Abstract Accurately simulating the post-peak behavior of brittle rocks remains a challenge for statistical damage constitutive models. This study investigates the sensitivity of post-peak simulation to model components by evaluating combinations of three strength criteria and three micro-element strength distributions against triaxial test data from five rock types. Results indicate that the probability distribution of micro-element strength governs simulation accuracy, while the strength criterion exerts minimal influence. Based on these findings, a relative brittleness index-based method for selecting the optimal distribution is proposed, along with a dual-parameter collaborative correction method. Validation shows that this approach significantly enhances the fit to experimental post-peak curves, eliminates criterion-induced interference in parameter adjustment, and remains effective across multiple distribution modes. The proposed method offers a solution for improving the reliability of damage models in predicting brittle rocks failure.
Shen et al. (Sun,) studied this question.