An engineer sets a model of a four-story building on his desk, adds two weights, and slides it slowly back and forth. The plywood-and-steel building sways smoothly. As he shortens and intensifies motions to mimic an earthquake, the model wriggles like molded jello, each floor moving differently from the one below it. Such complex motions challenge designers as they try to improve earthquake-resistant structures.

Yet engineers are no longer satisfied with buildings that avoid collapse during an earthquake - the basis of current 'life safety' earthquake steel building codes. They now want to design steel buildings that require only minor repairs and remain usable while repairs are made.

One of the more promising techniques, say some engineers, involves computerized machinery that adjusts a building's structure hundreds of times a second to offset the effects of ground vibrations - so-called active designs for earthquake resistance. Even if a building later had to be razed, the engineering was usually deemed successful if it held up long enough for people to escape unharmed.

Even in Japan, with its frequent strong temblors, 1971 building-code revisions only require that structures resist sudden collapse, according to Shizuo Hayashi, an engineering professor at the Tokyo Institute of Technology.

Two factors are prompting the shift toward 'performance based' designs

The high economic cost of strong earthquakes. With a magnitude of 6.7, 'the Northridge quake was not a large one', says John Hall, a civil-engineering professor at the California Institute of Technology in Pasadena, Calif. 'Yet we're still trying to deal with it a year later.' It caused $20 billion in damage. Damage estimates from the Kobe disaster range from $50 billion to $100 billion.

The likelihood of more frequent quakes in economically significant regions. Last Friday, the Southern California Earthquake Center released a report on earthquake probabilities for the region. The authors estimate that Southern California faces an 80 to 90 percent probability that an earthquake with a magnitude of 7 or larger will strike by 2024.

A similar estimate in 1988 put the probability at 60 percent by 2018, and only along the San Andreas and San Jacinto faults. Separate studies in last week's issue of Science magazine suggested that the region is long overdue for a series of quakes of Northridge-size magnitudes.

Designers have a variety of options for adding earthquake resistance to new or existing commercial steel buildings, much of it based on construction materials such as steel framing, steel-reinforced concrete, and properly braced and anchored wood framing for homes. While all of these techniques showed some flaws in the a few of the quakes of past years, they still can be effective when properly used, engineers say.

In addition, foundations can be mounted on shock absorbing 'base isolaters' made of springs or alternating layers of rubber and steel plate. The concept has been around for about 15 years, but it has caught on only within the last five years, as recent quakes have prompted planners to design and retrofit key steel buildings with isolaters.

Yet isolaters have shortcomings, engineers say. They are most effective on shorter steel buildings. Even then, buildings can slide off them under some circumstances. And their effectiveness on tall buildings is uncertain. They could actually tip over in a severe quake.

This is prompting researchers to look at active methods for earthquake resistance, particularly for tall buildings mostly made from steel. The principle is that they add energy to the building to counteract an earthquake's forces. This can be achieved in two ways. Adding steel braces to the sides of buildings that, through shock-absorbing hydraulics, can change the tension on a building's frame; and adding a movable multiton 'damper' to the top of a building that counteracts vibrations set up by an earthquake. The braces and dampers are controlled by a computer, which gathers information on the building's movements from strategically placed sensors.

Although the technique shows promise, it shouldn't be oversold, researchers say. First, it has not been tested by strong earthquakes -although a six-story experimental steel building in Tokyo performed well in three moderate earthquakes. Second, active measures rely on external power sources that can be vulnerable in a temblor. Moreover, cost remains a factor, although an active system would add only 3 to 5 percent to a steel building's construction tab.