Axial compressive behavior of actively confined and FRP-confined concretes.

Abstract

Since the 1920s, a significant research effort has been dedicated to the understanding of the improved compressive behavior of concrete under lateral confinement. Upon the introduction of fiber-reinforced polymer (FRP) composites to the construction industry, the use of FRP as confinement material has received much attention. To that end, a great number of studies have been conducted in the past two decades on the axial compressive behavior of unconfined, actively confined and FRP-confined concretes, resulting in the development of over 110 stress-strain models. These models are classified into four broad categories, namely design-oriented, analysis-oriented, evolutionary algorithm, and finite element models. In the present study, existing models in each category were carefully reviewed and assessed using comprehensive experimental test databases assembled through an extensive review of the literature. The databases cover more than 7000 test results of unconfined, actively confined, and FRP-confined concrete specimens from 500 studies. A close examination of the assessment results has led to a number of important findings on factors influencing the performances of existing models. For each model category possible areas for further improvement were identified and new models were proposed. First, an empirical model in simple closed-form expressions that are suitable for engineering design purpose was developed using the database of FRP-confined concrete. The distinct feature of this design-oriented model includes its applicability to normal- and high-strength concretes with cross-sections ranging from circular to rectangular. The model also considers the observed dependency of the hoop rupture stain of the confining jacket on the material properties of the concrete and FRP. In addition, a novel concept, referred to as the confinement stiffness threshold condition, was incorporated into this model to allow for an accurate prediction of post-peak strain softening behavior of FRP-confined concrete. Following this, using the combined database, a unified analysis-oriented model that is capable of predicting the complete stress-strain and dilation behaviors of unconfined, actively confined, and FRP-confined concretes was developed. It was found that, at a given axial strain, lateral strains of actively confined and FRP-confined concretes of the same concrete strength correspond when they are subjected to the same lateral confining pressure. However, under the same condition, concrete confined by FRP exhibits a lower strength enhancement compared to that seen in companion actively confined concrete. On the basis of this observation, a novel approach that incorporates the confining pressure gradient between the two confinement systems was established and a unified stress-strain model was developed. Other distinct features of this highly versatile model are its applicability to concretes ranging from light- to normal-weight and low- to high-strength. In addition, it is also applicable to specimens with various sizes and slenderness. To improve the capability of handling complex databases with a large number of independent variables, a third category of model that uses evolutionary algorithm and soft computing techniques was considered. In this study, a genetic programming approach that is capable of gradually refining the solution while maintaining the versatility of the model in closed-form expressions was used to develop an evolutionary algorithm model for FRP-confined concrete. Lastly, the finite element modeling approach was investigated. A review of the existing literature revealed that the failure criterion and flow rule considered by existing FE models for confined concretes are based on limited test results and they are not very sensitive to the variations in unconfined concrete strength and confining pressure. Based on the comprehensive experimental databases assembled, a concrete strength-sensitive finite element model applicable to concrete subjected to various confining pressure levels was developed. Comparison with experimental test results show that the predictions of all of the proposed models are in good agreement with the test results, and the models provide improved accuracy compared to the existing models. Comparison of models in different categories indicates that model accuracy generally improves with the size of database and the complexity of modeling framework; however, the choice of models depends mainly on the suitability of their end use.