Hoping to prevent past mistakes during the rebuilding process, civil engineers gathered in New Orleans to determine how Hurricane Katrina breached the city's levees, flooding 80 percent of the Big Easy (search).

The central issue they grappled with: Did Katrina (search) overwhelm the city's flood defenses with a torrent they weren't designed to contain? Or did faulty construction or maintenance cause them to burst open at water levels well within their capacity?

"The whole rebuilding of this infrastructure is, I think, a critical issue for us to come to grips with," said Robert Bea (search), a University of California, Berkeley, civil engineer who is part of an investigation team sponsored by the National Science Foundation.

Experts who gathered in the city last week say it is too early to know exactly what caused the barriers to fail Aug. 29. But much of what they saw suggests that better design and construction might have prevented the catastrophe.

If that is the case, then public institutions or contractors involved in building and maintaining the levees around New Orleans could be vulnerable to billions of dollars in lawsuits.

Government agencies are notoriously difficult to sue for damages, because federal law grants them various forms of immunity. But any case that could show the storm surge from Katrina never exceeded the walls' specifications "might have a chance of getting somewhere," said Mark Wasser, a Sacramento, Calif., attorney who has successfully litigated several cases against state and local flood control agencies.

Officials from the U.S. Army Corps of Engineers have said the barriers were never intended to withstand a storm as powerful as Katrina. Congress had instructed them to build a network of levees and floodwalls that could withstand a Category 3 storm similar to Hurricane Betsy, which flooded New Orleans in 1965. Katrina was a Category 4 hurricane when it hit, so it would be expected that floodwaters would pour over the levees.

That is evidently what happened on the east side of New Orleans, where an earthen levee was overwhelmed in numerous places by floodwaters surging in from the Gulf of Mexico. But farther west, along the 17th Street and London Avenue canals, engineers have seen few signs that the water ever got high enough to pour over the storm barriers.

Now, experts are asking if the levees failed because the floodwaters rose above them, or if they crumbled when the water was still well below their tops. The issue is critical because engineers do not want to repeat mistakes when they rebuild New Orleans' flood defenses.

Because taller levees need broader bases, the earthen mounds along the 17th Street and London Avenue canals are topped with concrete walls that are designed to increase the barriers' height without taking up land in adjacent neighborhoods. Up to 11 feet high, the walls are anchored to the ground by steel sheets driven into the earthen levee.

Immediately after Hurricane Katrina, Army Corps of Engineers officials hypothesized that parts of the wall had been undermined as the flood poured over them, cascading down the barrier's landward side like a waterfall. The force of the falling water would have scoured out dirt along the wall's foundation and undermined it.

Many of the levee breaches appear to have happened that way, said Raymond Seed, a University of California, Berkeley, civil engineer who headed the National Science Foundation team.

"That was a very common mode, and one lesson there is to prevent the erosion," he said. "I anticipate a number of wall sections will be armored for that in the future."

But Paul Kemp, a professor at the Louisiana State University School of the Coast and the Environment, and a number of colleagues are convinced that in many locations the water never reached the tops of floodwalls.

They appeal to evidence such as the bathtub ring-like high-water mark that can be seen in many places, indicating that the water never rose more than partially up the walls. They point out that the water never got over the levees along Lake Pontchartrain, where the walls are the same height.

Computer simulations of Katrina performed by researchers at the LSU Hurricane Center also suggest the water never rose high enough to pour over the walls, though in some places it could have gotten close.

A number of engineers suspect a process known as heaving undermined the floodwalls in the London Avenue and 17th Street canals. The pressure exerted by water in the canals would have squeezed soil out from underneath the floodwalls. In some places, Seed said, entire sections of levee embankment appear to have moved as much as 35 feet.

Engineers would be expected to consider heaving when designing the walls and reinforcing the soil beneath them, said J. Michael Duncan, Virginia Tech geotechnical engineering professor.

"You would design against it, and you would use factors of safety to try to ensure that this wouldn't happen," Duncan said. "But there are always uncertainties involved."