PREDICTION OF PERCENTAGE OF STEEL FOR SINGLY AND DOUBLY REINFORCED BEAM SECTIONS USING ARTIFICIAL NEURAL NETWORK

Abstract

This research paper is about an extensive computational work carried out to compare prediction of percentage of tension steel for singly reinforced beam sections, percentage of tension and compression steel for doubly reinforced beam sections using design charts, design tables developed for a wide range of Mu/bd2 for various grades of concrete and steel and these values are compared with Artificial Neural Network (ANN) predicted values. Various grades of concretes used for generating percentage of steel values for singly and doubly reinforced beam sections are M20, M25, M30, M35, M40 and M45. Various yield strength values of steel considered are 250 N/mm2, 415 N/mm2, 500 N/ mm2. An application of ANN is made to predict percentage of steel values for concrete grades M20 to M35 and yield strengths of steel Fe250, Fe415 and Fe500 for a wide range of Mu/bd2 values. An increment of 0.01 is adopted for successive Mu/bd2 values. Both the charts and neural network is used to predict the percentage of steel for various concrete and steel grades. Inputs for singly reinforced and doubly reinforced beam sections include grade of concrete and Mu/bd2 values. Outputs include percentage of tension steel for Fe250, Fe 415 and Fe500. Outputs comprise of percentage of compression and tension steel for various d’/d values of 0.05, 0.1, 0.15, 0.2. A five layer neural network for combined data predicted the values with an error range of 6.93% to 1.413%.