Optimum topological design of simply supported composite stiffened panels via genetic algorithms

Abstract

In the present paper the topological optimal design of isotropic/orthotropic thin structures performed via genetic algorithms is shown. Examples involving structural weight minimization under compressive load or buckling load maximization are presented. A modified finite strip method was developed and used to analyze parametric structures arranged in form of plates or stiffened panels with almost arbitrary cross-section shapes. Specific design variables were defined to assure a robust control over geometrical and topological features. In particular, a semi-analytical formulation for the determination of eigenvalues and eigenvectors was adopted in order to reduce computational efforts requested by the optimization task. A mesh-independent solver, involving a reduced number of degrees of freedom, was implemented and interfaced with a genetic optimizer for the purpose. The optimization procedure was based on a specific bit-masking oriented genetic algorithm, able to handle in parallel different genetic operators expressly conceived to process with proper metrics discrete and continuous design variables. As preliminary example, the buckling load maximization of a metallic plate with an arbitrary grid-shaped cross-section is described first. Then a topological optimization concerning the weight minimization of a composite stiffened panel subject to constraint about buckling load is illustrated and discussed in detail about parametric model definition and genetic procedure.