An inverse analysis method for determining tensile softening relationship of concrete considering local response

Abstract

This study presents an inverse analysis algorithm for deriving the softening relationship of concrete. The algorithm can not only reproduce the influence of local response on the σ-w curve by introducing additional load criteria but also reduce the dependence of optimization results on the initial guesses by combining global best-fitting with the automatic parameter updating technique, improving the prediction accuracy and lowing the possibility of obtaining pseudo-solutions. Then, the fracture tests were carried out on the wedge splitting specimens. The validity and versatility of the algorithm were verified using experimental data. The findings show that the simulation responses agree well with the test ones in all regions of the load-CMOD curve, and the values of fracture energy determined by the load-CMOD and σ-w curves are consistent. The derived σ-w curve is independent of the initial guesses of parameters. The widely used bi-linear σ-w curve tends to overestimate the load in the post-peak of load-CMOD, and the tri-linear model is recommended after considering the accuracy, computation time, and convergence. The tensile strength evaluated by the inverse analysis method is 75% of the splitting tension test and 91% of the fracture theory method. Further, the proposed algorithm exhibits outstanding versatility, applying to various concrete materials and specimen configurations.