Papers »

Seismic response assessment of structures in the presence of modeling uncertainty

Seismic response assessment of structures is an essential part of performance based earthquake engineering. The robustness of seismic response assessment is affected by the presence of uncertainties. Explicit consideration of variability in ground motions and structural modeling is essential in order to make robust predictions of seismic response. In this study, we focus on the uncertainty in structural modeling in seismic response assessment. We first demonstrate the impacts of modeling uncertainty on seismic performance through a set of code conforming and non-code conforming structures. The impact of modeling uncertainties is most pronounced when the structure approaches collapse state. We then discuss various methods for propagating modeling uncertainties through collapse risk assessment. These methods are presented focusing on their ability in characterizing the influence of modeling uncertainties and variability in component behavior on overall system response in comparison with their computational demand.

The balance of accuracy and computational expense is an important consideration in uncertainty propagation. In this study, we also present a Bayesian approach for efficient collapse response assessment of structures. This approach uses prior information on the collapse response of the structure, which can be informed from building design criteria, simplified linear dynamic analysis or nonlinear static (pushover) analysis. The results from a small number of nonlinear dynamic analyses are combined with the prior information in a Bayesian setting, which is observed to improve statistical and computational efficiency in collapse risk prediction.

Author(s):

Beliz Ugurhan Gokkaya    
Stanford University
United States

Jack Baker    
Stanford University
United States

Greg Deierlein    
Stanford University
United States