Structure assembling by stochastic topology optimization

Topology optimization is of significant importance to the design of truss- and grillage-like structures. Conventional topology optimization procedures are usually based on the ground structure approach. Starting from highly connected structures the uneconomical links are eliminated during the course of optimization. In this paper we show that, additionally, stochastic methods offer the possibility to build-up structures starting from simple initial configurations with few elements. Stochastic optimization methods (simulated annealing, evolutionary algorithms, random cost) are applied to the topology design problem on the basis of appropriate local structure variations. Results and performance comparisons are given.