Genetic algorithm in structural shape design and optimization

The design and optimization of skeletal structures has traditionally required the use of conceptual designs, which strongly influence the generated designs. If the design could be freed from the need for conceptual designs, it would present a potential for generating new and innovative designs, especially when faced with more complex design problems. A methodology using such a design approach is presented in this paper, which uses Genetic Algorithm to evolve designs which are free to assume any geometry and topology. It is capable of addressing single or multiple loadings, and plane or space structures. The sizing, configurational and topological aspects are simultaneously addressed. Discrete member sizes are used. A string representation scheme, a generalized penalty function and a fitness function are introduced. The methodology is illustrated by applying to the minimum weight design of a single span plane truss, and the result described.