In the design of laminates for multiple components with good characteristics of manufacturability two main issues that must be addressed are as follows: definition of laminate zones (regions with independent elastic properties) and how to maintain adequate laminate continuity across distinct zones. The design of blended composite laminate panels is of discrete nature, typically associated to determining the thickness or number of layers in certain directions, making the genetic algorithm a natural tool for this solution. This work is intended to present a methodology for the structural optimization of a preliminary composite wing based on automatic zone modeling strategy followed by GA iterations.
The automatic zone modeling approach is based on MSC.NASTRAN topometric optimization capability in which the thicknesses of the composite layers [0º, ±45º, 90º] of each finite element are the design variables, with weight minimization as objective and buckling load factor as constraint. The zones are generated based on the gradient of the thickness and the ply percentage distribution of the panels. On a second step the resulting structure is the base of a GA optimization (with modeFRONTIER
), in which the population is generated such that the resulting thickness is within a range around the initial design. A constraint is also imposed to guarantee a fully blended laminate solution. A comparison study is done to evaluate the effects of the zone modeling and the blending on the optimal solution. This multi-step approach seems to be adequate do generate efficient optimum laminate design for a composite wing structure. The automatic zone modeling based on the topometric optimization seems to be an effective ingredient for designing components aiming to reduce both weight and complexity. From an initial design of 18 zones to a design of 14 zones the weight reduction was of 17%, showing that a smart selection of zones has a great advantage in weight saving. The GA allows for an easy implementation of the blending constraints with flexibility to change several parameters. Further analysis is required to reduce the sampling space, to guarantee that the optimal solution is found and to check how the blending constraints affect the weight.