Global laminate optimization on geometrically partitioned shell structures

A method aimed at the optimization of locally varying laminates is investigated. The structure is partitioned into geometrical sections. These sections are covered by global plies. A variable-length representation scheme for an evolutionary algorithm is developed. This scheme encodes the number of global plies, their thickness, material, and orientation. A set of genetic variation operators tailored to this particular representation is introduced. Sensitivity information assists the genetic search in the placement of reinforcements and optimization of ply angles. The method is investigated on two benchmark applications. There it is able to find significant improvements. A case study of an airplane's side rudder illustrates the applicability of the method to typical engineering problems.