Abstract
This paper presents a method for synthesizing multicomponent beam structural assemblies with maximum structural performance and manufacturability. The problem is posed as a relaxation of decomposition-based assembly synthesis, where both topology and decomposition of a structure are regarded as variables over a ground structure with nonoverlapping beams. A multiobjective genetic algorithm with graph-based crossover, coupled with FEM analyses, is used to obtain Pareto optimal solutions to this problem, exhibiting trade-offs among structural stiffness, total weight, component manufacturability (size and simplicity), and the number of joints. Case studies with a cantilever and a simplified automotive floor frame are presented, and representative designs in the Pareto front are examined for the trade-offs among the multiple criteria.