The microstructure of a soil-rock matrix (SRM) is a determinant of its macroscopic physical and mechanical attributes. The influence of the shape and volume fraction (VF) of rock blocks on the dynamic properties of the SRM has not been subjected to quantitative analysis. Consequently, a suitable construction technique was developed for the fabrication of small-scale triaxial specimens incorporating artificial rocks of various shapes. A series of homogeneous SRM specimens with differing rock VFs and shapes were fabricated. These specimens were then exposed to a long-term dynamic load consisting of 15,000 cycles at a frequency of 1 Hz. The principal findings are summarized as follows: The construction method proposed is capable of producing small artificial rocks with dimensions of 3 mm or 5 mm in arbitrary shapes while maintaining consistency with the prototypes. The method holds significant promise for application in geotechnical testing. Under long-term dynamic loading, the rock VF effectively elevates the threshold cyclic stress ratio of the SRM, diminishes the Pore Water Pressure within the mixture, enhances the dynamic stiffness, and mitigates the cumulative strain. SRMs composed of rock shapes with increased angularity and reduced block sizes exhibit higher dynamic stiffness and cumulative strain. The threshold cyclic stress ratio for an SRM with a 40 % rock VF is approximately 0.04, and the pore pressure increment in the SRM exhibits a gradual change, which contrasts with the test outcomes for pure clay. The exponential-hyperbolic model provided a satisfactory fit for the pore pressure data, while the hyperbolic model yielded good fitting results for the cumulative strain of the SRM with a low rock VF. These findings contribute to an enhanced comprehension of the dynamic properties of railway subgrades filled with SRM under cyclic train loading conditions.

This study investigated the influence of sample preparation methods, moist tamping and wet pluviation, on the erodibility and mechanical behaviour of gap-graded soils with three gradations: fully stable, unstable, and on the borderline of stability. Drained triaxial tests were performed using a modified erosion-triaxial apparatus, followed by micro-CT scanning to assess pore network properties. The results indicated that for fully stable and fully unstable samples, the preparation method had minimal impact on both erosion and mechanical behaviour. However, for the samples on the borderline of stability, wet pluviation method resulted in fine particle segregation, creating a heterogeneous structure with reduced pore connectivity. This led to lower erosion rates (0.4 gr/min reduction compared to the moist tamping technique), but mechanical properties remained largely unaffected, as confirmed by similar intergranular void ratios and stress-strain responses. Micro-CT scanning quantified differences in pore structure, showing that wet pluviation samples exhibit lower connected porosity compared to those prepared by moist tamping. These findings highlight the critical role of specimen preparation in assessing suffusion susceptibility and erosion behaviour, particularly for soils near the threshold of instability.

Specimen preparation methods significantly affect indoor sand soil's properties. A theoretical model was developed based on granular thermodynamics to describe the undrained shear characteristics of calcareous sand. We validated the model using specimens prepared through four different methods under various relative densities, stress paths, and confining pressures. The hardening index and granular temperature of this model effectively reflect the impact of specimen preparation methods on the strength characteristics of calcareous sand. The model successfully captured these differences in shear behavior resulting from four preparation methods by adjusting the hardening parameters. Furthermore, this model can accurately describe the influence of stress path changes on the mechanical responses of the calcareous sand granular system by incorporating the third elastic strain invariant.