FRP-Confined Concrete Stress-Strain Calibration with Bayesian Uncertainty Quantification

Abstract

The application of fiber-reinforced polymer composites for the confinement of concrete structures has become a widespread technique in structural engineering rehabilitation and construction. Despite the proliferation of theoretical and empirical stress-strain models, the inherent material variabilities and measurement uncertainties present substantial challenges for reliable performance prediction. This paper presents a comprehensive framework for the calibration of fiber-reinforced polymer-confined concrete stress-strain models utilizing Bayesian uncertainty quantification. By transitioning from traditional deterministic approaches to a probabilistic paradigm, this research systematically integrates prior knowledge with experimental database observations to yield posterior predictive distributions. The methodology employs Markov chain Monte Carlo sampling techniques to evaluate the statistical distributions of key model parameters, notably the dilation properties and ultimate strain capacities. The analysis demonstrates that Bayesian calibration not only refines the accuracy of the ultimate stress and strain predictions but also provides robust confidence intervals that quantify both aleatory and epistemic uncertainties. The resulting probabilistic models offer a rigorous foundation for reliability-based design codes, ensuring structural safety while optimizing material utilization. The findings advocate for the paradigm shift towards stochastic modeling in structural materials engineering, laying the groundwork for more resilient infrastructure systems that can safely accommodate natural material variations and construction tolerances.