Decomposition-based assembly synthesis based on structural considerations.

Most structural products have complex geometry to meet customer's demand of high functionality with enhanced structural stability. However, manufacturing those products in one piece requires sophisticated methods of manufacturing processes that will increase the total production cost. For this reason, most structural products are multi-component structures: they are made of number of components and these components are assembled into the final structure. Generally, designing a multi-component structural product requires designers to decompose overall product geometry in to the components set at some point during the design process. The components set decided there will be assembled into the final product. This research presents methods that systematically decompose product geometry into the components set considering the structural stiffness of the end product. Methods of parameterizing components set of a structure are represented by using the structural topology graph. The structural stiffness of the assembled structure for the given components set is evaluated as one of the objective functions by using the FE analysis. In order to model the assembled structure with different components set in FE, the methods of modeling the joinings between components are presented. In addition to the structural stiffness, the methods of evaluating the manufacturability and assembleability of given components set are discussed and used as the other objective functions. The Pareto optimal decompositions for this optimization problem are obtained by combining the FE analysis with a multi-objective genetic algorithm using graph-based crossover, exhibiting trade-offs among the objectives including the structural stiffness, component manufacturability (size and simplicity), and the assembleability.PhDApplied SciencesMechanical engineeringUniversity of Michigan, Horace H. Rackham School of Graduate Studieshttp://deepblue.lib.umich.edu/bitstream/2027.42/124251/2/3137885.pd