Predicting the Torsional Strength of Reinforced Concrete Beams Strengthened with FRP Sheets in terms of Artificial Neural Networks

Reinforced concrete (RC) structural elements such as peripheral beams, beams supporting the canopy, ring beams of circular slabs are common members which could encounter torsional moments. One of the up-to-date and modern techniques for strengthening such beams is using FRP sheets. The increase of service loads, degradation of mechanical properties and updates in code regulations would cause to the need for retrofit. Using FRP sheets as a strengthening technique would increase the flexure, shear and torsion capacity. It would also change the failure mode and/or plane. In this article, torsional strength prediction of RC beams strengthened by FRP using artificial neural networks has been investigated. Input parameters of ANNs are RC beam width, height, FRP sheet thickness, modulus of elasticity of FRP sheet, yielding stress of longitudinal and transverse steels, the compressive strength of concrete, the effective width of shear strips along beam length, center-to-center space of FRP strips, the angle of wrapping and number of FRP layers. The target parameter is the capacity of the beam in bearing the torsional moment. The results indicate that the idealized neural network having definite number of neurons in hidden layer, can predict the torsional strength of RC beams externally bonded with FRP with a high degree of precision. Considering the mentioned precision, the method could be an efficient alternative for time-consuming and highly cost experimental programs.