Flexural capacity estimation of FRP reinforced T-shaped concrete beams via soft computing techniques

Abstract

This paper discusses a framework for predicting the flexural strength of prestressed and non-prestressed FRP reinforced T-shaped concrete beams using soft computing techniques. An analysis of 83 tests performed on T-beams of varying widths has been conducted for this purpose with different widths of compressive face, beam depth, compressive strength of concrete, area of prestressed and non-prestressed FRP bars, elasticity modulus of prestressed and non-prestressed FRP bars, and the ultimate tensile strength of prestressed and non-prestressed FRP bars. By analyzing the data using two soft computing techniques, named artificial neural networks (ANN) and gene expression programming (GEP), the fundamental parameters affecting the flexural performance of prestressed and non-prestressed FRP reinforced T-shaped beams were identified. The results showed that although the proposed ANN model outperformed the GEP model with higher values of R and lower error values, the closed-form equation of the GEP model can provide a simple way to predict the effect of input parameters on flexural strength as the output. The sensitivity analysis results revealed the most influential input parameters in ANN and GEP models are respectively the beam depth and elasticity modulus of FRP bars.