Probabilistic-based assessment for tensile strain-hardening potential of fiber-reinforced cementitious composites

Abstract

This paper presents a novel probabilistic-based approach considering material heterogeneity to assess the tensile strain-hardening potential of fiber-reinforced cementitious composites (FRCC). Multivariate adaptive regression splines (MARS) method is used to explicitly express the performance indices governing tensile strain-hardening. First order reliability method (FROM) is then carried out to evaluate tensile strain-hardening potential of FRCC. Results show that strain capacity of FRCC has a negative correlation with failure probability and it increases exponentially with decreasing failure probability. Analysis of variance (ANOVA) decomposition of MARS model indicates increasing fiber strength and volume, reducing fiber modulus, and moderate interface frictional bond are effective means to improve tensile strain-hardening potential of FRCC. The proposed approach is thus able to consider uncertainty in evaluating tensile strain-hardening potential of FRCC by treating micromechanical parameters as random variables and taking heterogeneity into account in the probabilistic-based model.