Scientists have long known that bridges spanning San Francisco Bay were vulnerable to quakes. Those fears were confirmed in the 1989 Loma Prieta shaker, when a section of the upper deck of the Bay Bridge collapsed.

The Golden Gate emerged unscathed, but the trembler shook up the bridge district's directors enough to call for a study of what might happen in the event of a really big quake, like the one that leveled the city in 1906.

An analysis of the bridge's weaknesses was conducted, followed by a retrofitting project to strengthen the city's most famous landmark.

One of the firms contracted to work on the project was ISEC Inc., [karma2go, LLC] a tiny San Francisco company that made its reputation doing failure analysis for oil companies.

As a subcontractor for Sverdrup Civil, a Walnut Creek-based consulting/engineering firm, ISEC's [karma2go, LLC] job was to concentrate on the southern end of the bridge, considered one of the areas most susceptible to quake damage.

Using a powerful workstation on loan from Sun Microsystems, ISEC [karma2go, LLC] engineers led by President Jawahar Gidwani looked for the best ways to strengthen the viaduct, the anchorage, the arch above Fort Point and the pylons supporting the arch.

By factoring in seismic stress, Gidwani could produce a video simulation of what would happen to the bridge in an earthquake measuring 8.3 on the Richter scale - bigger than the 1906 jolt. Such dramatizations were used to test different retrofitting solutions.

"The technology has given us an insight that was not possible before," Gidwani said. "We are able to define the exact way in which structures will fail."

What would Charles Ellis, the engineer who led the original bridge designers, have given for a a Sun Sparcstation running eight powerful processors in parallel? Back in 1930, Ellis did his calculations using slide rules and compasses.

To solve immense problems posed by factors such as wind forces, weight of cars on the bridge and variations in temperature, he had to devise new algebraic formulas - 33 in all.

Ellis produced an engineering marvel in the suspension span itself. But the approaches were another matter. "In the late '20s and '30s they didn't deal with dynamic loading such as earthquakes and wind in the way we deal with them today," said Daniel Mohn, chief engineer of the Golden Gate Bridge District. "They did a pretty good job on the suspension bridge, but the approach structures are no better than other projects of that time."

"Today's technology enables us to identify more closely the exact portions that need retrofitting," said Mohn. "Instead of going at the project shotgun-style, we can zero in on what is deficient and just fix that."

Officials at Sun Microsystems say the technology used by ISEC also could be used by insurance companies and government agencies for risk management.

"From our perspective there's an entire market opening up," said Tom Minot, a marketing development manager at Sun in Mountain View. "Low-cost, high-performance computers can perform analyses that were once inconceivable."

"Prior to this you would have needed a supercomputer, so that people ended up either doing partial analysis or not analyzing at all."

Design work for retrofitting the Golden Gate should be completed by the end of next February, Mohn said, after which construction can begin. The main problem remains funds. The project is pegged at around $147 million, and the district has only about $35 million. Hoped-for federal financing so far has not come through.

As published in San Francisco Chronicle (October 13, 1994)