Truss’ topology and shape optimization by genetic algorithms

In the paper the global optimization problem of truss systems is studied. The genetic algorithms are employed for the optimization. As the objective function the structure mass is treated; the constraints include equilibrium, local stability and other requirements. All the truss system characteristics needed for genetic algorithm are obtained via finite element solution. Topology optimization of truss system is performed using original modified genetic algorithm, while the shape optimization – using ordinary genetic algorithm. Numerical solutions are presented. The obtained solutions are compared with global extremes obtained using full search algorithm. All the numerical examples are solved using original software.

Experimental performance of optimized trusses

<p>A series of six 3D printed discretely optimized truss specimens and two warren truss specimens were experimentally loaded until failure. The results were compared to the theoretical failure loads and stresses determined using Maxwell’s Method. Each set of truss specimens were loaded in a simple span condition, with a point load applied at the center of the span. Each truss specimen was configured into pairs in order to prevent lateral torsional buckling (LTB) while testing. Strain, load, and displacement data was gathered for each truss specimen tested. These results were compared to the predicted results calculated by Maxwell’s theorem. Of the 6 specimens tested, all of the trusses failed within 1% - 20% of the analytical vales. The trends in the experimental results support efficacy of previously developed theories of optimized truss topology in order to increase strength and efficiency of lateral systems in high rise structures.</p>