Structural Health Monitoring Using Ambient Vibrations

Abstract

Ambient vibration tests can be used efficiently, economically and unobtrusively to identify the small-amplitude periods and modeshapes of the lower modes of vibration of a structure. It is important to determine whether identified changes in the modal parameters can be used in a global structural health monitoring scheme which has as a goal the detection, location and assessment of structural damage by vibration monitoring. A successful health monitoring method would allow "hidden" damage to be detected and located in a structure almost immediately after it experiences a damaging load event such as an earthquake, hurricane, explosion or impact. It should also allow progressive structural damage from environmental deterioration, such as fatigue damage in steel off-shore structures, to be detected and located whenever the severity of damage exceeds some threshold level. Rapid detection of damage would eliminate the threat of a structure remaining in an undetected weakened state for months or years after an earthquake, as occurred after the 1994 Northridge and 1989 Lorna Prieta earthquakes. So far, no technology has been demonstrated to reliably provide these functions for actual structures. Development of a workable technology is hampered primarily by inadequate methodologies and lack of opportunities for field testing. One approach to structural health monitoring is to "invert" the changes in modal parameters to get the corresponding local changes in stiffness, which are taken as a proxy for damage. Usually a model-based system identification method is used for this inversion. The identified modal parameters are, however, incomplete because of limitations on the number of observed degrees of freedom and the number of modes that are excited in ambient vibrations. They are also "noisy" because of the effects of measurement noise in the original vibration data, the absence of excitation measurements, and nonlinearities and other modeling errors. These difficulties make the inverse problem ill-conditioned and often non-unique. A new probability-based approach will be described to treat these difficulties which focuses on the probability of damage in a substructure or member of a structure. Ambient vibration surveys (AVS) were performed were performed by the authors on an eleven-story steel frame building in Los Angeles. Nearly 30% of the steel connections in the building failed as a result of the Northridge earthquake and the building was tested in its damaged state and again after it was repaired to restore it to its original condition. Modal parameters identified from the AVS are presented and their use in indicating the presence of damage in the structure is discussed.

Additional details