SAC Phase 1 Analytical Studies of Building Performance


Project Title:
Ambient Vibration Surveys of Three Steel-Frame Buildings Strongly Shaken by the 1994
Northridge Earthquake
James A. Beck, California Institute of Technology
B. Scott May, California Institute of Technology
David C. Polidori, California Institute of Technology
Michael W. Vanik, California Institute of Technology
Project Summary:
Ambient vibration surveys (AVS) can be used efficiently, cheaply and unobtrusively to identify the small-amplitude periods and modeshapes of lower modes of vibration of structures. These surveys sere performed on three steel-frame buildings which experienced strong shaking during the 1994 Northridge earthquake.

The primary objective of this study was to identify the small-amplitude modal parameters of the building for assessment of the analytical models constructed by others. A preliminary comparison of the modal periods obtained form the AVS with those calculated from the analytical models is presented. Although the periods identified from the AVS were considerably shorter than the model periods, the ratios of the identified to model periods for each mode were quite similar. The differences in periods are thought to be primarily because the analytical models treat only the structural frame, ignoring the stiffness of nonstructural components such as architectural partitions and building cladding.

This studyalso serves as a first step in the testing of proposed structural health monitoring methodologies. Since one of the tested steel-frame buildings was damaged and has not yet been repaired, its data represent an "after" damage state. The tests on this steel-frame buildings can be followed later by an AVS if the structure is repaired. These "before" and "after" AVS results would provide valuable data to test proposed global structural health monitoring methodologies whose goal is to detect, locate and assess damage by monitoring ambient vibrations. A successful structural health monitoring method would allow "hidden" damage to be detected almost immediatedly rather than weeks after an earthquake.

We were successful in identifying the modal parameters for the lower modes of vibration from the ambient vibration surveys conducted on one four-story and one eleven-story steel-frame building. The identified modal periods were compared with those from analytical models of the two buildings and found to be shorter.

An explanation for the difference in periods is that the models considered only the structural frame in the stiffness calculation. At low levels of vibration, such as in an AVS, non-structural elements like architectural partitions or building cladding can contribute significantly to the stiffness, thus shortening the natural periods. As vibration amplitudes increase, the effective stiffness decreases due to loosening or slipping of non-structural components, thereby increasing the natural periods of the building. Therefore, as the excitation level increases, the observed periods approach the periods computed from an analytical model treating only the structural system. This behavior is consistent with limited past studies (Beck,1978; McVerry, 1979) which showed that the identified periods from seismic response were closer to the model periods than to the identified AVS periods.

The ratios between AVS and analytical modal periods obtained in this study were similar for nearly all identified structural modes of a building, although they were different for each building. Assuming that measured seismic periods approximately reflect the model periods, prior studies corroborate this result. For example, a study of five buildings in the San Francisco Bay area indicated that the ratio of periods for different modes from ambient and strong motion vibration records may be characteristic of a given building (Phan et al., 1992). Another study (McVerry, !979) demonstrated this result for a number of buildings in the Los Angeles area. The period ratio does change from building to building, but no studies have been found which investigate how differences in the design and construction, or damage to the building, affect it.

Recent testing of the Millikan Library building on the campus of the California Institute of Technology used an eccentric mass shaker to excite the structure. For low excitation levels, the natural periods found were similar to those determined during a previous AVS (Beck et al., 1994a). With increased excitation levels, the periods of the first north-south, east-west, and torsional modes increased. Prelliminary analysis of the data suggests that for a given excitation level, the ratios of corresponding AVS and forced vibration periods for each of the identical modes are similar. This imples that the excitation level as well as the building characteristics affect the period ratios.

The period-ratio data imply that the lower mode periods of a building scale uniformly with changing response amplitude. This behavior could be caused by a nearly constant proportional decrease in stiffness over the building as the response level increases. A nearly uniform shedding of nonstructural stiffness would provide such a change in stiffness.

This study has shown that a considerable amount of modal information can be identified using ambient vibration data. Further work will be needed in order to show the usefulness of ambient vibration surveys in model validation and health monitoring.

SAC Logo SAC Home

SAC Steel Project
c/o Earthquake Engineering Research Center
1301 South 46th Street
Richmond, CA 94804
(510) 231-9477
FAX: (510) 231-5664