System identification of torsionally coupled shear buildings using linear and hysteretic time domain data.

摘要

Vibration-based structural health monitoring methods aim to assess the state of structures through analyses of measured dynamic response. The first part of this dissertation describes the development of a non-iterative algorithm for estimating the stiffness properties of torsionally coupled shear buildings via direct analyses of linear response time-histories. The inherent characteristics of the reduced-order models of multi-story buildings are exploited to identify lateral and torsional stiffnesses of each story, independent of the others. The stochastic properties of random structural vibrations are utilized to extend the proposed algorithm so that it can accept ambient vibration data as input. The algorithm is verified and validated through the analyses of simulated and experimentally measured responses of a four-story quarter-scale steel frame benchmark test structure. These studies show that the algorithm is able to consistently estimate the loss of stiffness in subtle, moderate and severe damage scenarios, even in the presence of reasonable measurement noise.;The second part of this dissertation offers a parametric approach for identifying the nonlinear characteristics of torsionally coupled shear buildings, observed during moderate earthquakes. The nonlinear hysteretic force-displacement loops of the lateral load resisting systems (e.g., structural frames) are modeled with the Bouc-Wen approach. Using an Unscented Kalman filtering method, and data gathered during a damaging event, the nonlinear response of individual lateral-load-resisting systems are identified. The application of the proposed method is demonstrated through numerical simulations, where the identified and true hysteretic loops agree well. It is also shown that combined effects of estimation error in velocities and restoring forces may lead to slight deviations in the estimates of dissipated energy from the true values. The algorithm is also capable of estimating permanent drifts, which provide a quantitative measure of damage even in the absence of permanent stiffness degradation.