Enhancing the optimization of material distributions in composite structures using gradient architectures

Abstract

Many structural components encounter service conditions and hence, material performance requirements, which vary from location to location within the component resulting in a composite structure. It has been shown that abrupt transitions in material properties within a composite structure can cause concentrations of deformation, which are mitigated by gradually varying the microstructure and/or composition of materials in a gradient architecture. Structural optimization techniques, such as the homogenization method, have yet to take full advantage of gradient architectures. Many of these structural optimization techniques employ robust mathematical techniques, such as genetic algorithms (GAs), with finite element simulations to optimize material distributions in composite structures through computationally intensive stochastic, global explorations of the design search space defined by a multitude of design variables associated with the discrete representation of the composite structure. Using gradient architectures, it is demonstrated that GAs can be enhanced for composite structures by constraining the design search space through a reduction in the number of design variables, thereby substantially reducing the computational effort. For more complex material distributions, a “material gradient optimization method” is proposed that produces multiple gradient architecture constraints with more optimal solutions than obtained without using these constraints, but whose uniqueness will be dependent upon the number of layers used in the finite element simulations.