A homogenization‐based state‐dependent model for gap‐graded granular materials with fine‐dominated structure

Abstract Gap‐graded granular soils are common natural soils composed of cohesionless granular matrix and rock aggregates. Since they are widely used as construction materials worldwide, their deformation behavior is crucial for the design of civil infrastructures. There have been rare constitutive models reported for gap‐graded granular soils in the literature. This study presents a homogenization‐based state‐dependent model within the elastoplastic framework. The model features a homogenization equation with a structure parameter and a novel state‐dependent dilatancy function to describe the behavior of granular matrix. Simulations of the model reveals that the initial stiffness and peak shear strength of gap‐graded soils rely on both the coarse fraction and the initial density of granular matrix. The structure parameter is related to the internal structure of gap‐graded soils, and it varies with the particle shape, and particle size distribution of rock aggregates. A higher value of the structure parameter indicates a more distinct coarse fraction effect of rock aggregates, that is, higher initial stiffness, higher peak shear strength, and higher residual shear strength. For a given coarse fraction, the critical states can be well fitted by a straight line through the origin, and the critical state strength parameter M m increases with the rising rock fraction. A practical method is further proposed for the critical state line of gap‐graded granular soils in e ‐ p ′ compression plane. The proposed model is validated by using experimental data from the literature, including the soils with various densities of sand matrix, different particle size, and particle size distribution of rock aggregates. Comparison between measured data and model predictions indicates that the proposed model can well reproduce the stress‐strain relationship and volumetric deformation behavior of sand‐gravel mixtures and soil‐rock mixtures.