I-35W Bridge Collapse

The busy Interstate 35W bridge over the Mississippi River in downtown Minneapolis, Minnesota, collapsed at the peak of rush hour on Wednesday, August 1, 2007. The investigation into the cause of the disaster took 16 months, with the final report being published on December 19, 2008.

The NTSB concluded that the root cause of the bridge collapse was a design error that resulted in gusset plates at 4 locations on the bridge being too small to carry the load that was placed on the bridge. This design flaw resulted in a structure that was doomed from day one.

Note—these materials are covered by copyright and may not be used without written permission.

The photo above is a view of Bohemian Flats, located on the west bank of the Mississippi River near the University of Minnesota campus. The NTSB reassembled key sections of the bridge in the parking lot of what normally is a river cruise operation. Other interesting parts of the bridge were also stored at this location, with less important pieces being moved to a MN-DOT gravel pit in Afton, MN. A few pieces were sent to Washington, DC, for materials testings. The steel remains sitting in this parking lot two years after the disaster as the lawsuits slowly move through the court system.

The Washington Avenue bridge is visible in the background. In an ironic twist of fate, shortly after the I-35W bridge collapse, it was discovered that the steel supporting the upper deck on this bridge was built with too light of material, and the bridge was at risk of collapsing under its own weight.

The photos above and below show bridge beams laid out for study as part of the bridge forensic reconstruction. The truss sections nearest the south bridge piers were the most heavily studied sections after it was learned that the collapse started in this section of the structure.

The photos above and below are two more sections of the main truss that was reconstructed by the NTSB. These sections are pieces of the truss that were vertical, and runs from the pier along the edge of the river towards the middle of the bridge. The bridge deck would have been on the straight edge on the outside of this pair of truss sections, with the curved part that faced the river being next to each other on the inside. These two pieces would have been lifted from the outside, folded up parallel, and cross-braced when the bridge was still standing.

The truss section members towards the back of the photo were once vertical and those members were attached to the bearings that were attached to the bridge piers. The U10 joints that failed would have been along the straight edge that once supported the deck, on the outsides of this pair of truss sections, about halfway along these sections. The actual pieces that failed are not present in this layout. I suspect that these pieces are among those that were taken to Washington, DC, for further analysis.

Note that in the second and third photos, those truss sections are flipped compared to these two photos, with the member above the piers being in the front of the photo, and the parts in the back of the photo heading towards the middle of the span.

These two photos show the steel laid out in the rear of the reconstruction yard. These are parts from the middle of the river span. The large connected pieces are the sides of the main truss where it arches over the center of the Mississippi River. The smaller triangle parts in the front of the photo below are the cross-braces that connected the two faces of the truss and kept the truss members parallel where they could carry the most load.

The pyramid shaped pieces of metal in the front of the photo above are the mounts that once sat on top of the concrete piers. These metal pieces supported the bearings, which supported the bridge. Bearings were supposed to allow the bridge to move as it expanded and contracted with temperature changes. The NTSB determined that several of the bearings showed no signs of moving, yet were not frozen in place. The result is that the bridge was flexing, and it did not slide on its bearings as it designed to do.

The NTSB considered a number of other factors in the bridge collapse. The trigger appears to have been 6 loads of rock and sand that were placed on the bridge at about 2:30 PM on the day of the collapse. These materials were being staged as part of a concrete pour that would begin at 7:00 PM. The concrete pour was part of a project to fix pot holes in the bridge deck and install a 2 inch pavement overlay. The materials, water trucks, and other construction equipment were positioned directly over the U10 node on the south end of the bridge.

These two photos are wide-angle views of the reconstructed truss sections. It gives an overview of the amount of work that was done to sort out the pieces, and shows how the yard was organized.

The NTSB did considerable computer modeling of the bridge. In this work, they discovered that the bridge would likely still be standing today if the U10 gusset plates were built with 1-inch think material rather than 1/2-inch steel. This finding lead to a number of conclusions. First, previous modeling did not consider the strength of the gusset plates, so the flaw was missed in previous computer models. Second, a design review prior to construction also missed this error, so the design review process needs to be improved. Third, MN-DOT had no policy for storing construction materials on a bridge deck. In the past, it was up to the opinion of the project bridge inspector, and was not based on engineering calculations.

Even more startling is that the I-35W bridge was modified in previous years to add lanes, an anti-ice system, and a thicker deck. None of these projects included a review of the bridge structure to determine if it could support the additional weight with a reasonable safety margin. This challenges the past mentality that once a bridge is built, you can do anything you want to it without concern for the structure.

The photo above is an overview of the south end of the yard. In this area, the NTSB reconstructed sections of the bridge deck support members. There was a theory that the bridge deck had pulled apart as the first step in the bridge failure, but the wreckage did not support that theory. The bridge to the left is the Washington Avenue Bridge. The concrete structure on the river bluff is the opening to the elevator for the Anderson Library, which is located deep underground. The photo below is an overview of an area that contains a large number of cross-braces.

The NTSB also ruled out a number of factors. There were no issues with the piers. The piers that were pushed sideways were pushed during the collapse. Temperature was not a factor. While it was unusually hot that day, additional forces in the bridge due to heat were very small. Traffic load was also not an issue. The bridge often had far more traffic on it when all 8 lanes were in use. The fact that traffic was using the outside lanes was considered, but there is no evidence in the wreckage that the bridge tipped sideways. Some people speculated that the construction project was the cause. Up until the excess weight was put on top of the U10 node, there was no indication that any of the construction activities caused any damage to the bridge.

The photo above is a close view of the storage area for the random cross-braces that once connected the two truss sections.

The photo below is an overview of the yard showing part of the skyline in the background. The buildings in the foreground is a city housing project. The building in the back to the left is a University of Minnesota administration building, while the building in the back to the right is a condominium tower located along 19th Avenue South.

After 16 months of waiting, we have the final verdict. The I-35W Mississippi River Bridge was built with a design flaw where a particular set of gusset plates in the U10 nodes were too thin. When excess weight was placed on the bridge over one of those U10 nodes, that node failed within 3 hours and 30 minutes. The non-redundant nature of the design saw the excess loads transferred to the other U10 nodes, which all failed, resulting in the truss failing and the bridge span dropping into the river.

The photo below is the river flats area as seen from the Northern Pacific Bridge #9 in early 2012. MN-DOT was afraid to move the bridge remains because these steel beams could be evidence in future lawsuits, and disturbing the beams any further could make the state liable for destruction of that evidence. MN-DOT finally moved the beams to a site in Afton in late 2010. The park was cleaned up and restored in early 2011. The tour boat operation did not return, which had been generating rents of $20,000 and parking fees of $16,000 per year. The city was compensated for these losses and restoration costs.

On November 15, 2012, the Minneapolis StarTribune reported that the State of Minnesota settled the last lawsuit related to the failed I-35W bridge. Jacobs Engineering agreed to pay the state $8.9-million. Jacobs was not directly responsible, but they are a successor company to the firm that did design the bridge. Jacobs argued that the statue of limitations had long since expired, but the state nullified the statue of limitations for this case, and the US Supreme Court refused to hear Jacob's challenge. In 2010, URS Corporation, the firm that inspected the bridge prior to its collapse, paid $52-million to the state.