Thursday, November 7, 2019

NTSB results concerning the FIU Pedestrian Bridge Collapse

I have already researched the FIU bridge collapse. These notes discuss the NTSB results.

Video: NTSB Board Meeting: Pedestrian Bridge Collapse Over SW 8th Street, Miami, Florida

Except for the amendment procedures, once I got into the staff presenting their slides, I found this video to be more interesting than I would have guessed.

Five alphabet soup organizations watched cracks quickly (just a few days) grow to 40 times an acceptable width. All of five had the authority to stop traffic under the bridge and order shoring to be placed under the bridge. But none of them did. As the NTSB chair observed, the bridge was "screaming" that it was going to fall.

Multiple design errors were made such as treating the truss as path redundant when it was not, And an unqualified reviewer was chosen. The FDOT web site had erroneously listed that reviewer as qualified. The materials used in construction were fine and the construction followed the plan correctly. The problem was the plan (design).

The cracking and ultimate failure was in the 11-12 node on the north side where an additional span would have been added.
Slide
The designers, FIGG, created four analytical design models. But they chose the model that had the lowest demand (load) force. To be conservative, they should have used the highest demand force. During the investigation, they were not able to explain this choice. I'll observe that if you are pulling a design out of thin air, you should choose the most conservative design.
Slide
Evidently NTSB hired FHWA to do an independent design. If FIGG had chosen their conservative design, they would have been in the same ball park as the FHWA design.
Slide
Unfortunately, they chose the design that underestimated the demand force in the outer nodes by about half. And the post-tensioning action taken in the "hope" of stopping the cracking was a design change that was not peer-reviewed. I say "hope" because nobody could explain why node 11-12 was cracking. Concrete cracking was heard while they were still removing falsework in the casting yard. And it continued to dramatically crack after the move. The NTSB determined that the move itself did not add to the damage of the bridge. The tensioning procedure was done well, but it increased the demand force that was already excessive and triggered the collapse. The "fix" was like throwing kerosene on a burning fire. Since no one stopped the traffic on the road while the emergency fix was being applied, people got killed.

I don't understand the details of the capacity (support) load design errors. But the bottom line seems to be that they designed with a factor of 1.25 instead of  the conservative factor of 0.90.
Slide
So the design had a double-whammy concerning the shear force at node 11-12: the bridge was both more heavy and more weak than assumed in the design. In fact, the design had 5 sq. in. of rebar to resist the shear force when it should have had 18 sq. in! The result of the insufficient shear resistance was that the concrete blew out of the north end. Since there was no path redundancy, that failure propagated to a total collapse.
Slide
This was the first attempt to build a bridge with a concrete truss. Some of the NTSB recommendations are that bridge design guidelines be augmented to provide guidance for concrete trusses if they are going to be used in practice.

A problem was identified with the cold joints between the truss members and the deck. Specifically, the roughening of the joints did not conform to specifications. In fact, there was inconsistency in the applicable specifications. The most definitive spec stipulated that the concrete surface should have been roughened with a 0.25" amplitude. The surfaces has some roughness because they were not finished (troweled). But I gather some sort of chiseling is needed to achieve 0.25" indentations in the concrete. But this is a red herring issue because it was determined that even if this spec was followed, the joint would not have been strong enough to compensate for the design errors.

Another red herring was that there were pipes in the node. If the node had been built with 18 sq. in. steel across the shear plane, the weaker concrete load paths would not matter. Because there was only 5 sq. in. of shear resistance, the pipes determined where the node would fail. They effectively supplied "score lines" for the failure. The analogy in my mind is a chain. Assume every link is more than strong enough to hold the design load. But if an excessive load is applied, the chain will fail at the weakest link. But the link is not at fault, the excessive load is the fault. In this case, the pipes are not at fault, the shear force far exceeding the rebar design is at fault. (Although I never did figure out what the purpose of the four vertical pipes was.)
Slide
Some notes I made as I watched it:
5 vs 18 steel sq. in. of steel reinforcement:  https://youtu.be/fdUf-_el9vA?t=7993
probable cause:  https://youtu.be/fdUf-_el9vA?t=11023

The irony of having several organizations responsible for safety is that it seems you effectively have no one responsible.



No comments:

Post a Comment