Link to site: Bob Marshall, Staff writer, March 17, 2006

Return to: watercenter.org sciencefaircenter.com watercenter.net

Highlights:
- The combination of forces that brought the structure down was its finding that one of the main triggers for that failure -- extremely low soil strengths under the toe of the levee -- would have been detected had the design team done soil borings in that area
- "The factor of safety would have been (low enough) to where they would have changed the design,"
- The task force said rising water pushed the wall away from the canal, eventually creating a crack, separating the wall from the canal-side levee. Water pressure building inside the crack began pushing down on soil layers under the wall, which required support from the levee on the land side of the canal and the soils adjacent to it. The weak soils beneath the toe of the levee couldn't stand up to the rising pressure and began slipping, bringing the levee and the floodwall down.


The key to learning why the 17th Street Canal floodwall failed during Hurricane Katrina may lie more in what designers didn't do than in what they could have foreseen, experts now say.

Lost in the controversy swirling around a government panel's comment last week that the designers of the floodwall could not have anticipated the combination of forces that brought the structure down was its finding that one of the main triggers for that failure -- extremely low soil strengths under the toe of the levee -- would have been detected had the design team done soil borings in that area, an official with the Army Corps of Engineers said Thursday.

Had the weakness at the toe of the levee been included in the analysis system used by the project designers, "The factor of safety would have been (low enough) to where they would have changed the design," said Reed Mosher, a researcher at the corps' Engineering Research and Development Center in Vicksburg, Miss., and a member of the corps-sponsored Interagency Performance Evaluation Task Force that is investigating the failures. The options considered probably would have included a T-wall, or a much larger levee, he said.

The task force said rising water pushed the wall away from the canal, eventually creating a crack, separating the wall from the canal-side levee. Water pressure building inside the crack began pushing down on soil layers under the wall, which required support from the levee on the land side of the canal and the soils adjacent to it. The weak soils beneath the toe of the levee couldn't stand up to the rising pressure and began slipping, bringing the levee and the floodwall down.

Review team members said the designers did "few if any" soil borings at the toe of the levee, a finding John Greishaber, acting chief of the engineering division at the corps' New Orleans district, said was not normal. He said his office normally required designers to take borings at the center line as well as at the toe of levees.

"This is the preferred method," he said. "There are items when this is not done. You have to get into specifics (for each case) as to why it is not."

Greishaber said that when borings aren't made, engineers can estimate the soil strengths at the toe of a levee.

Engineers use a standard formula for estimating the soil strengths at the toe based on the known strength of soils at the center line of the levee, where the soil strengths are highest. That means soil borings at the toe usually aren't necessary unless the center line values are below a certain threshold, task force members said.

And that is where the designers made obvious mistakes, said J. David Rogers, a professor at the University of Missouri-Rolla who is a leading expert on levee failures and a member of a National Science Foundation investigation into the disaster.

"Looking at their calculations on the slope stability analysis, they used the same high figure from the center of the levee and projected it out to the toe, without any diminution in value," Rogers said. "That was one of the first things we picked up when we started working on this.

"When we tried to find out what factor they used for diminution with increasing distance from the toe, it didn't appear they used any. They were using maximum strength all the way to the toe. That's the part everyone will take issue with."

More surprising, Rogers said, is the fact that obvious mistake was missed by the corps in New Orleans, as well as its superiors in Vicksburg.

"I can't explain how this went through," he said.

Making waves

Although the quality of the engineering done by local firms and reviewed by the corps has been the focus of scrutiny since shortly after the walls collapsed, it was pushed from the headlines last week when the task force released an interim report identifying how the walls collapsed and saying the combination of forces responsible could not have been anticipated by the project designers. That provoked criticism from independent investigators.

But this week Ed Link, project director for the task force, said his panel's statements had been misconstrued by the media.

"Our position on this is that, very simply, whoever did the design just did not consider this particular mechanism," said Link, a University of Maryland senior fellow who is head of the corps-sponsored Interagency Performance Evaluation Task Force. "We, IPET, made no value judgment whether it should have been considered or could have been considered.

"If that was inferred by our comments, it was inaccurate."

Link added that the corps has made no attempt to interfere or steer the investigation by the panel, which lists more than 150 members from academia, private industry and other state and federal agencies.

"The only pressure the corps has put on us is to find out what has happened and put it in the public domain," he said. "I'm telling you as an engineer, as a professional, I would not work in this environment if I felt there was anything political or adverse pressure on what we are doing."

The executive summary of the task force report, which Link said he wrote without input from the corps, said "this failure mechanism was not anticipated by the design criteria used."

When task force panelists and corps engineers were asked if that meant the design systems used by the engineers of the day could not have foreseen this type of failure, they answered "yes."

Link said that while the individual components of the failure are well documented as concerns for engineers doing stability analysis of levees and floodwalls, the combination of those factors coming together at the same time is not. He also said methods of analysis used by engineers at the time would not have included all those factors in testing a design for stability.

Point of contention

Task force panelists at the press conference also said a "search of the literature" turned up no examples of this specific failure mechanism.

Those claims were quickly challenged by members of the engineering community. Most notably Ray Seed and Bob Bea, University of California-Berkeley professors and members of the National Science Foundation team investigating the levee failures, issued a response calling the task force statements "unfortunate" and inaccurate. They called attention to a 1986 report done by the corps, known as the E-99 report, that showed the separation -- "tension cracking" -- of the wall as well as the build-up of high pressure at the base of the floodwall after the cracking.

They also cited two 1997 papers published in an industry journal analyzing the 1986 test. One of papers' authors was Mosher, who is a member of the task force.

Link said Thursday that his reference to the "literature" meant a review of the corps engineering manuals, which design teams are required to use.

"We were looking at the design criteria to see if there was a process like this described in the corps' design manuals that (the design team) missed," he said. "We didn't see anything that described this mechanism, that would have alerted (the design team) to look for this when doing their analysis."

Link and Mosher disagreed with Bea and Seed's analysis of the importance of the 1986 study. Mosher, who analyzed the E-99 report, said it was not designed to look at levee stability, but at how much a sheet pile "moved at the top as water increased."

The fact that the test also showed there was evidence of tension cracking and high pressure at the toe of the wall was not given much attention at the time, Mosher said, "because the study was not designed to look at the stability of the levee." He also said the evidence of cracking and increased pressure was minimal.

Rethinking strategies

Mosher said Katrina has made the report important today.

"When I go back now and look at E-99 knowing the other pieces of information about the 17th Street failure, I can make a better interpretation of what's in E-99," he said. "Now I can say I understand how all this relates."

Mosher and Link said the lessons learned from the investigation already are being put to work.

"We're going back and doing borings at the toes of the levees in the system anywhere we think this failure mechanism might be present," he said. "We're already doing re-evaluations of the stability analysis done by the (original design teams).

"Now that we know what to look for, we're out there looking for it."

. . . . . . .

Bob Marshall can be reached at rmarshall@timespicayune.com or (504) 826-3539.