UCLA

This dissertation addresses a wide spectrum of topics associated with resilience-based seismic evaluation and design of reinforced concrete structures. First, a comprehensive framework was developed enabling users to select, scale, and modify mainshock and aftershock ground motions (GMs) based on a set of different criteria. Various hazard-consistent target intensity measure metrics were utilized based on a set of conditioning criteria (IMi's). Then, a statistical approach was utilized to pull realization samples from a multivariate distribution of multiple intensity measures (IMi's) using both Monte-Carlo (MC) and Latin Hypercube Sampling (LHS) techniques. A comprehensive database of seismic records—namely, the PEER NGA-WEST2 database—was utilized to select those records whose IMi' s match the conditioning targets by assigning a different set of weights to different IMi's in a least-squares sense.

The effects of different GM selection strategies on a range of engineering demand parameters (EDPs) were investigated for a pair of 4-, 8- and 12-story ductile and non-ductile reinforced concrete (RC) buildings. In such studies, a very large number of nonlinear response history analyses (NRHAs) is unavoidable. Therefore, the analyses were performed using a parallel computing approach, which was specifically designed for the task at hand and carried out at the Texas Advanced Computing Center’s “Stampede2” supercomputer.

Another topic that was addressed in this dissertation has been the development of a novel framework to select earthquake records based on a spectral shape matching approach. The effects of different GM selection strategies based on a pre-existing spectral shape matching approach—namely, the response spectrum matching method—versus the newer approach developed in the present study was studied. The same ductile and non-ductile RC buildings mentioned above were utilized for this task and a variety of damage limit states (including the collapse) were used for comparison of fragility functions obtained using the two approaches.

Finally, an optimization framework was developed to reduce the effects of epistemic uncertainties associated with wide range of structural modeling parameters, on the probabilistic seismic responses of RC structures. To this end, a non-dominated sorting genetic algorithm (NSGA-II) was integrated with OpenSeesMP to determine optimal values of several design variables that minimize the median peak inter-story drift ratios (IDRs) at two different performance levels—namely, Immediate Occupancy (IO) and Collapse Prevention (CP)—simultaneously.