Remarks of Honorable Mark V. Rosenker, Acting Chairman
National Transportation Safety Board
Investigation of the I-35W Bridge Collapse
Minneapolis, Minnesota, August 1, 2007
Transportation Committee, National Conference of State Legislatures
December 12, 2008
Good morning Senator Dibble and Committee Members. I am very happy to be here with you today to talk about one of the largest and most important highway accident investigations in the Safety Board’s history, the collapse of the I-35W bridge in Minneapolis, Minnesota. I am extremely pleased that you have invited me to present details about our investigation – so we hopefully won’t have to relive this tragedy again.
While the Safety Board is most known for its investigations of aviation and other major transportation crashes, we are charged by Congress to investigate accidents of significant safety problems in all modes of transportation.
On August 1, 2007, all of us were shocked to see the images on our televisions – the remains of an Interstate highway bridge that had collapsed during rush hour in one of our nation’s major metropolitan areas. I would like to take a moment to show several pictures taken shortly after the collapse.
Tragically, 13 people lost their lives and nearly 150 others were injured when the I-35W bridge plummeted into the Mississippi River. Understandably, the accident raised many questions in the minds of Americans about the condition of our nation’s highway infrastructure.
In the 16 months since the accident, there has been much speculation about the causes of the collapse, and much discussion among policymakers and others about its larger policy implications. Indeed, I am sure many people in this room have had to face tough questions about the condition of bridges in your States. It is for this reason that I want to explain our investigation findings in some detail so that you fully understand the lessons that this tragedy does, and does not, hold for policymakers.
The investigation the Safety Board launched the night the I-35W bridge collapsed was to involve the cooperation of Federal, State and local authorities. It would take more than three months just to recover the mangled wreckage of the 1,900-foot-long, 8-lane bridge where it collapsed into the Mississippi. Thousands of tons of steel were removed from the river and laid out for detailed inspection at the Bohemian Flats Park down river from the collapse site. This picture of the bridge structure laid out in the park gives some account of the details of this investigation. Eventually this investigation would include visits to 14 other States to analyze the nature of their bridge assessment programs. It would involve the development of complex computer models of the entire bridge and of key gusset plate connections. And just to make sure an extra set of eyes looked at what we were looking at, we asked one of the nation’s premier laboratories, Sandia National Laboratories, to conduct a detailed peer review of our methodology and our conclusions.
Among the organizations that worked with us as parties to the investigation over the past 16 months were the Federal Highway Administration (FHWA) and its Turner Fairbank Highway Research Center, the Minnesota Department of Transportation, Jacobs Engineering Group, the successor to the original bridge designer Sverdrup & Parcel (S&P); and PCI, the bridge construction contractor. We contracted with the State University of New York at Stonybrook and the SIMULIA company to assist our computer modeling efforts, and with the University of Minnesota to support on-scene work.
In order to understand the causes of the collapse, it is important to review some history of this bridge. It was built during the mid-1960s and opened to traffic in 1967. The collapsed portion of the bridge was a steel deck truss, which was considered “fracture critical,” meaning that a failure of any one of the major structural elements in the bridge would cause a collapse of the entire bridge. There are approximately 465 steel deck truss bridges in the National Bridge Inventory, according to the Federal Highway Administration. In the years since it opened, the I-35W bridge experienced two major renovations, in 1977 and 1998. As part of these renovations, the average thickness of the concrete deck was increased from 6.5 inches to 8.5 inches, and the center median barrier and outside barrier walls were increased in size and weight. These changes added significantly to the weight on the structure. On the day of the collapse, the bridge was undergoing repaving operations and there was construction equipment and material on the bridge.
The deck truss portion of the bridge consisted of steel beams that were connected to each other at nodes, or joints, by gusset plates. There were two gusset plates at every node for a total of 224 gusset plates on the main trusses of this bridge. During the wreckage recovery, we encountered fractured gusset plates from eight different nodes located in the main center span; all 16 gusset plates from those eight nodes were fractured. The other major gusset plates in the main trusses were generally intact. The damage patterns and fracture features uncovered in the investigation indicated that the collapse of the deck truss portion of the bridge was related to the fractured gusset plates, and in particular originated with the failure of the gusset plates at node U10. As you might imagine, this created questions about the materials used to construct the bridge. However, materials testing performed during the investigation found no deficiencies in the quality of the steel or concrete used in the bridge.
Subsequently, the Safety Board and FHWA conducted a thorough review of the design of the bridge, with an emphasis on the design of the gusset plates. This review determined that the design process led to a serious error in sizing some of the gusset plates in the main trusses; specifically, the gusset plates at the U10 nodes I mentioned earlier. Basically, those gusset plates were too thin to provide the margin of safety expected in a properly designed bridge such as this. The gusset plates were roughly half the thickness that would be required – half an inch thick rather than an inch thick.
