ABSTRACT
ABSTRACT: Performance-Based Design in earthquake engineering accounts for structural and ground motion uncertainties in optimizing design parameters, satisfying multiple performance criteria with target reliabilities and minimizing an objective. This implies a nonlinear structural dynamic analysis for earthquakes likely to occur at the site, to obtain the demand responses of interest. The responses are represented here via neural networks and used in the performance criteria for estimation of achieved reliabilities by Monte Carlo simulation. An optimization is finally implemented to obtain optimal parameters, using a gradient-free algorithm. An application is shown for a reinforced, multi-story concrete frame. The demand is simulated using artificially generated ground motions, with the peak ground acceleration corresponding to the city of Mendoza, Argentina. Performance requirements (operational, life safety or collapse) are specified in terms of damage severity, using maximum displacements, inter-story drifts, local and global damage indices. Optimization objectives are minimum cost, minimum dimensions and/or minimum steel reinforcement.
