UC Berkeley

This dissertation investigates the issue of selecting and scaling ground motions as input

excitations for response history analyses of buildings in performance-based earthquake engineering. Many ground motion selection and modication (GMSM) procedures have been developed to select ground motions for a wide variety of objectives. In this research, we focus on the selection and scaling of single, horizontal components of ground motion for estimating seismic demand hazard curves (SDHCs) of multistory frames at a given site.

In Chapter 2, a framework is developed for evaluating GMSM procedures in their ability to provide accurate estimates of the SDHC. The notion of a benchmark SDHC is introduced, enabling biases caused by GMSM procedures to be isolated from other sources of bias. More importantly, the ability to quantify bias facilitates the identication of intensity measures (IMs) that are sufficient. However, this approach is limited by the availability of recorded ground motions and of prediction models for engineering demand parameters (EDPs) of structures.

The framework developed in Chapter 2 is applied to synthetic ground motions in Chapter

3, where biases in estimates of SDHCs caused by GMSM procedures can be estimated for any structural system and any EDP. However, the use of synthetic ground motions gives rise to the issue of developing benchmark-consistent ground motion prediction models. Based on the results from Chapters 2-3, it is hypothesised that the potential bias in any SDHC estimate is caused directly by two important properties of the particular selection of ground motions: (i) hazard consistency, and (ii) IM sufficiency.

A novel ground motion selection procedure, rooted in the theory of Importance Sampling,

is developed in Chapter 4 that allows: (i) hazard consistency of the selected motions to be directly enforced for a user-specified collection of IMs, and (ii) SDHCs of a structure to

be estimated from a single ensemble of ground motions, with the option of avoiding record scaling altogether. This procedure, together with two other contemporary GMSM procedures -- (i) "exact" Conditional Spectrum and (ii) Generalized Conditional Intensity Measure -- are evaluated in Chapters 5-6 for a variety of structural systems and EDPs at a specified site. In these chapters, the amount of effort involved in implementing these procedures for estimating SDHCs is summarized in a step-by-step form, and the magnitude of biases caused by these procedures are documented.