Let me take a moment to show what our modeling revealed for the stresses in the U10 gusset plate on the bridge. Notice how much of the area is under high stress as indicated by red and orange. Now I would like to show the stresses under the same load using a properly designed gusset plate. These slides clearly show how the ½ inch gusset plates were stressed nearly to yield. Now I want to show you how the U10 gusset plate fractured. See how the two lines of fracture occurred where modeled stress was highest.
Once we determined how the failure occurred, we turned to why it occurred. Why were these gusset plates under-designed?
Investigators examined documents on the bridge design to establish what led to the under-designed gusset plates. First the Board looked at fabrication. An error by the fabricator or erector could have included a material mix up during part manufacture, or something as simple as a transcription error between the final plans and the shop drawings. However, the final design plans specified that the U10 gusset plates were to be ½ inch thick. Inspection of the shop drawings and testing of the actual gusset plates confirmed that they were indeed ½ inch thick steel. So errors in fabrication or construction were ruled out.
At this point the investigation focused on how the design firm made the error. The Board considered four possible sources of the design error: a simple drafting/transcription mistake where the material or thickness was incorrectly annotated on the drawing, a material change that was not applied to the gusset plates, a simple mathematical error in calculations, or a failure to perform the proper calculations at all. The Board’s investigation methodically ruled out each of the first three possible scenarios. In the end the Board found that the error was that S&P failed to perform all of the necessary calculations.
Further, and perhaps more importantly, S&P failed to catch the error through its quality control procedures. The Board found insufficient quality control procedures for detecting this omission of calculations, and insufficient Federal and State procedures for reviewing and approving bridge design plans and calculations. Thus, last January, the Board made recommendations to the American Association of State Highway and Transportation Officials (AASHTO) and FHWA to develop and implement a bridge design quality assurance/quality control program, to be used by the States and other bridge owners. This program should include procedures to detect and correct bridge design errors before the design plans are made final. The program will provide a means for verifying that the appropriate design calculations have been performed, that the calculations are accurate, and that the specifications for the load-carrying members are adequate with regard to the expected service loads of the structure.
Now, why did no one notice the under-designed gusset plates for forty years?
As we learned from the investigation, the I-35W bridge was flawed from the beginning. However, the bridge received numerous inspections and load ratings in the 40 years the bridge stood. Why did none of these inspections find the inadequate capacity of the gusset plates? Quite simply, load rating and inspections do not address gusset plate capacity. The computer programs used to identify the weakest link in the bridge did not include gusset plates. The investigation identified photographs from two bridge studies, one in 1999 and a second in 2003, that showed bowed gusset plates. One inspector indicated that he saw the bowed gusset plate but assumed the bowing occurred during construction. Therefore, the Board recommended that AASHTO include gusset plates as a commonly recognized structural element and develop guidance for bridge owners in tracking and responding to potentially damaging conditions in gusset plates, such as corrosion and distortion. Also, we recommended that AASHTO include gusset plates in the load rating process outlines in the Manual for Bridge Evaluation and include guidance for conducting load ratings when bridges are first built. The Board recommended to the FHWA modification of the bridge inspector training program to include a greater emphasis on gusset plates.
Finally, why did the bridge, that had been standing for the last 40 years, collapse on that day? The board meticulously measured the construction materials and vehicles on the bridge and incorporated that data into our bridge computer models. What we found was that the cumulative weight from previous construction modifications plus the concentrated weight of the construction materials and vehicles plus the traffic weight on that day exceeded the capacity of the U10 gusset plates.
The Board also learned that very few States had guidance on the placement of construction materials and vehicles. Let me be clear though, had the gusset plates been designed correctly, the bridge would not have collapsed under the loads of August 1. Still, the Board is concerned that heavy loads or concentrated loads can be placed on bridges with little guidance. Therefore, the Board asked AASHTO to develop specifications and guidelines for use by bridge owners to ensure that construction loads and stockpiled raw materials placed on a structure during construction or maintenance projects do not overload the structural members or their connections.
A final note about a factor that did not cause the collapse, the Board identified corrosion on the I-35W bridge that was much more extensive than what had been identified in previous routine inspections. However, the locations of this corrosion were not on the gusset plates that originally failed – the U10 plates and played no role in the collapse. The Board, nevertheless, is concerned with the amount of corrosion that went undetected. The area in the red is what was detected in inspections but the Board found the corrosion extended the whole length of the gusset plate. The investigation revealed that the corrosion was occurring in spaces that were difficult or impossible for inspectors the see. For example, water can run down the diagonals and pool behind a gusset plate. As a result the Board recommended to FHWA that States and other bridge owners make better use of technology for accurately assessing the condition of gusset plates on deck truss bridges.
In summary, the I-35W bridge collapsed due to the inadequately designed gusset plates. The Board has recommended better quality control of the bridge design, expanded inspection guidelines for gusset plates, creation of guidelines for the placement of construction materials, and the use of technology to better identify corrosion on bridges. Hopefully these changes will prevent a similar tragedy.
Thank you. I would be happy to answer any questions you might have.