DISCRETE-VARIABLE OPTIMAL STRUCTURAL DESIGN USING TABU SEARCH

Tabu search is a discrete-variable optimization algorithm with the ability to avoid entrapment by local optima and hence continue searching for a global optimum. In this paper tabu search is applied to the optimal structural design, in terms of weight minimization, of two standard (test) structural configurations; a 10-bar planar truss and a 25-bar space truss. The design variables are the cross-sectional areas of the bars, which take discrete values. An implementation of tabu search in a structural design context is presented which features beta-depth neighbourhoods and a search back-track facility. Investigations show that tabu search may readily cope with problem formulations that include buckling, gravity effects and design variables with realistic values. Furthermore, compared to previous research, superior (i.e. lower) minimum weights may be obtained. It is shown that tabu search is a technically viable technique for use in optimal structural design, although, for practical use, current (significant) execution times may inhibit its utilisation.