Federal investigators are trying to determine whether 91-degree heat caused expansion that put too much pressure on the corroded gusset plates that held the I-35W span together.
Ten weeks into their probe of the Interstate 35W bridge collapse, National Transportation Safety Board investigators have intensified their inspection of a long-corroded gusset plate that was located in the section of the bridge that fell first.
In addition, authorities are analyzing what role the 91-degree heat on Aug. 1 might have played in increasing stress on the already-weakened L-11 gusset plate, which connected four steel beams located near the bridge's south end.
In 1993, a state inspector found that the half-inch gusset plate had lost nearly half of its thickness in some spots due to corrosion along an 18-inch line, but no repairs were ordered, according to Minnesota Department of Transportation records.
Two structural engineers who have viewed the wreckage said in interviews this week that the L-11 gusset plate is one of three closely situated connections that could hold the secret of what caused the bridge to collapse. All three joints appear to have been damaged by some primary force -- not from secondary impacts sustained during the collapse, the engineers said. They spoke to the Star Tribune on the condition that their identities would not be disclosed. NTSB officials declined to be interviewed.
The three damaged gusset plate connections can be clearly seen lying along the secured West River Road site where federal experts have laid out the bridge parts most critical to their investigation. Earlier this week, two officials spent hours examining and photographing the L-11 connection and the beams that were once riveted to that corroded plate.
Within a week of the bridge disaster, NTSB Chairman Mark Rosenker announced that stress on gusset plates may have been a factor in the collapse. A month later, he said that "a failure in one of these plates could have catastrophic consequences." Now it appears that of more than 100 gusset plates on the bridge, the NTSB is focused on just a few.
One of the structural engineers who has viewed the wreckage and knows the design of the bridge said runoff of salt and de-icing chemicals from the bridge deck could have contributed to the corrosion in the L-11 gusset-plate connection. That's because a diagonal, H-shaped beam running into the joint could have acted to channel the liquid toward the gusset plate, the engineer said.
Also, the engineer said that two of the three damaged gusset plates that appear to be of primary interest to the NTSB are half an inch thick. The thickness of gusset plates used in the bridge varied between half an inch and 1 inch. That could be an important issue, because a consulting firm hired by the state has said that some half-inch gusset plates may not have been strong enough to hold the bridge up.
According to MnDOT documents, state bridge engineers worried for years about fatigue cracking in the heaviest beams of the steel superstructure. As recently as January 2007, MnDOT engineers were planning to reinforce the beams that were deemed by a consultant to be most critical to ensuring the bridge's safety. But MnDOT opted instead to closely inspect the beams every year. When the bridge collapsed, inspectors had completed only half of their 2007 inspection of the fracture-critical beams.
Both structural engineers interviewed by the Star Tribune said the issue of fatigue cracking is not currently at the forefront of the NTSB's investigation. Rather, they say federal authorities are examining whether intense heat on Aug. 1 triggered a chain reaction of force that overpowered gusset plates in crucial locations -- such as the one at L-11.
Flexing was planned
The I-35W bridge was designed to flex, to handle expansion in extreme heat and contraction in bitter cold. But that design assumed that roller bearings would move accordingly.
Roller bearings are mounted on top of bridge piers. They support the weight of the bridge and contain steel cylinders that allow the bridge to roll smoothly back and forth as it expands and contracts with temperature changes.
MnDOT officials had long known from their inspections that the bearings were not working correctly because of corrosion and buildup of debris. And in July 2006, a consultant's report highlighted the problem. "The bearings are not allowing the structure to move linearly with changes in the ... temperature," the report said.
Roller bearings recovered from the river are stacked neatly in the NTSB's secure area for inspection. One of the structural engineers who asked not to be named said the bearings appear to be lacking marks of wear, indicating they may have been locked up or their movement restricted by debris and corrosion.
The NTSB has provided occasional updates on its investigation, but the agency has said repeatedly that it could take 12 to 18 months to reveal a probable cause for the disaster, which killed 13 people and injured 144 others. While critical bridge components are expected to be laid out on the riverside until at least November, the NTSB has been shipping less important pieces of the wreckage to a gravel pit in Afton owned by the state.