And Speaking of Large Earthquakes. . .
How will
Southern California’s steel-frame, earthquake-resistant high-rises
fare when the Big One hits? That’s a very complicated question that
a team led by Caltech postdoc Swaminathan Krishnan (PhD ’04) has
answered with unprecedented specificity using a supercomputer model—the
first to combine 3-D seismological simulations with 3-D nonlinear analyses
of building motions.
The model “ruptured” a 290-kilometer section
of the San Andreas fault between Parkfield, located in the Central Valley,
and Southern California. Two magnitude-7.9 earthquakes were simulated—one
rupturing southward and the other northward. The model calculated the
resulting motions for a grid of 636 points spaced 3.5 kilometers apart
and covering the Los Angeles basin, which includes the San Gabriel Valley
and Orange County, and the San Fernando basin, which is a geologically
separate entity, and applied them to two structures: an actual 18-story
building that was designed according to 1982 building code standards and
suffered significant damage in the 1994 Northridge earthquake when welds
failed, and the same building designed to the stricter 1997 standards.
The model predicted each building’s “peak interstory drift,”
which measures the structure’s distortion as it sways—for
example, for a 10-foot-high story, a drift of 0.10 means that the ceiling
is displaced one foot in relation to its floor. Zero-point-one is also
approximately the threshold of collapse, while anything over 0.06 indicates
severe damage.
Not surprisingly, L.A. fared worse in the south-propagating
rupture, where peak drifts for the 1982 building far exceeded 0.10 in
the San Fernando Valley, Santa Monica, and West Los Angeles, as well as
the areas around Baldwin Park, Compton, and Seal Beach. Peak drifts were
in the 0.06–0.08 range in the Huntington Beach, Santa Ana, and Anaheim
areas, and in the 0.04–0.06 range everywhere else, including downtown
Los Angeles. The 1997 version did better—although peak drifts in
some parts of the San Fernando Valley still exceeded 0.10, they were in
the range of 0.04–0.06 for most of the Los Angeles basin. In the
south-to-north rupture, both buildings scored in the 0.02–0.04 range,
suggesting damage enough to close the building but little danger of collapse.
Such hazard analyses could be performed on specific existing
and proposed buildings for a range of earthquakes, providing detailed
information for developers, building owners, city planners, and emergency
managers. “We have shown that these questions can be answered, and
they can be answered in a very quantitative way,” Krishnan says.
A southward-propagating 7.9 earthquake hit the San Andreas
in 1857, and seismologists think an event of that size—propagating
in either or both directions—could happen every 200 to 300 years.
To put this in context, the Northridge earthquake was a mere magnitude
6.7, yet caused 57 deaths and economic losses of more than $40 billion.
The results were published in October issue of the Bulletin of the Seismological Society of America. The other authors are Chen Ji (MS ’99, PhD ’02), now at UC Santa Barbara, Dimitiri Komatitsch of the University of Pau in France, and Jeroen Tromp, Caltech’s McMillan Professor of Geophysics and director of the Seismological Laboratory. Online movies of the earthquakes and building-damage simulations can be viewed here. —JP
