Design-driven harmony search (DDHS) in steel frame optimization

Recent research efforts in the field of steel frame optimization have demonstrated the successful implementation of stochastic methods such as genetic algorithms, simulated annealing, ant colony optimization, particle swarm, and harmony search in obtaining least-weight designs. The stochastic nature of these methods is effective in scouring the large design space of discrete variables intrinsic to steel frame optimization problems. However as the randomness of these methods can limit the optimality of the frame designs obtained, the frequency with which these designs are obtained, and the computational effort required to obtain them, no one algorithm has distinguished itself as most effective. This paper presents a design-driven harmony search (DDHS) algorithm for optimization of steel moment frames. Based on the harmony search method, DDHS incorporates intelligence in the stochastic search DDHS by using constraint satisfaction or violation data from previous trial solutions to steer the optimization in the direction of larger or smaller sections as sorted by an appropriate section parameter. Thus, DDHS generates random trial solutions within intelligently specified search neighborhoods, resulting in improved performance in steel frame optimization as measured by the optimality of the designs obtained, the consistency with which successive runs obtain these designs, and the number of structural analyses required to obtain them. This paper demonstrates the efficient performance of DDHS as applied to three benchmark problems – a 15-member, a 105-member, and a 168-member planar frame optimization.