Date of Award

8-2013

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Legacy Department

Civil Engineering

Advisor

Atamturktur, Sez

Committee Member

Pang , Weichiang

Committee Member

Huang , Yongxi

Abstract

In structural design, if robustness is overlooked in design process, the acquired design is likely to have large variation in its performance due to uncertainties. In that there is no design specification that explicitly considers the robustness against such uncertainties, this dissertation elucidates design methodologies for use in selecting the optimal design parameters to minimize the effect of the hard to reduce or irreducible uncertainties on structural performance, i.e., maximizing the robustness. However, due to limited resources, structural designs are also constrained by available resources and budget. Consequently, in that a tradeoff relationship exists between the robustness and cost (i.e. the more robust the design, the greater the cost). Therefore, optimizing robustness and cost are conflicting objectives. Thus, an explicit consideration of this tradeoff relationship between robustness and cost necessitates formulating a robust design optimization (RDO) as a multi-objective optimization problem, with robustness, cost and other metrics of interest as objectives. The outcome of an RDO is a Pareto front, the optimum set reflecting tradeoff between competing objectives, in which the acquired Pareto front designs are more robust and more economical than all other designs. Furthermore, with the acquired Pareto front, a more informed decision can be achieved. This dissertation applies proposed RDO to two distinct problems, in which the reliability index in foundation design and seismic demand in steel moment resisting frame design are the considered performance measures. These measures, in turn, lead to a confidence level based robust design optimization, second order reliability based design optimization, and robust design optimization of steel moment resisting frame.

Share

COinS