An assessment of retrogressive behavior in soil-rock mixture slopes from MPM modelling to failure

Soil-rock mixture (SRM) slopes frequently undergo highly destructive retrogressive collapses. However, the retrogressive mechanisms remain poorly understood due to inherent structural heterogeneity and large-deformation behavior. This study systematically investigates the retrogressive behavior of SRM slopes using the material point method (MPM). Firstly, the rock block contours are acquired by close-range photogrammetry, and the rock block morphologies are quantified with the computational geometry algorithm, so a digital rock block database is established. Secondly, MPM models of SRM slopes are modelled using the digital image processing techniques. Finally, parametric studies are conducted to evaluate the effects of rock block content (RC), gradation, and soil strain-softening property on the retrogressive behavior of SRM slopes. In addition, the influences of grid size and particle density are also assessed through an experimental–numerical comparative analysis. A coupling effect between RC and gradation in governing retrogressive modes is revealed by the results, and the influences on the run-out distance are exhibited by the softening modulus and residual strength. Moreover, reference values for both the model height-to-grid size ratio and particle density are established in accordance with experimental validation. The findings can provide a theoretical foundation for the safety assessment and reinforcement design of SRM slopes.