The variability in particle morphology significantly impacts the mechanical properties of rockfill materials. To enhance the understanding of this influence, this study collected basalt rockfill particles from 6 different site sources, with their morphology captured by 3D scanning technology, and then the morphological characteristics categorized through cluster analysis. True triaxial tests for these 6 particle groups were simulated using discrete element method (DEM), and the effects of elongation, flatness, convexity, and intermediate principal stress coefficient on the stress-strain relationship and peak strength were qualitatively assessed through principal component analysis (PCA). Further, by controlling the elongation, flatness, and convexity, 3D reconstructed particle models were created by spherical harmonics (SH) analysis, and the true triaxial tests on these models were simulated to quantitatively clarify the influence of morphological parameters on the macroscopic stress- strain relationship, peak strength, microscopic contact, anisotropic evolution, and other characteristics. Considering the size effect in rockfill materials, multi-scale models incorporating particle morphology were further evaluated across four sample scales. The results indicate that, on the macro scale, the three morphological parameters and the middle principal stress coefficient each have substantial effects on peak strength independently, while the interaction among these parameters does not have a notable influence on the strength. With increasing convexity, the peak strength of samples gradually decreases, while an increase in elongation and flatness leads to a trend of initially increasing and then decreasing strength. On the micro scale, the increase in both elongation and flatness results in a more uniform fabric in the main and lateral directions, while the coordination number shows a trend of initially increasing and then decreasing before stabilizing gradually. The influence of elongation on the main direction fabric is slightly smaller than that of flatness, while convexity has minimal effect on these microscopic features. Additionally, the morphological parameters not only impact the deformation capacity of samples but also demonstrate heightened sensitivity to the strength-size relationship of the sample due to interlocking and boundary constraints between particles. This underscores the pivotal role of morphological parameters in governing the mechanical motion of particles during the sample size scaling process, consequently influencing the strength of the material.

A great challenge in the construction of braced excavation in unsaturated residual soil is the emergency caused by rainfall-infiltration, which may affect the safety of the braced excavation and the serviceability of adjacent underground structures. Therefore, prediction of the deformation of braced excavation induced by rainfall-infiltration is becoming one of the major tasks in the design of underground engineering. Numerical simulation is gaining popularity, and many constitutive models are available nowadays to analyze excavation problems. However, it is not clear to select a suitable soil constitutive model to describe responses of the braced excavation under rainfall-infiltration in unsaturated soil. This paper investigated the optimal selection of different soil constitutive models. These models included the hardening soil model (the HS model), the Mohr-Coulomb model (the MC model), and the Tresca model. A series of laboratory tests back-analysis using the above three soil constitutive models were conducted to evaluate the ability to predict the mechanical response of soils. And a case of braced excavation in unsaturated soil was simulated by using the different constitutive models. The optional selection of constitutive models was discussed by comparing simulation results of wall deflection, earth pressure, and stress path of soils. These investigations confirmed that the HS model had the best predictions of soil stress state and excavation deformation. This result may be since the HS model could more reasonably predict the pore water pressure and critical failure state of soils under rainfall and excavation conditions.