Over the last five parts of our Miami Bridge collapse series, we’ve explored the events surrounding the failure of the Florida International University (FIU) bridge on 15 March 2018.
In this part of the series, we’ll try to make sense of this tragedy, not only from a technical perspective, but also from the perspective of how failure investigations should be undertaken in the future.
But to begin, let’s step back and recap our story, starting with the bridge’s design.
The Technical Causes
In Parts 4 and 5 of this series, we covered the bridge’s extensive design errors: the actions in the critical Member 11 and 12 to deck joint were higher than the designers assumed and the strength of this key joint was lower than the designers calculated. This under-designed connection would ultimately go on to play a key role in the failure.
Then came the peer review process. As discussed in Part 3, the Florida Department of Transport (FDOT) required a peer review of the bridge’s design by an independent third party. This was to ensure the bridge was designed correctly. However, this peer review failed to check the design of the bridge’s connections, which allowed this under-designed connection to make its way through to the construction phase of the project.
Cracks in the bridge then became evident at an early stage in construction and continued to grow as the project progressed. As we know from Part 2, despite clear and documented evidence of cracking – even up to the morning of the failure – the parties didn’t appear to consider it to be a safety concern, and didn’t stop construction.
Instead, the design engineers, FIGG, made the decision to restore the post-tensioning that had been released in Member 11. Their logic appears to have been that because releasing the post-tensioning earlier in construction caused the cracking to get worse, restoring it would make things better. (In other words, applying a compressive load to the Member 11 to 12 and deck joint would close up the large cracks.)
The NTSB investigation, however, would conclude this simply wasn’t a solution. There was no way such a badly cracked section of the bridge could be restored to behave like a solid block of concrete again just by compressing it together. Instead, by applying a significant force to a badly cracked section, they ran the risk of making the joint explode.
As this post-tensioning force was being applied, workers were standing on top of the bridge and vehicles were traveling freely beneath it. And it was during this re-tensioning that the stresses in the Member 11 and 12 to deck joint become too much for the connection. It catastrophically failed, leading to the collapse of the bridge and the deaths of 6 people.
A Different Question
What are the major takeaways from this failure?
Well, the technical lessons are reasonably obvious: multiple errors in the bridge’s design were made, the formal review process failed to identify them, and when warning signs of failure became glaringly apparent, in the form of cracking, no one halted the project.
While this is what happened, the investigation sheds little light on why it happened.
For example, why did FIGG calculate the correct action in the Member 11 and 12 to deck joint, but then use a lower value for this action in their design? And why didn’t those involved in the project stop construction when the bridge’s cracking became so bad?
We don’t know the answers to these questions, but in many ways, they are not the most useful questions to ask. Instead, rather than asking why these decisions were made, it’s better to ask why the processes we have to prevent these poor decisions failed.
After all, for every technical issue that culminates in failure, there are always a range of systemic issues that allow these technical issues to make their way through the entire design and construction phases of a project, despite our safeguards. And in order to answer these types of questions, we need to understand the organisational causes of the failure.
Now, some of these organisational causes of the bridge’s collapse are readily apparent from the NTSB investigation, with the peer review process being one of them: why did the peer reviewers fail to check the connections? Well, as discussed in Part 3, Louis Berger won the work based on a greatly reduced fee and timeframe, and on that basis, they appear to have made the decision not to check the design of the bridge’s connections.
But this is hardly a new finding: we’ve always known from engineering failures that when time and budget pressures are high, rational engineering decisions are not always made. The peer review process should have identified the design errors, but not checking the connections meant this opportunity was lost.
But while we may understand the reasons why the peer review failed, we have very little insight into why other key aspects of the failure occurred. For example, take the design errors. We know the management of human error is a constant battle in engineering design, but why were these errors actually made?
And the answer is we don’t know.
And what about the warning signs? Significant cracking appeared in the bridge, as we discussed in Part 2, and these cracks grew throughout the bridge’s short life – with some of them reaching more than 40 times the permissible crack width limits. They did generate plenty of discussion, emails, and concern, but this didn’t translate into anyone taking action to actually stop the work. Why was this the case?
And the answer, again, is that we don’t know.
So what does this failure really teach us?
And how should we approach the investigation of such failures in the future?
Firstly, it teaches us that while we engineers may pride ourselves on being rational human beings that follow sensible processes put in place to manage risk and respond to warning signs of failure, we’re just as susceptible as anyone else to making mistakes. We need systems that are robust enough to recognise this, manage this, and withstand the time and costs pressures that the world of design and construction inevitably brings.
Secondly, when failures occur, we need to investigate them in such a way that allows us to understand how these very systems failed. In other words, getting to the bottom of the organisational causes of a collapse are just as important as getting to the technical causes.
The NTSB investigation of the FIU bridge collapse leaves many important organisational questions unanswered as to why these systems failed. Questions that, without adequate answers, leave us ill-equipped to glean the critical lessons we need to learn from this failure.
For me, the most important learning from this tragic collapse is this: if we really want to learn as a profession, we need investigations that focus on the organisational causes as much as the technical ones.
If you are interested in a more detailed discussion of this failure then be sure and listen to Episodes 34 – 36 of the Brady Heywood Podcast.
Photo: Miami-Dade Fire Rescue