Feature-driven method for structural topology optimization

Abstract

We propose a new method by which features are regarded as the basic design primitives for structural topology optimization. This method originates from the concept of feature-based design in CAD community. Implicit level-set functions such as Max-Min function, R function, KS function or Ricci function are used to describe the feature model of a structure. Structural topology optimization is converted into layout and shape optimization of features within the prescribed design domain. This method makes it possible to avoid the drawbacks of traditional topology optimization methods that seek the material distribution only for the optimization of mechanical performance and thus result in abstract topology optimization results with the lack of engineering features. The minimum compliance problem subjected to volume constraint is considered. Structural response analysis and sensitivity analysis are formulated within the framework of fixed mesh. It is shown that the sensitivity accuracy is independent of the mathematical expressions of the related level-set function. Finally, a holed cantilever beam structure and aero-engine bracket structure are studied as numerical examples. By means of super-elliptical/super-ellipsoidal features, the effectiveness and advantage of the developed method are demonstrated.