Structural Seismic Fragility Study Based On Model Updating

Seismic fragility analysis refers to the failure probability of structural members or systems given in the ground motion parameters, and as an effective way to the analysis of seismic risk assessment, which gets more and more attention around the world. In this paper, the fragility analysis is carried out based on Bayesian theory for a steel-concrete test model which combined the shake table test data with the numerical simulation data. On the other hand, using a health monitoring Benchmark structure as an example, the effects of damage identification to the structural fragility are analyzed. The main work are as follows:(1) The Bayesian theory is applied to the structural seismic fragility analysis, and the basic parameters of the non-information prior distributions and the normal linear regression model are determined, including position parameter, scale parameter and the Fisher information matrix without prior information. By the theory, the mathematical model and the seismic demand model are linked together and thus the experimental data can be used to update the demand model, in addition the result can update according to the newly acquired data.(2) A fiber model was established for the steel-concrete test model through LS-DYNA software and the probability function of the seismic demand can be achieved through numerical simulation. By means of the Bayesian theory, posterior probability of the seismic demand can be achieved combined with the shake table test data. The test data can be applied to the fragility analysis by the normal linear regression model of the seismic demand. This method can take advantage of the numerical simulation which can get large amounts of data and the reliable test data to evaluate the structural damage state due to different earthquakes more accurate.(3) The effects of damage identification to the structural seismic fragility, considering9different analysis cases, are studied for a steel frame model which uses Benchmark as an example. Results indicate that using the monitoring data to update the model can represent the current state of the small damaged structure, but for the moderate damaged structure, the uncertainties of identified damage extent and damage location have larger influence on the structural fragility analysis results, and the uncertainty of identified damage extent is more sensitive.