Risk and uncertainty: How safe is safe enough?

Chapter 10 Case Study: Genoa Viaduct Collapse and Responding to Risk

The New York Times ran an article in October 2018 over a catastrophic bridge collapse that killed a few dozen people in Genoa, Italy, interviewing both victims and engineers. The Genoa viaduct was designed by Riccardo Morandi and had a light design that made left very little margin for error. The bridge consisted of three A-frame towers paired with just 12 stays, which suspend the road from the towers. The design was praised for its elegant beauty, but the neat and trim design came with a cost. Critics pointed out that there was no redundancy—no possibility of redistribution of the load should one part fail. In other words, if one part of the bridge failed, it would likely lead to a total collapse. Andrew Herrmann, a structural engineer who is president of the American Society of Civil Engineers, said simple that “If you lose one stay the whole thing comes down.” Most cable-stayed bridges today have numerous stays so that the loss of one does not lead to collapse.

In addition to the small number of stays, the structure of the stays created a special problem on its own. The steel cables were encased in prestressed concrete. While this was intended to reduce stay, it also had the consequence of making structural deterioration of the cables difficult to monitor. As it happened, the concrete did not protect the steel from rain, and it was known in the 1970s when the bridge was still new that water was already corroding the steel. The easternmost tower required maintenance in the 1990s, and as the steel was repaired, the concrete was jettisoned for a protective sheath. In 1999, a private company, Autostrade, took over maintenance of the bridge from the government during a budget crunch. Autostrade did not make the same refurbishments to the other two towers. Doing so would have replaced bad steel and then protected the new steel from water while making it easier to identify corrosion that does occur.

In October of 2017, Autostrade asked Carmelo Gentile to test for corrosion hidden away in the concrete encased stays of the other towers. Gentile uses an unusual method for diagnosing problem in the steel of bridges: He listens to them. Professor Gentile has performed his tests on over 300 bridges around the world. He tests the vibrations that pass through the steel cables and listens to their frequencies. A smooth wave indicates that the steel is continuous and in good condition; when the sound does not display such consistency, that indicates corrosion and a lack of structural soundness. Professor Gentile reported the findings and recommended that sensors be installed to monitor for corrosion. Autostrade admitted to being briefed on the concerns but claimed that it did not receive a warning of urgency. Autostrade did make plans to perform maintenance on the bridge to address its deteriorating condition, but never scheduled it—the bridge collapsed first.

On August 14, 2018, the stays on the south side of the bridge broke. This is all that was supporting the road on that side, and so the entire road began to rotate at an angle toward the south. Connecting pieces of road between that held up by the stays were then disconnected from the part of the road tilting south and collapse. Now the remaining roadway supported by the stays on the north snapped those stays and collapsed. The tower itself would collapse onto the rubble. More than 30 cars were on the bridge during the catastrophic failure, and 43 people were killed.

The bridge collapse was preventable had Autostrade acted sooner to fix it. But the costs of repair were extremely expensive, and there was talk of simply replacing the bridge rather than continually fixing it. While there was plenty of evidence of structural deterioration, there was no exact assessment of the likelihood of structural failure. One engineer the New York Times interviewed, Gary Klein, explained that ascertaining the condition of internal cables is extremely difficult and imprecise.

Given the evidence that there was structural damage and the inability to precisely identify the risk, what sort of risk approach should Autostrade have taken?

Case study by Robert Reed


Back to top