Identification and Finite Element Modeling of Alcazar Building in Chile Using Earthquake Vibration Data

Abstract eng:
Earthquake vibration data collected from an instrumented structure contain important information for estimating the dynamic characteristics and updating models of the structure. In this work we use earthquake acceleration time histories from three low amplitude earthquake events recorded for the 16-story Alcazar building office in Viña del Mar (Chile) at the far field, basement and various floors, in order to identify the building’s modal characteristics (modal frequencies, modal damping ratios and mode shapes). For this, modal identification techniques developed to handle non-classically-damped modes are outlined. A three-step modal identification approach is proposed. In the first step, conventional least squares complex frequency algorithms along with stabilization diagrams are used to automatically estimate the modal frequencies and the damping ratios of the modal model and to distinguish between the physical and mathematical modes. In the second step, two approaches are proposed for computing the mode shapes and the effective participation factors, one very efficient one which does not require iterations, and an iterative one that uses the estimates of the first approach as initial values to accelerate convergence. In order to improve the estimates for closely-spaced and overlapping modes, the full nonlinear optimization problem is solved in the third step by using the initial estimates of the parameters obtained in the first two steps. The modal properties are useful in updating finite element (FE) models. A high-fidelity FE model is also developed in SAP and the modal properties are compared with the ones identified using the measurements. Up to 7 modes were reliably identified from the analyses of the three seismic events. The lowest two bending modes in the EW direction, the three bending modes in the NS direction and the second torsional mode. Due to the sparse sensor grid, the type of some of the identified modes could not be easily estimated without the use of the FE model as a guide. For the majority of the identified modes, the damping ratios ranged from 2% to 3%. The modes were found to be very close to classically-damped modes. The analysis of the three events gave consistent results for the modal properties. Most of the modal frequencies have relatively small variation over the three seismic events. The values of the identified modal frequencies are close to the values predicted by the high-fidelity FE model. The modal frequency values predicted from the FE model are higher, indicating that the model is stiffer than the actual behavior of the building. The identified modes are representative of the condition of the building at low-amplitude vibration levels and can be further used for model updating, linear and nonlinear analyses, as well as structural health monitoring purposes. Matching the mode shapes in an FE model updating methodology presents challenges mostly from the fact that there is no measurements available from the 7th up to the 16th floor of the structure. FE model updating using hierarchical Bayesian modeling to account variabilities from the different seismic events is left for a future work.

• Export as: BibTeX | MARC | MARCXML | DC | EndNote | NLM | RefWorks
• Add to your basket