Normal Accidents: I-35W Bridge, Minneapolis, Minnesota
Over the past two days everyone has been talking about the Interstate freeway bridge collapse in my hometown of Minneapolis, Minnesota. While the National Transportation Safety Board (NTSB) will be investigating this disaster and it’s expected that we won’t see the investigation concluded for at least a year, it’s interesting to speculate already on what factors contributed to the bridge’s collapse. In a number of reports over the past decade, including a March 2001 study for the Minnesota Department of Transportation conducted by the Department of Civil Engineering at the University of Minnesota, questions have been raised about the structural integrity of the bridge. In light of these reports, what we can expect to see in the coming days is any number of claims that the bridge obviously should have been rebuilt. But any experienced civil engineer knows that’s not the way things work, as eloquently described in this editorial in the NYT by Samuel Schwartz, who was the NYC Department of Transportation’s chief engineer from 1986 to 1990. In an era of tightened budgets and lack of political will for dealing with the prosaic elements of day to day life in modern technological cultures, infrastructure projects rarely get funded except in the face of a crisis like this. And so engineers have to look for other approaches.
In the end metal fatigue, which has been implicated in the reports and even in the most casual conversations with engineers I know who are familiar with fracture mechanics, is surely going to be one of the key factors. One of the fundamental strategies for managing fatigue is to inspect rather than repair — good engineering practice involves designing the product so that the so-called critical crack length is a size that can be readily seen by the naked eye. In fact, this is a core feature of any aircraft design since the days of the de Haviland Comet, which was one of the keystone design debacles that contributed to the field of fracture mechanics and was the driver that made this design strategy universal in the aerospace industry. What aerospace engineers do is very carefully identify the points where the key crack formation is critical to the integrity of the overall structure. They also design in many mechanisms for arresting crack propagation. (This in fact is why there are so many rivets in airplanes and why the process of making a rivet is so very carefully managed and circumscribed — as much as possible, each of those little circular holes is a crack arrestor.) In the case of the I-35W bridge and the myriad other bridges that have been classified as “structurally deficient” by the Bureau of Transportation (see stats), the strategy has been to secure budgets for routine inspection (as was the case with the I-35W bridge) rather than to ask for funds for the kind of massive infrastructure redevelopment that needs to be done. Unfortunately, the I35-W bridge is an early form of a platform on truss design, which the University of Minnesota clearly asserts has only one level of redundancy – the absence of which means that a crack anywhere has the potential to be catastrophic for the entire structure. The problem, however, is that visual inspection of every square inch of a structure of this size is simply impractical. Any engineer watching the video of the bridge collapse that has been made available would find it hard to believe that a critical crack propagation is not core to the failure mode of this bridge. (I happened to be at the Massachusetts Institute of Technology when the video was released yesterday morning and the materials science researchers I was with all found it hard to believe that a structure of that size would fall so quickly for anything less than a fatigue failure.) What do you think? Watch the video and comment below. Of course, as we have learned in virtually every major failure of a large-scale engineering structure (the Challenger Space Shuttle, the New Orleans levees), the technical basis for the failure is frequently only the primary (albeit catastrophic) symptom of the actual reasons for the failure of the system in question. (See Charles Perrow’s Normal Accidents, whose title forms the basis for this entry’s lede.) We will undoubtedly find that there are economic, institutional and political factors that led to this technical vulnerability in the bridge and the ultimate resolution to this problem is going to depend on far more careful strategies than simply saying “make it a bit thicker here.”