Refurbishment of the viaduct at Kinzua was about halfway complete in 2003. It had stood for 100 years, but its time was up when a tornado swung round from an unusual direction.
Gene ComossWhen the structure was designed to withstand the prevailing winds, the design provided for the towers on the west side of the structure to be fixed with anchor bolts into the pedestal bases.
On the... east side... the towers were fitted with roller bearings to allow for some movement either from wind load or thermal expansion.
An intensive system of severe weather is moving into Pennsylvania from Ohio. There are reports that this system is packing tornados, among winds of well over 50 miles per hour, the system is tracking North by North West.
Gene ComossWhen the tornado struck the bridge repairs were approximately 50% complete. As a result of the tornado striking the bridge... the Department convened a board of inquiry to investigate the cause of the failure.
I led the board of inquiry investigation. Our first task was to assemble a professional team. Our task consisted of forensic engineers from my firm, Gannett Fleming. We also engaged the Meteorological Department of Pennsylvania State University and the fracture, uh, engineers of Atlas Lehigh University.
In the course of the investigation, we looked for forensic markers. What we disclosed at the site was four specific forensic markers. The four markers we found were order markers.
Order markers are looking for clues of what member is on top of what member. We looked for directional markers; directional markers such as the orientation of the trees that collapsed, the direction of the debris field.
We noticed separation markers, clean separation of the superstructure from the substructure. And finally we looked for fracture markers; telltale signs of cracking in members that may have precipitated the collapse.
Finding forensic markers like these is a routine way of investigating large structural failures. There was, however, another piece of evidence that was nearly missed.
One important part of our investigation was making good use of the high resolution aerial photography. As we were trying to determine precisely the sequence of collapse, I was viewing on my computer screen the high resolution photography and I noticed, quite suddenly, large skid marks.
These are the marks where the towers actually had gone airborne, slid down the hill and made large depressions in the ground. These depressions were unnoticed by any of us who had wandered all over the site during our day of our board of inquiry investigation. These were crucial in determining the precise sequence of collapse
The testimony of workers at the bridge site was important in determining the sequence of events as the tornado struck. As they were retreating at the time the tornado hit they could only report on what they heard.
Some employees had no idea anything had happened. This gentleman here only talks about raining and thunder and lightning, uh, but Mr Quillin’s report, in particular, is…has some more information.
... He says he had sent the crew home because of weather and he was leaving the site when the winds picked up, and he had a crew member with him, and they were leaving the site “when I heard four to five loud booms. I wasn’t sure what the booming was” ... but at that time he went and determined that part of the bridge had actually fallen down.
We determined the sequence of events for the collapse in three distinct episodes. Let me describe those episodes.
During the first episode, as the tornado touched down and moved in a northerly direction, leading edge winds affected the structure.
The first group of towers occurred a separation failure at their base and were displaced a little to the west and were held momentarily in an upright... position by the rails that were still attached to the towers.
The next group of towers as the tornado moved northward were affected by the leading edge winds, separated at the base and toppled in a westerly direction, followed by the northern most section of towers which similarly fell in a westerly direction.
Finally, as the inflow winds attacked the structure, the southerly group of towers then collapsed in a northerly direction.
The booms which were heard were almost certainly the sound of each group of towers hitting the ground. But the bridge was designed to resist lateral tornado-strength winds of up to 100 miles per hour, so what caused the towers to collapse?
Gene CosmossIn the structural design of the towers, the load was transmitted down through the tower legs to the pedestal bases and the tower legs were fastened to the pedestals by ... use of an anchor bolt that was embedded in the pedestals and then bolted to the tower legs.
In the pedestal bases higher up on the flanks of the gorge the original base bolts from the 1882 bridge were of sufficient length to bolt the towers down. But this wasn’t so down at the bottom of the gorge, where the towers failed, where the bolts had been modified.
Gene CosmossIn the towers that failed, anchor bolts similar to this anchor bolt was the principal cause of failure. ... The reason for the failure was the method, the designers of the 1900 reconstruction chose to fasten the towers to the, uh, 1882 pedestals.
In their fastening, they extended the anchor bolt by approximately six inches and then fastened that six inch extension to the original anchor bolts by use of shims and a coupling.
So the weakest link in the load path occurred where the bolts were lengthened.
To the original bolt embedded in the base, spacing shims were added and then a wrought iron collar, or coupling, was attached. Into this the additional length of bolt, required for tying-down the new towers, was screwed into place.
These joints were unprotected, water percolated into the collars leading to corrosion fatigue as the rust expanded – resulting in vertical cracking.
Evidence recorded by the inquiry, shows the degree of the corrosion suffered by the collars.
Cracks like this weren’t safety critical because the base of a tower will have been held down by gravity but they became the weakest link with very high lateral wind-loads.
The forensic investigation had established the sequence of events that took place when the tornado descended on Kinzua. The third and final part of the video uses the investigation’s findings to examine another, much earlier, disaster.