Open AccessPredicting Shear Strength of RC Columns Using Artificial Neural Networks
Author(s) -
A Said,
N Gordon
Publication year - 2019
Publication title -
journal of building materials and structures
Language(s) - English
Resource type - Journals
eISSN - 2600-6936
pISSN - 2353-0057
DOI - 10.34118/jbms.v6i2.69
Subject(s) - artificial neural network , shear (geology) , structural engineering , shear strength (soil) , reinforced concrete , column (typography) , geotechnical engineering , computer science , mathematics , materials science , geology , engineering , artificial intelligence , composite material , soil science , connection (principal bundle) , soil water
A primary objective in the seismic design of structures is to ensure that the capacity of individual members of a structure exceeds the associated demands. For reinforced concrete (RC) columns, several parameters involving steel and concrete material properties control behavior and strength. Furthermore, it is unrealistic to simply consider the shear strength calculation as the sum of concrete and steel contributions while accounting for axial force when, in fact, all those parameters are interacting. Consequently, it is challenging to reasonably estimate the shear capacity of a column while accounting for all the factors. This study investigates the viability of using artificial neural networks (ANN) to estimate the shear capacity of RC columns. Results from ANN are compared with both experimental values and calculated values, using semi-empirical and empirical formulas from the literature. Results show that ANNs are significantly accurate in predicting shear strength when trained with accurate experimental results, and meet or exceed the performance of existing empirical formulas. Accordingly, ANNs could be used in the future for analytical predictions of shear strength of RC members.