Multi-scale deterministic optimisation of blended composite structures: case study of a box-wing

Abstract

This work presents a multi-scale design methodology for the deterministic optimisation of thin-walled composite structures integrating a global-local approach for the assessment of the buckling strength and a dedicated strategy to recover blended stacking sequences. The methodology is based on the multi-scale two-level optimisation strategy for anisotropic materials and structures. In the first step, focused on the macroscopic scale, several design requirements are included in the problem formulation: lightness, feasibility, manufacturing, blending, buckling failure, static failure and stiffness. The second step, which focuses on the laminate mesoscopic scale, deals with the recovery of blended stacking sequences, for the structure at hand, matching the optimal geometric and elastic properties determined in the first step. As a case study, the unconventional PrandtlPlane box-wing system is used to show the effectiveness of the proposed design methodology.

A novel parallel method for layup optimization of composite structures with ply drop-offs

“…symmetry, balance, contiguity, disorientation, minimum percentage, and damage tolerance constraints). In order to obtain the layups for large composite structures with ply dropoffs, the DLBB method which incorporates a dummy layerwise technique with the branch and bound method is developed [20]. This DLBB method performs a logical search rather than a stochastic search conducted by a heuristic algorithm.…”

“…The dummy layers embedded in the layerwise technique do not contribute to the stiffness but are used to impose the blending constraint. An illustrative example is given in [20] .…”

“…Although GAs have become the most popular technique for this discrete layup optimization problem, the disadvantages of conducting a random search and the fact that the predetermined parameters have significant influence on optimization performance cannot be ignored. Several alternative methods have therefore been developed, such as the constraint satisfaction programming method (CSP) [16,17] based on a branch and bound tree, the mixed integer linear programming (MILP) method [18], the multi-scale two-level optimization strategy (MS2LOS) [19][20][21], and topology inspired methods such as the discrete material and thickness optimization method (DMTO) [22,23]. A detailed review of the methods discussed above can be found in [24].…”

Design optimization of laminated composite structures using artificial neural network and genetic algorithm

“…Montemurro et al [43,44,50] conducted a series of research on polar parameters which are directly related to the symmetries of the tensor. Based on these polar parameters, a multi-scale two-level optimization strategy (MS2LOS) was developed for optimizing flat laminates [51], stiffened laminates [52,53], blended laminates [52,54,55], and variable angle tow laminates [56,57].…”

Stacking sequence optimisation of an aircraft wing skin

“…Physical constraint violations are observed for the discretised design even when applying blending constraints in the first stage of the optimisation. Recently, Scardaoni et al (2022) applied a two stage process using polar parameters to the sizing of a box-wing. Buckling constraint violations occurred in some of the sub-components studied.…”