Computationally efficient discrete sizing of steel frames via guided stochastic search heuristic

• A guided stochastic search (GSS) technique is proposed for sizing of steel frames.
• The principle of virtual work is utilized for handling the displacement constraints.
• The GSS locates promising solutions using an insignificant computational effort.

Recently a design-driven heuristic approach named guided stochastic search (GSS) technique has been developed by the authors as a computationally efficient method for discrete sizing optimization of steel trusses. In this study, an extension and reformulation of the GSS technique are proposed for its application to problems from discrete sizing optimization of steel frames. In the GSS, the well-known principle of virtual work as well as the information attained in the structural analysis and design stages are used together to guide the optimization process. A design wise strategy is employed in the technique where resizing of members is performed with respect to their role in satisfying strength and displacement constraints. The performance of the GSS is investigated through optimum design of four steel frame structures according to AISC-LRFD specifications. The numerical results obtained demonstrate that the GSS can be employed as a computationally efficient design optimization tool for practical sizing optimization of steel frames.