PREDICTIVE CAPABILITY OF NONLINEAR STATIC ANALYSIS PROCEDURES FOR SEISMIC EVALUATION OF BUILDINGS

The recorded response of a number of reinforced concrete buildings to real earthquakes are used to test the predictive capability of nonlinear static procedures (NSP). Response parameters such as drifts and inter-story shears are obtained by blending measured acceleration signals and model based information, using an observer. The buildings are represented by 3D nonlinear Finite Element models with elastic stiffness updated from eigenproperties identified from small amplitude response. The fidelity of the models for behavior at large amplitudes is validated by contrasting time history predictions with measured strong motion acceleration response. As one anticipates from theoretical considerations, the prediction accuracy of NSP is found to be significantly higher in the lower levels of the buildings considered. The ratio of “measurements ” to predictions over the full height for the cases analyzed has a mean and coefficient of variation of around {1.05 and 0.22} for shears and {1.2 and 0.45} for inter-story drifts. Differences in predictive capability between the various NSP are found statistically insignificant