Artificial Neural Network Model for Shear Strength of Fibrous RC Beams

Abstract

This study has investigated the modeling of shear strength using the artificial neural network (ANN) approach. The Results of 128 samples of steel fiber reinforced concrete (SFRC) beams without stirrups were collected gathered and used to generate a four-layer feed forward neural network using the back-propagation learning algorithm available in the MATLAB program. Nine parameters for SFRC beams, namely, beam height, beam depth, beam width, steel cross-sectional area, shear span-to-depth ratio, concrete compressive strength, volume fraction, fiber length, and fiber diameter, were considered as input variables for the ANN. ANN output representing the shear strength were compared with those observed experimentally using regression analysis approach. Results indicated that the ANN modeling technique is effective in simulating the behavior of SFRC beams. In addition, a parametric study shows that shear span, compressive strength of concrete, volume fraction, and fiber length are playing the major role in the behavior of SFRC beams.