Robust topology optimization of truss structures with random loading and material properties: A multiobjective perspective

In this paper an approach to robust topology optimization for truss-like structures with material and loading uncertainties, and discrete design variables, is investigated. Uncertainties on the loading direction and magnitude, as well as spatially correlated material stiffness, are included in the problem formulation using spectral stochastic finite element analysis, taking truss element length into account in the uncertainties on the material properties. In this way a more realistic random field representation of the material uncertainties can be achieved, compared to classical scalar random variable approaches. Additionally, a multiobjective approach is used to generate Pareto optimal solutions to the truss optimization problems. In an example it is shown how the mean and standard deviation of the compliance can be considered as separate objectives, avoiding the need for a combination factor which may be difficult to choose correctly in the classical single-objective approach. Further examples considering the mass and deflection of the structure are solved, showing the effects of the novel truss-like structure material modelling on the robust optimal designs.

Publication year:2015

Keywords:Robust optimization, structural Optimization, multiobjective genetic algorithms, topology optimization

CSS-citation score:2