CS and RCPT Prediction of Concrete Samples Using Bayesian Inference and Performing Different Reliability Analyzes

Abstract

Compression strength (CS) and rapid chloride permeability test (RCPT) are very significant parameters in mechanical and durability properties in concrete, respectively. Analytical methods such as formulas and graphs for prediction and reliability of CS and RCPT in concrete samples are gathered with many problems. Many soft computing methods are very accurate in the prediction of CS and RCPT but these methods are deterministic or have not reliability tools. For these reasons, Bayesian inference is used which is a probabilistic and linear method. For this purpose, according to some of the concrete samples, a probabilistic relation is proposed for each CS and RCPT. The accuracy of each proposed formula is tested, and after verification of them, reliability analysis is performed. In this study, the first-order reliability method (FORM), Monte-Carlo sampling (MCS), and histogram sampling are used for reliability analysis. Each of these methods has unique properties that FORM is linear and has a very short time-consuming. MCS and histogram sampling are nonlinear and have high time-consuming but their accuracy are very high. Histogram sampling is similar to MCS but in this type of analysis, reliability results for any outcomes are given, and time-consuming in this method is very high. A three-method analysis of CS and RCPT showed that the results are closed together. So, using FORM because of use easily and save time-consuming can be a reasonable choice for reliability analysis of CS and RCPT in concrete samples.