California has spent billions retrofitting freeway overpasses and public buildings to make them safer in earthquakes.
But in the absence of a real quake to test their strength, it has been impossible to say for certain how the structures would hold up. Some older regulations thought to protect structures against earthquakes have been proved by real temblors to be inadequate.
Scientists are trying to change that by building elaborate models of freeways and office towers and testing how they would respond to extreme shaking.
UCLA researchers spent $1 million building the foundation of the freeway overpass near Los Angeles International Airport and on Tuesday put it to the test.
They gave it a gigantic shove, pressing 2 million pounds of force against the foundation to see how far it would move.
Scientists were surprised by the results: The makeshift structure -- which looks like a giant block of concrete with metal rods running through it -- held up better than expected.
The data showing that the foundation slipped about an inch and a half on impact will be factored into future computer models of how bridges behave in earthquakes.
Earlier tests at UC San Diego used a massive shake table to jiggle life-size or near-size models of every part of the new San Francisco-Oakland Bay Bridge to make sure that it will hold up during the next big temblor.
One of the most elaborate experiments came last fall, when UC San Diego researchers built a seven-story building and tried to knock it down by replicating the shaking of the 1994 Northridge earthquake.
The work is important, said Tom Jordan, director of the Southern California Earthquake Center at USC, because otherwise the complex computer models used to develop building codes, insurance premiums and even architectural designs would not be grounded in real data.
The problem, Jordan said, is that there is a huge variation in how the soil and structures on it will react to any given earthquake. Simple mathematical models -- in which, say, it is assumed that a building will move twice as far if it is hit with twice as much force -- just don't work.
"Nobody is yet comfortable modeling these things on computers to predict real behavior," Jordan said. "We're always skeptical of simulations because materials behave in a very complex way -- especially when subjected to very strong forces."
