Optimum design of planar steel frames under LRFD-AISC specifications using a step-by-step descent algorithm

This paper presents a novel descent algorithm based on the step-by-step iterative principle, applied to the optimum design of steel frames. The search consists on finding the direction which decreases the structural weight most quickly. As the design problem includes discrete variables, the optimum is found by evaluating the structural weight gradient step by step. The step size is controlled in such a way that convergence towards infeasible or suboptimal solutions is avoided. By properly choosing the initial solution, it is possible to increase the efficiency and the convergence speed of the algorithm. Many strategies, for the choice of initial design point, by making use of engineering intuitions or using optimized design obtained by other algorithms are discussed. Furthermore, it is confirmed in this study that the proposed algorithm can be used to improve optimum designs found by metaheuristic algorithms. The optimization results, relative to several weight minimizations problems of benchmark planar steel frames designed according to Load and Resistance Factor Design, American Institute of Steel Construction (LRFD-AISC) specifications, are compared to those obtained by different optimization methods. The comparison proves the efficiency and robustness as well as the prompt of convergence of the proposed descent algorithm developed in this paper.