Probabilistic calibration of strength and strain enhancement models for FRP-confined concrete in circular section: Fid-Bau Portal

Probabilistic calibration of strength and strain enhancement models for FRP-confined concrete in circular section

Chen, Qi-Sen / Yu, Bo / Li, Bing

Highlights • Probabilistic models for simulating the strength and strain enhancement of FRP-confined concrete for circular section were proposed based on Bayesian updating and MCMC. • A total of nine models for SEE and six for SAE were chosen for review in this paper regarding the Credible Interval and Confidence Level. • Different types of FRP, including AFRP, CFRP, GFRP, HM, UHM wrapped and Tube-encased were investigated. • Sensitivity of different cases of $ε_{co}$ , $ε_{h, r u p}$ , $f_{co}$ , and $K_{l}$ were discussed with scatter plots.

Abstract Lateral confinement with fiber-reinforced polymer (FRP) jackets can effectively improve the axial compressive behavior of concrete but with considerable variability in outcomes. Therefore, that there is a need for calibrations of deterministic models for strength enhancement (SEE) and strain enhancement (SAE) based on proposed probabilistic prediction models and comprehensive available databases. The probabilistic models that incorporate the essential factors identified from the previous study were updated based on the Bayesian theory, and Markov Chain Monte Carlo (MCMC). Moreover, nine representative deterministic SEE models and six SAE models were evaluated by credible interval (CI) and confidence level (CL) under different conditions of the axial strain of unconfined concrete in the literature, the hoop rupture strain, the peak axial compressive stress of unconfined concrete, and the lateral confinement stiffness. Analysis of different FRP types was also conducted individually for more critical results. The proposed probabilistic models are capable of predicting the characteristics of ultimate axial stress and corresponding strain and providing an efficient approach to calibrate the confidence level and computational accuracy of deterministic models in literatures.