Evaluating and proposing models of circular concrete columns confined with different FRP composites

Abstract

Commonly used fiber-reinforced polymer (FRP) includes Carbon, Glass, and Aramid FRP composites. Meanwhile, some new FRPs such as PBO (Polypara-phenylene-Benzo-bis-Oxazole), PET (Polyethylene Terephthalate/Polyester), Dyneema, and Basalt have been gradually applied in recent years. Over the past 20 years, there has been extensive research on modeling of stress–strain response of confined concrete using the common types of FRP. In this study, most popular and recent models are investigated to evaluate their general practical application in predicting the response of FRP-confined concrete with strain-hardening performance without any restriction on the fiber used. The aim of the study is twofold. In case of different types of FRP composites, providing equivalent confinement modulus (lateral stiffness), five models are employed to find the FRP-confined concrete stress–strain relationship of three scale-model circular columns. Second ascending part (second stiffness) of the stress–strain relationship of FRP-confined concrete with strain-hardening performance is evaluated in the light of available database from the existing literature using these analytical models. The results showed that the examined models do not satisfy the fact that the slope of the second ascending branch of the stress–strain curve of FRP-confined concrete is independent of its types, provided the design confinement modulus is the same. A comparison of predicted values of the second stiffness with the collected test results of 257 cylinder specimens confined with the common types of FRP composites revealed the necessity for a more accurate model. Based on the discussion of the features and accuracy of these models, a model considering the effect of FRP lateral stiffness is proposed. Because of the variability observed in the test data, however, it appears impossible to develop simple empirical models based on the current database with less than approximately 27% mean absolute error for the second stiffness, and 16% mean absolute error for ultimate strength.