Ultimate Confined Compressive Strength of Carbon Fiber-Reinforced Circular and Rectangular Concrete Column

Abstract

Retrofitting circular and rectangular concrete columns with Carbon Fiber Reinforced Polymer (CFRP) has been proven to be a method of increasing the ultimate confined compressive strength \( f^{\prime}_{cc} \). The present study uses a hybrid of Analytic Hierarchy Process (AHP) and Backpropagation Artificial Neural Networks (BP) that assesses the strength performance of circular and rectangular columns confined with CFRP, steel ties and a hybrid method using both materials. The data used for the modeling came from available references in journal articles. The process of AHP was first used in determining the best set of independent parameters which are as follows: unconfined concrete strength \( f^{\prime}_{co} \), ultimate CFRP strength \( f_{CFRP} \), volumetric ratio of CFRP \( \rho_{CFRP} \), and steel transversal strength (\( {\text{fsfs}} \)). Additional independent parameters such as the diameter (\( {\text{DD}} \)) and corner radius (\( {\text{rr}} \)) were also included since these were found to have the highest significance on circular columns and rectangular columns, respectively. After utilizing AHP, BP was used as a tool for modeling the \( f^{\prime}_{cc} \) as an output parameter. From this, two best models were produced for circular and rectangular columns out of twenty BP that were trained, validated and tested. These models were selected based on the Pearson’s correlation coefficient (\( {\text{RR}} \)) which performed better compared to previous existing models from past literature. The BP models attained were C_5_520_5 for circular columns and NC_5_415_6 for rectangular columns. Additionally, a parametric study of the BP models was done to determine the effect of the important input parameter \( f^{\prime}_{co} \) with the output parameter \( f^{\prime}_{cc} \).