Topology redesign for performance by large admissible perturbations

Abstract

A methodology is developed for optimal structural topology design subject to several performance constraints. Eight-node solid elements are used to model the initial structure, which is a uniform solid block satisfying the boundary conditions and subjected to external loading. The Young modulus of each solid element or group of elements is used as redesign variable. A minimum change function is used as an optimality criterion. Performance constraints include static displacements, natural frequencies, forced response amplitudes, and static stresses. These constraints are treated by the large admissible perturbation methodology which makes it possible to achieve the performance objectives incrementally without trial and error or repetitive finite element analyses for changes in the order of 100–300%. Thus, the optimal topology is reached in about four to five iterations, where each iteration includes one finite element analysis and setting of an upper limit for the value of the modulus of elasticity to produce a manufacturable structure. Several numerical applications are presented using three different benchmark structures to demonstrate the methodology and the impact of performance constraints on the generated topology.