MODELING AND OUTPUT-ONLY IDENTIFICATION OF LINEAR DAMPED STRUCTURES

The research focuses on the modeling and the structural identification of linear damped structures. In particular, the attention has been set to models for which the dissipative forces depend either on the instant values of velocity (viscous damping) and on the history of velocity (non-viscous damping), in order to accurately reproduce the dynamic behaviour of of structures with energy dissipation devices. Specific interest has been pointed also towards output-only identification techniques, because in real life applications it is not always feasible neither to measure the input acting on a structure nor to excite it with an input able to overcome the noise. The objective of the thesis consists in the definition of a unitary output-only identification methodology for linear systems with viscous and non-viscous damping. The methodology is subdivided into three steps. The first is the identification of a linear first-order black-box model through the technique Data-Driven Stochastic Subspace Identification (Data-SSI); a modal model with complex eigenvalues and eigenvectors is subsequently extracted; nally, the definition of a physical model is carried out by using a sensitivity-based model updating procedure. The thesis is subdivided into two parts. Within the first, the direct problem and subsequently the inverse problem are formulated. In the direct problem, the equation of motion are written in the state-space and the study of the modal properties is oriented towards the output-only identification. In the inverse problem, several aspects are faced: the definition of rules to choose the user-dened parameters of Data-SSI; the evaluation of the performance of Data-SSI when responses to non-stationary excitations are used, and the inuence of four dierent weighting matrices; the definition of a model updating taking into account the complex eigenvalues and eigenvectors. The performed experimental activity in laboratory and in situ through which it has been possible to validate the proposed methodology is presented in the second part of the thesis. Two experimental campaigns on shaking table have been conducted in laboratory on scale models of a steel frame structure and on a structure equipped with high damping rubber isolators. Other two have been carried out in situ and have pointed out the potentialities of the methodology in the eld of structural health monitoring. The rst has concerned ambient vibration tests on a footbridge in Rieti equipped with high damping rubber isolators and exhibiting excessive vibrations, while the second has concerned ambient vibration tests on a part of the "San Camillo de Lellis" hospital in Rieti, a strategic building structurally complex due to its extension and configuration, the presence of an expansion joint and its seismic vulnerability. The proposed methodology has resulted eective for all the studied cases.