Keywords: robust design optimization, optimization under uncertainty, uncertainty, genetic algorithm, mean-variance.

There are numerous uncertainties in the structural design, such as the randomness or variation in the loading, structural parameters, geometric parameters, operation conditions, etc. The Robust Design Optimization (RDO) methodology aims to determine an optimal solution corresponding to the insensitive system performance when subjected to these uncertainties. Available RDO approaches can effectively take into account these uncertainties. Still, accuracy and computational cost in evaluating the mean and variance (robustness measures) are prohibitive for designing complex and realistic structural systems. To obliviate this limitation, a novel Stochastic simulation-based approach is proposed in this paper. The newly developed approach is constructed based on the 'augmented optimization problem,' in which design variables are artificially considered as an uncertain parameters. Furthermore, for optimization, a two-stage approach is adopted. Firstly, the design space size is reduced by formulating an unconstraint optimization approach followed by any standard random search method (KN direct search method) to determine the optimal solution within the reduced design space. As the mean and variance frequently conflict with each other, so to obtain the Pareto optimum, a linear scalarized objective function is adopted. Three optimization problems: quadratic function and six-hump camel-back function, and 10-bar truss structure subjected to uncertain loading and uncertain material properties are solved with the proposed approach to demonstrate the efficiency of the proposed approach. The results obtained indicate that the proposed approach is as accurate as of the conventional Monte Carlo simulation approach. This paper allows the designers to design insensitive structure systems. Moreover, the proposed RDO approach is general and not limited to the civil structures only, but it can also be enforced in the design of any realistic linear/nonlinear systems.

download the full-text of this paper (PDF, 10 pages, 691 Kb)

go to the previous paper
go to the next paper
return to the table of contents
return to the volume description