WASHINGTON - Mystified about how a simple human error occurred four decades ago, leading to the deadly collapse of the Interstate 35W bridge, federal safety officials recommended new rules Friday to prevent such error in the future.
In the end, the National Transportation Safety Board attributed the failure of the downtown Minneapolis span to a math calculation that wasn't made. The board found that the designers of the doomed bridge failed to perform some necessary calculations on the critical steel beam gusset plates that eventually gave way.
That failure, combined with inadequate oversight by state and federal regulators, led to the construction of a bridge that would not withstand all the weight that was added over its 40-year life, which ended with a total collapse on Aug. 1, 2007, killing 13 rush-hour motorists.
"It's intriguing to me that an error of omission of that magnitude would have been made," said Steven Chealander, one of five board members who signed off unanimously on the final report. "It's inexplicable."
The board's long-awaited findings, released Friday in summary form, were accompanied by a recommendation for a nationwide bridge design quality assurance program. Federal and state officials vowed immediate action.
The NTSB's final conclusions are likely to provide fodder for a spate of lawsuits that have been filed this week by relatives of the dead and about 145 others who were injured in the accident. While the NTSB concluded that the primary cause of the bridge collapse was its under-designed gusset plates, it also highlighted the heavy construction loads that were placed on the bridge the day it fell.
The design flaw was pinned in large part on a now-deceased engineer working for Sverdrup & Parcel, which was contracted to design the 35W bridge in 1962. The firm was bought by Jacobs Engineering a decade ago.
NTSB investigators who pored over 10,000 pages of documents from Jacobs Engineering and state highway officials from the 1960s found no evidence of completed calculations for the design of the gusset plates, which tie the bridge's steel beams together.
The lack of documentation, and the relative ease of the gusset plate calculation, led investigators to rule out simple calculation error as the source of the bridge's fatal design flaw.
Ultimately, the 1/2-inch thickness of the plate was half of what it should have been. "Had they done the calculations, you wouldn't have a half-inch gusset plate," said Joe Osterman, the managing director of the NTSB investigation. "It's not a complicated calculation."
Compounding the error was a lack of systematic review by the Minnesota Department of Transportation and the Federal Highway Administration engineers, who apparently relied on S&P's quality assurance practices and did not check the gusset plate calculations.
"While the design firm did not check its own work, neither did MnDOT and the FHWA," said NTSB structural analyst Joseph Epperson.
Although current state and federal review processes have become more sophisticated, NTSB officials said they still may not be adequate to detect design errors, however uncommon they may be.
"What we've uncovered is there's not a systematic quality assurance process," said NTSB board member Kitty Higgins, who has family in the Twin Cities. "This is a wake-up call."
A full NTSB report, to be released in several weeks, will call for new federal and state inspection standards and review procedures of bridge designs, including gusset plates.
"We won't wait for national standards," said MnDOT Commissioner Tom Sorel, who attended the NTSB's final two-day hearing. "We're going to move forward in Minnesota."
U.S. Transportation Secretary Mary Peters said federal officials are "acting immediately" to work with states to improve quality controls during the bridge design process.
The NTSB report also faults the load rating systems periodically used to determine the strength of bridges. The 35W bridge underwent its first load rating in 1979, some 12 years after it opened. It underwent another review in 1995, after further enhancements added more weight to the bridge. But neither review looked at the adequacy of the flawed gusset plates.
New rules are now being refined in Minnesota and across the nation. "Progress has been made," said NTSB chairman Mark Rosenker.
Rust not a factor
While the 35W bridge had been classified as structurally deficient since 1991, the NTSB found that the aging and corrosion that led to that poor rating was not a factor in its failure.
"A structurally deficient bridge is not ready to fall down," Rosenker said.
U.S. Rep. Jim Oberstar and other Minnesota DFLers had raised concerns about aging infrastructure, corrosion, and inspections on the bridge, which was scheduled to be replaced between 2020 and 2025.
The NTSB found that the bridge had been inspected more frequently than federal guidelines required, although in at least one instance corrosion to one gusset plate -- not one of the ones that failed -- was "grossly under-reported." Photos from two prior inspections showed bowing in the gusset plates that ultimately did fail, but the defects were not noted in the reports.
"Nobody understood it," Rosenker said. "We now have a better understanding."
Some critics, including NTSB member Higgins, raised concerns that the federal probe did not consider the quality of Minnesota's bridge inspection program.
Mark Bagnard, the NTSB's lead investigator, said current bridge inspections are designed to assess the bridges' conditions, not their original designs.
Oberstar said the NTSB findings support his efforts to pass legislation beefing up bridge inspections, including assessments of their original design features. "It should be very clear there was a design flaw," he said. "That's an important lesson for the future of bridge inspections."
Also among the key recommendations of the NTSB report is a call for greater attention to the effects of rust and corrosion on the structural integrity of bridges. "They are critically important," Rosenker said. "At the same time, this accident was not the result of corrosion and aging infrastructure. It was the tragic result of an inadequately designed gusset plate."
While NTSB investigators expressed confidence in their conclusion, they acknowledged the difficulty of proving a negative: That not all the necessary calculations on the bridge were done.
"We're looking back 45 years and inferring an error of omission based on a lack of documentation," said Vern Ellingstad, the NTSB's director of research and engineering. "It's a fair inference."
But the NTSB's Chearlander noted that it's a circumstantial conclusion. "We don't categorically know they didn't do it."
Another intriguing question that may never be answered is why.
In the end, Epperson said, "We didn't come down on why it wasn't done. ... It's very difficult for staff to understand an error of such a basic thing. But we're all human."
Kevin Diaz • 202-408-2753