basic mismatch between the structural form of the bridge and the material used for its construction.
Putting aside anthropomorphism (“what does this structure want to be?”), all materials are better suited to some forms of structure. The use of a truss structure for the FIU bridge was fine, maybe even the preferred form given the span and the goal of creating a signature element for the campus. Trusses in general have members in both tension and compression, and Warren trusses* usually have load reversals in each web diagonal. Reinforced concrete as a material performs best in compression and worst in tension, which is why concrete trusses are rare. (The McMillin Bridge, pictured above, is both a rare example of a concrete truss and is pretty enough that it deserves to be seen.)
The concrete truss was the main load-carrying element of the bridge, with the pylon and diagonals above (which would have given it the appearance of a cable-stayed bridge) mostly for show. The use of a single truss down the centerline of the bridge is another rare but not unique design decision. Finally, because the main span was constructed nearby and then moved to position** there were several different possible load and span conditions for the bridge, requiring among other things, the post-tensioning and then un-post-tensioning of the main deck. A lot of the design and construction sequence was worked around controlling tension in a brittle material so as to ensure ductile behavior. You know what’s easier than that? Using a ductile material.
The truss form used had no redundancy. The failure of any one joint or any one member would lead immediately to overall failure. That’s not necessarily true of all trusses. Double-diagonal Warren trusses, for example, can usually survive the failure of one member. More importantly, the use of concrete meant that ductile failure was only possible if the stresses are in a pre-determined location and direction: reinforced concrete behaves in a ductile manner when there is reinforcing running in the right direction. Steel behaves in a ductile manner most*** of the time. So steel trusses with continuous top and bottom chords have some**** accidental redundancy from yield failure at the nodes.
The report mentions that the engineers who wrote the scope for the design/build RFP suggested using steel. The report mentions the changes in analysis based on four different stages in the construction sequence. The report mentions the use of a non-redundant structure. None of these design issues directly led to the failure – a single concrete truss built off-site could have worked safely – but they all point in the direction of a mismatch between the structural form (the expected structural action) and the material used to achieve it.
* The main structure of the FIU bridge was a Warren truss, despite its unique geometry.
** Similar to the construction sequence I described for the West Thames Street bridge.
*** Unless it’s overly constrained or loaded to full stress across thick, less homogenous flanges.
**** Maybe not much, but more than zero.