Performance-based optimum seismic design of steel frames using the hybrid learning based-jaya algorithm

Metaheuristic optimization methods have been implemented as efficient tools for solving complicated optimization problems over the last three decades. Teaching-learning-based optimization (TLBO) and JA algorithm (JA) are among the most popular metaheuristic methods. TLBO mimics the teaching and learning process in a class in which learners are interacted with the teacher and themselves. JA implements the simple strategy that it tries to approach the best solution and moves away from the worst solution during the search process. The performance-based optimum seismic design of steel frames is one of the most complicated and time-consuming optimization problems. In this study, TLBO, modified TLBO (MTLBO), JA, modified JA (MJA) and the hybrid learning based-JA algorithm (HLBJA), which combines the JA and the learning phase of TLBO, are proposed for the performance-based optimum seismic design of planar steel frames. Moreover, the proposed methods are also implemented for optimum seismic design of spatial frames. The objective of performance-based optimum seismic design of planar steel frames is to minimize the weight of frames under the interstory drift and strength constraints. Three planar steel frames previously designed by particle swarm optimization (PSO), improved quantum particle swarm optimization (IQPSO), firefly (FA) and modified firefly algorithm (MFA), TLBO and JA are used to demonstrate the competence of proposed methods. The pushover analyses for performance based optimum seismic design of planar frames are performed by OPENSEES structural analysis programme while seismic analyses of spatial frames are perfomed by SAP2000 structural analyses software. The interstory drift, strength constraints and geometric-size design constraints are used for the optimum seismic design of spatial steel frames. Optimum seismic design of 325, 504 and 499-member space steel frames are realized in this thesis. The results obtained by the proposed methods are compared to each other and other optimization methods in terms of optimum weight and number of structural analyses. Several statistical parameters obtained from different executions of proposed methods such as average weight, worst weight and standard deviation are also presented in the design examples. The comparisons prove that HLBJA could find lighter designs with less computational effort than MTLBO, MJA, JA, TLBO and other methods in the literature.