Shear strength prediction of short circular reinforced-concrete columns using soft computing methods

Abstract

In this article, it has been aimed to predict the shear strength of short circular reinforced-concrete columns using the meta-heuristic algorithms. Based on the studies conducted so far, the parameters dominantly affecting the shear strength include axial force, longitudinal and transverse reinforcement, column dimension ratio, concrete compressive strength and ductility. In this respect, first, 200 numerical models of the short circular reinforced-concrete column incorporating various effective parameters so that a sufficient number of outputs could be provided, are analyzed by ABAQUS software to compute their shear strengths. Then, the gene expression programming and particle swarm optimization algorithms are employed to predict the shear strengths and by means of each algorithm, a relation was proposed accordingly. Then, using the experimental data, these relations are evaluated by comparing with those specified in ACI 318 and ASCE-ACI 426. The results indicate that the percentage of relative error between the experimental data and the values obtained from ACI 318 and ASCE-ACI 426 is respectively equal to 25% and 30%, which have been reduced to 13% and 9% through the gene expression programming and particle swarm optimization algorithms implying the satisfactory performance of these two algorithms. Finally, a comparison of the gene expression programming and particle swarm optimization is investigated in terms of convergence rate, degree of accuracy, and performance mechanism.