Keywords: composite, blending, stacking sequence design, genetic algorithm.

Genetic algorithms (GA) are widely used for discrete stacking sequence design of composite laminates. Design of a large structure entails subdivision of the problem into several smaller local panel design problems. Independent design of local laminate leads to stacking sequence mismatch between adjacent panels leading to a blending issue.

Soremekun et al. [1] developed a two step adhoc design methodology to impose blending via a sublaminate zone approach. Adams et al. [2] developed a guide based GA to impose blending. The panel stacking sequence is formed by taking a certain number of plies from the same guide, which is determined locally by one dimensional optimization. Local panel loads are responses of the whole structure and changing the number of plies at any panel will result in internal load redistribution requiring expensive global analyses. Seresta et al. [3] successfully used a guide based GA with local improvement operator for wingbox design. The local improvement operator deletes or adds one ply per panel assuming that it would not significantly change the internal load distribution. However, above approaches becomes inapplicable in absence of local constraints or in a situation where both global and local constraints are competing with each other to dictate overall design of the structure or if local improvement operator results in significant internal load redistribution.

A standard GA works on a number of individuals (designs) collectively called population. Each individual has a chromosome (array) of user defined lengths where each gene (element of array) represents a discrete ply angle by an integer. In this paper, we propose a multi-chromosomal GA to impose blending. The first chromosome will represent the stacking sequence of the guide. Then, for each panel a binary chromosome is added where 1 indicates the corresponding ply in the first chromosome is on and 0 indicates off. The blending is imposed by forcing all the on plies to the beginning of the chromosome. This representation uses standard GA crossover operators to arrive at the minimum number of plies without any local operation. A single panel optimization is carried out to validate the proposed represention with traditional GA representation. A fifteen panel design problem is formulated such that each panel has the same dimensions and inplane loads. One blended optimum solution is single panel optimum design being repeated for all the panels. The proposed scheme found the absolute minimum easily. Third, an eighteen panel horse-shoe configuration design problem is solved and compared with published results [1].

1

G. Soremekun, Z. Gürdal, C. Kassapoglou, D. Toni, "Stacking Sequence Blending of Multiple Composite Laminates using Genetic Algorithm", Composite Structures, 56(1), 53-62, 2002. doi:10.1016/S0263-8223(01)00185-4

2

D.B. Adams, L.T. Watson, Z. Gürdal, C.M. Anderson-Cook, "Genetic Algorithm Optimization and Blending of Composite Laminates by Locally Reducing Laminate Thickness", Advances in Engineering Software, 35(1), 35-45, 2004. doi:10.1016/j.advengsoft.2003.09.001

3

O. Seresta, Z. Gürdal, D.B. Adams, L.T. Watson, "Optimal Design of Composite Wing Structures with Blended Laminates", Composites Part B. Engineering, 38(4), 469-480, 2007. doi:10.1016/j.compositesb.2006.08.005

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £145 +P&P)