Design of composites using a generic unit cell model coupled with a hybrid genetic algorithm

Structural optimization for any kind of laminate or weave composite material is a computationally expensive and intense process. In this study, two research objectives are addressed: (i) to introduce architecturally independent, adaptive unit cells and representative volume elements, and (ii) to use a hybrid genetic algorithm procedure for composite structural optimization. By employing this approach, state parameters and non-uniform homogenization are adopted, resulting in high prediction accuracy and low computational cost without loss of the local (micro) detail of the composite structure. Furthermore, genetic algorithm optimization procedures necessitated in determining the type of reinforcing material, architecture (stacking sequence of laminates, woven type of fabric), shape and size of the final product in order to meet certain design deliverables are explored. The prospective unit cells contain the information required to form a laminate, or woven, composite and record different modes of crack propagation. Unit cells are built such that they become discrete entities, easily adopted to build a geometric architecture. The attention is primarily focused on the applicability of genetic algorithms in promoting improved analytical and finite element computations of specific objective functions. This paper demonstrates the process of building and combining UCs to form any desired structure and architecture, and illustrates its implementation and effectiveness with the developed hybrid genetic algorithm with memory model. To this end, issues such as efficient coding, selection procedures and effects of anisotropic material properties are particularly emphasized.