Zipper. Test of vertical member between upper floors to improve post-yield response was run at three separate labs. ( Images courtesy of Georgia Tech/NEES)

Broadband communications and faster computers are changing the way seismic engineering research is done. The combination enables simultaneous collaboration not only between individuals, but also between ever more sophisticated physical and computational experiments at simulation labs across the country. Such linking is letting researchers get closer to analyzing the complex response of systems in seismic events than ever before.

“We have very good tools for elastic or linear modeling and decent tools for mildly non-linear modeling, but the things that are difficult to handle currently in a computer are events when you have extreme loads that are fast, and when you have highly non-linear behavior,” says Clifford J. Roblee, executive director of the George E. Brown Jr. Network for Earthquake Engineering Simulation (NEES), headquartered in Davis, Calif.

Roblee offers an example of a braced-frame steel building’s columns, beams and floors subjected to the cyclic loading of simulated ground motion. In the test, braces are beginning to buckle. Engineers are trying to use the deformation to dissipate energy and reduce damage elsewhere as the event continues. Scientists now can model such events by using tools enabled by linking data passing back and forth between separate physical models of discrete portions of the structure and combining it with computer simulations and computations via fast data transmission. Software can collect the seismic event and structural-response phenomenon into one virtual test. It can send feedback to hydraulic actuators on all the models to modify the loads in one part of the structure to simulate evolving response in another. Roblee calls it fast, numerical/physical hybrid testing.

“We’ll take the simple parts that there is good data on and run those in a computer,” explains Roblee. “But we take the complex parts that we don’t have good data on and run the physical tests—and they are interacting. We apply a load to the physical specimen, compute the response, put those loads into the computer algorithm and redistribute the load in near-real time. Depending on how one member buckles, we can load the computational model.”

The output from the physical test is used to shape the computational test, which in turn calculates new results to drive the hydraulic rams or other actuators loading the physical model again through the wracking cycles of the quake. The instant feedback, which can come from tests on different portions of the system at scattered locations, allows larger, far more complex experiments than any one lab could ever perform.

One example was a “zipper-frame” test conducted with different parts of the model at Georgia Tech, the University of California Berkeley and at the University of Colorado at Boulder. Roblee says a zipper frame adds a vertical member between upper floors at the intersection of angled braces. When the braces begin to deform load is taken up by the vertical member and successfully transferred to the columns, preventing collapse. “Simply by adding a couple of members you can get 10% to 20% better post-yield response,” Roblee says. “It was proposed as a concept in the 1980s but there has never been any systems-scale testing until two years ago, when it was used as one of the early demonstration projects for NEES.”

Another project is non-parallel hybrid testing, such as current research on a tall foundation interaction problem for a bridge. This collaborative project led by Sharon L. Wood, University of Texas at Austin, involves collaborators at UC Davis, University of Nevada at Reno and Purdue University, West Lafayette, Ind. It addresses the critical need to improve knowledge of soil-foundation-structure-interaction during seismic events.

Bridging

The structure investigated is a continuous bridge whose dynamic response is influenced by the ground motion and the nonlinear characteristics of the soil, foundation and super structure.

But testing a single physical model of the structure and reproducing all key aspects of the system is not possible, so a series of complementary, but not necessarily simultaneous physical model tests are under way at the four sites.

Related Links: Earthquake Simulation Network Rattles from Coast to Coast, April 17, 2006 Skyscraper Performance-Design Anxiety Is Turning Into Cautious Optimism, April 17, 2006 Seattle Makes 'Performance' Skyscraper Approval Look Easy, April 17, 2006 Quake Engineering Moves Toward Era of Empowerment, April 17, 2006 Bridges Include Elements to Make Them Stand after Shaking, April 17, 2006 More Multimedia: Seismic Strides: 100 Years of Achievement | View >>

Centrifuge tests of individual bents are being conducted in Davis to evaluate the nonlinear response of the soil and foundation system, while dynamic field tests of individual bents are being conducted in Austin using mobile shakers to evaluate the response of the soil, foundation and structure. Meanwhile, shake table tests in Reno are testing a half-scale, three-span model to evaluate the nonlinear response of the superstructure, and finally, laboratory tests of large-scale individual columns are being run at Purdue to evaluate strength degradation in flexure and shear.

By linking labs with real-time data exchange “the expert doesn’t have to be present to do the testing,” says Roblee. “We are finding broadband communications very helpful there.”