System and damage identification studies of large/full-scale structural specimens subjected to seismic shake table testing

The last decade has witnessed significant advances in sensing technology, structural identification, and structural health monitoring. Vibration-based, nondestructive damage identification makes use of changes in dynamic characteristics (e.g., modal parameters) to identify structural damage. Damage identification consists of: (1) detecting the occurrence of damage, (2) localizing the damage areas, and (3) estimating the extent of damage in the various damage areas. Numerous vibration-based methods to achieve these goals have been proposed in the literature. Extensive reviews on vibration-based damage identification were provided by Doebling et al. [1] and Sohn et al. [2]. A class of sophisticated methods for structural health monitoring consists of applying sensitivity-based finite element (FE) model updating for damage identification [3] where damage is defined as local losses of stiffness within the structure. These methods update the physical parameters of a FE model of the structure by minimizing an objective function expressing the discrepancy between FE predicted and experimentally identified (before and after damage) structural dynamic properties that are sensitive to damage such as natural frequencies, mode shapes, and mode shape curvatures. Optimum solutions of the problem are reached through sensitivity-based constrained optimization algorithms. While many researchers have successfully developed and applied structural identification and vibration-based damage identification approaches to numerical and/or small-scale laboratory models of civil structures, there have been relatively few applications to full/large-scale structures with realistic design details and damage scenarios [4-8]. Field and large/full-scale laboratory applications allow the various approaches to be examined under realistic instrumentation constraints, levels of measurement noise, sources of uncertainty such as estimation uncertainty and modeling errors, and damage conditions. Actual construction practices often cannot be reproduced in small/medium-scale test specimens.