Probabilistic calibration of stress-strain models for confined high-strength concrete

Abstract

A comprehensive probabilistic calibration of traditional deterministic models for peak stress, peak strain, and stress-strain curves of confined high-strength concrete (HSC) was investigated. The probabilistic models for peak stress and peak strain of confined HSC were first established by combining the Markov chain Monte Carlo (MCMC) method with the Bayesian theory. A probabilistic stress-strain model of confined HSC was then proposed to provide a probabilistic approach to calibrate the confidence level and computational accuracy of four typical deterministic stress-strain models of confined HSC. Analysis results show that the randomness of the stress-strain curve in the ascending branch is not obvious, but that in the descending branch after peak stress is significant. Deterministic stress-strain models can better predict tested stress-strain curves in ascending branches with a greater confidence level than descending branches. The tested stress-strain curves generally fall within the 50% confidence interval of the probabilistic stress-strain model, which implies that the proposed probabilistic stress-strain models can adequately describe the probabilistic characteristic of stress-strain curves of confined HSC.