Earthquake Damage Detection in Two Buildings.....

Earthquake Damage Detection in Two Buildings - Comparative Analysis of Several Structural Health Monitoring Methods Including a New Wave MethodSpeaker: Dr. Maria Todorovska,
Department of Civil and Environmental Engineering,
USCAbstract:The only true validation of real time structural health monitoring methods is in terms of sensory data recorded in instrumented structures during a damaging event. For civil structures, damaging events are e.g. earthquakes, explosions, blasts, etc. Due to the high cost of instrumentation and the low frequency of occurrence of such events, data in damaged civil structures are rare, but do exist, e.g. vibrational data recorded in buildings by seismic monitoring arrays during a damaging earthquake. Yet, the existing data are rarely used, and methods are usually tested only on numerically simulated data of response with postulated damage and assumed additive Gaussian noise. Results will be presented of a critical comparative analysis of several structural health monitoring methods applied to detecting damage in two instrumented buildings th the 1971 San Fernando and 1994 Northridge earthquakes, and for which records of 10 other earthquakes are also available.The first method is a new method, based on measuring wave travel times through the building using impulse response functions, applied to structural health monitoring for the first time in the presented work. This method is more reliable than monitoring changes in the apparent building frequencies, which are sensitive to the effects of soil-structure interaction and environmental factors, such as weather. Further, it is local and can detect the location of damage with relatively few sensors as compared to the methods based on detecting changes in curvature of the mode shapes of vibration, which require extensive instrumentation. Another relatively new method is based on detecting novelties in the recorded response using wavelets, previously tested on numerically simulated data but not conclusively on real data. This method is superior to all other methods in its resolution of the estimate of the time of the occurrence of damage. The analysis of this method also revealed that the noise in this method, consisting of high frequency pulses from the ground motion traveling through the building, contains useful information about the travel times through the building, and hence can also be used to infer about the changes of the state of health of the structure. The results by these two methods are compared with the distribution and degree of the observed damage, and with results of analyses based on other indicators of damage, such inter-story (dynamic) drifts estimated from the recorded accelerations, observed changes of the "instantaneous" apparent frequencies of vibration as functions of the amplitudes of response, estimated from the ridges and skeletons of the Gabor transform, and finally by analysis of simulated response using ETABS. The mutual consistency of the results by different analyses methods, and their consistency with the observed damage are discussed and several important conclusions are drawn.

Location: KLaprielian Hall 203

Audiences: Everyone Is Invited

Contact: Evangeline Reyes