A developed design optimization model for semi-rigid steel frames using teaching-learning-based optimization and genetic algorithms

This paper proposes a developed optimization model for steel frames with semi-rigid beam-to-column connections and fixed bases using teaching-learning-based optimization (TLBO) and genetic algorithm (GA) techniques. This method uses rotational deformations of frame members ends as an optimization variable to simultaneously obtain the optimum cross-sections and the most suitable beam-to-column connection type. The total cost of members plus connections cost of the frame are minimized. Frye and Morris (1975) polynomial model is used for modeling nonlinearity of semi-rigid connections, and the P-∆ effect and geometric nonlinearity are considered through a stepped analysis process. The stress and displacement constraints of AISC-LRFD (2016) specifications, along with size fitting constraints, are considered in the design procedure. The developed model is applied to three benchmark steel frames, and the results are compared with previous literature results. The comparisons show that developed model using both LTBO and GA achieves better results than previous approaches in the literature.