Modeling ultimate deformation capacity of RC columns using artificial neural networks

Abstract

This paper aims to explore the feasibility of the potential use of artificial neural networks (ANN) in deformation estimates of RC columns whose behaviour is dominated by flexural failure. Experimental data of 237 rectangular columns from an existing database were used to develop an ANN model. The input parameters were selected based on past studies such as aspect ratio, longitudinal reinforcement ratio, yield strength of longitudinal reinforcement, uniaxial (cylindrical) concrete strength, yield strength of transverse reinforcement, transverse steel spacing, ratio of transverse steel parallel to the direction of loading, axial load ratio, and confinement effectiveness factor. Ultimate displacement estimates of reinforced concrete columns by the ANN model were compared to the existing semi-empirical and empirical models. The ANN model was found to perform well. The promising results have shown the feasibility of using ANN models for deformation estimates of RC columns.