Multi-scale design of three dimensional woven composite automobile fender using modified particle swarm optimization algorithm

This paper proposes a multi-scale modeling and optimization strategy for the lightweight design of automobile fenders made of three dimensional (3D) woven composite. A Multi-scale modeling approach is developed to predict the mechanical behaviors of the 3D woven composite fender with various design parameters from material and structure. In order to alleviate the computational cost, Kriging modeling technique is adopted to generate the surrogate models of structural responses under multiple load cases. An optimization method integrating modified particle swarm optimization algorithm and Kriging surrogate model is employed to find the optimal combination of continuous and discrete design variables from different scales. The optimized automobile fender achieves a 20.65% weight reduction when satisfying all requirements on structural stiffness under multiple load cases.