SGC—a novel optimization method for the discrete fiber orientation of composites

Abstract

Using variable stiffness design to enhance the bearing capacity of composite structures has been an appealing strategy. In previous approaches for optimizing discrete fiber orientation, a penalty strategy is always adopted to drive the design variables to 0 or 1, which is unstable and does not guarantee the convergence of all fiber orientations. This paper proposes a novel optimization method termed sequential gradient chase (SGC) for optimizing discrete fiber orientations in composites based on stiffness matrix interpolation. We construct an optimization domain tightening criterion and a discrete direction search criterion. Constraints are incorporated into the solver by using an intriguing update rule for design variables, resulting in conversion from a constrained optimization problem to an unconstrained optimization problem. The suggested method is capable of achieving convergence of all fiber orientations without the use of any penalty measure and producing a clear fiber layout scheme devoid of gray-scale elements. Numerical examples demonstrate that the proposed method can be successfully applied to a variety of physical models for the calculation of in-plane stress and strain, plate bending, and shell torsion, and possesses a stable optimization ability and a high solving efficiency.