Topology design of planar cross-sections with a genetic algorithm: Part 2 - Bending, torsion and combined loading applications

A topology design approach provides freedom to design a structure of any size, shape and connectivity within a defined domain. The binary chromosome storage and global search capabilities of the Genetic Algorithm (GA) make it an excellent tool for structural topology applications. A companion paper (Fanjoy, D. W. and Crossley, W. A. (2000), Topology design of planar cross-sections with a genetic algorithm: Part I-overcoming the obstacles. Engineering Optimization, (this issue)) investigated and demonstrated a successful GA approach for topology design of planar cross-sections subject to bending, torsion and combined loading. In this paper, the structural topology design applications are investigated in greater detail. The finite element method used for section analysis is described, Several applications are presented, highlighting different features of the GA/finite-element method combination for topology design. Designs generated for simple bending and torsion problems are presented first, with discussion and comparison to theoretical or known solutions. A combined loading application is presented, and the generated solution is compared to a baseline design. Finally, a multiobjective problem demonstrates the ability of the GA to generate a family of design trade-off solutions; a capability not normally associated with topology design approaches. The GA method for topology design presented here shows promise for application to a wider range of structural design problems than previous GA approaches.