Abstract
A multi-level optimization process has been developed specifically to design thick laminated composite structures. The large number of design variables present in this type of structure makes such a process a necessity so that reasonable computational efficiency can be achieved. At the first level, the in-plane response of a global model is optimized using the stiffnesses of an equivalent orthotropic laminate as the design variables. At the second level, stiffnesses of sublaminates are used to optimize the out-of-plane response while maintaining the in-plane performance obtained in the first level. Finally, a detailed model of a critical local region is developed based on the results from the second level. This local model is used to optimize the laminate strength using ply orientations as design variables while maintaining the in-plane and out-of-plane performance obtained in the first two levels. The optimization process has been applied to a conceptual model of a composite femoral component for total hip joint replacement. Three-dimensional finite element models are used for the analysis at each of the three levels. The required computational efficiency is achieved by controlling the complexity of the models, the number of design variables, and the nonlinearity of the optimization problem at each level.