In order to predict the resilient modulus of compacted clays, the material indicator tests, repeated load triaxial test, and pressure plate test for compacted clays from South China were carried out. The soil–water characteristic curve (SWCC) was described using the Fredlund and Xing’s model. And a logarithmic function between the soil suction and resilient modulus was built. Then, a new variable named the minimum bulk stress was defined to separate the shear effect of soil samples from the bulk stress, which avoids the bulk stress reflects two contrary effects, namely hardening effect and softening effect. Subsequently, the influences of the degree of compaction, soil suction, minimum bulk stress, and octahedral shear stress on the resilient modulus were analysed. In the following, a new resilient modulus prediction model of compacted clays, which took the soil suction, minimum bulk stress, and octahedral shear stress as the model variables, was developed and verified using the data of different cohesive soils from this study and existing literatures. The results show that the new model matched these data well and have a high prediction accuracy, which indicates that this new model is reasonable and widely applicable.

The resilient modulus has been used to characterize the stress–strain nonlinear behavior of subgrade soils and is the primary property needed for pavement design and analysis. The degrees of stress and moisture content have a significant impact on the resilient modulus of compacted cohesive soils. Due to the hot, humid climate in South China, the moisture content in embankments will increase gradually from the optimum moisture content (OMC)—the design water content for embankments—to the equilibrium moisture content, which relates to soil properties and the surrounding climate. In this process, the resilient modulus of embankments will obviously decrease. In order to predict the resilient modulus of typical compacted cohesive soils in South China, repeated triaxial tests were carried out in this study. The soil matric suction was measured by the pressure plate test and the soil–water characteristic curve (SWCC) was described using the Van Genuchten model with a relatively high coefficient of determination. Then, the effect of the moisture content, degree of compaction, and stress state including the deviator stress, confining pressure, and octahedral shear stress to the resilient modulus were analyzed. In addition, a logarithmic function was utilized to build the relationship between the matric suction and resilient modulus. Subsequently, a new resilient modulus estimation model of compacted soils in South China, which took the bulk stress, octahedral shear, and matric suction as the model variables, was developed and verified using the data of different cohesive soils from different studies. The results show that the new model matches their data well, and the coefficients of determination are high, which indicates that this new model is reasonable and widely applicable. Finally, the correlations between the physical parameters of soil samples, such as the liquid limit, dry density, the maximum dry density, plasticity index, percentage passing through a 0.075-mm sieve, and the regression coefficients of the new model, were established. The resilient modulus can be predicted much more easily with these physical parameters of compacted cohesive soils rather than conducting triaxial tests.

The physical composition and stress state of soil-rock mixture (SRM) materials have a crucial influence on their mechanical properties, and play a vital role in improving the performance of subgrade. To reveal the resilient behavior and mesostructure evolution of SRM materials, triaxial tests and discrete element method (DEM) numerical analysis have been carried out. In the triaxial test section, the mechanical response of SRM materials was investigated by preparing samples under different stress states and physical states and conducting triaxial tests on samples. Simultaneously, a new irregular particle modeling method was developed and applied to the discrete element modeling process to analyze the mesostructure evolution of SRM materials under cycling loading. First, a cyclic triaxial test of SRM material is performed on the SRM material, and the effects of bulk stress, octahedral shear stress and rock content on the resilient modulus of the SRM material are analyzed. It is revealed that the resilient modulus increases with increasing bulk stress and rock content, and decreases with increasing octahedral shear stress. Based on a new resilient modulus prediction model, the relationships among the rock content, stress state and resilient modulus are established. Then, based on an improved DEM modeling method, a discrete element model of the SRM is established, and the influence of rock content on coordination number and mesostructure evolution of the SRM is analyzed. The results show that in SRM materials, the increase of crushed rock changes the mesostructure of the SRM material. With the increase of rock content, the internal contact force changes from “between soil and rock” to “between rocks”, and the skeleton formed in the rocks gradually develops overall stiffness. Under the condition of low stress, the anisotropy of the SRM material is mainly caused by the shape and grade distribution of crushed rock. The induced anisotropy caused by the change of stress state has little effect on its mechanical behavior, which may lead to the greater dispersion of multiple SRM test results.

Subgrade soils are often unsaturated and the resilient modulus (M_R) of subgrade soils is usually subjected to the climate environment and traffic loading in the field. Therefore, the Matric Suction (MS) and traffic loading are considered to be two important parameters associated to the M_R prediction model. To verify the M_R prediction model, the MS of the typical subgrade soil were determined through the pressure plate test. In this study, the soil-water characteristic curves were also described using the Fredlund & Xing’s model. Then, the dynamic M_R of the typical subgrade soil under various stresses and water contents was measured. After that, a new prediction model was proposed with the model variables including the minimum bulk stress, octahedral sheer stress and matric suction, and the validity of the new model was verified by previous research results. Finally, the correlations between the physical properties of subgrade soils including the percentage passing through the No. 200 sieve (0.075 mm), plasticity index, liquid limit, dry density and the regression coefficients of the new model were established. The results show that the new model can be used to predict the M_R well, and it effectively solves the problem that the bulk stress is equal with a different combinations of the confining pressure and deviator stress. At the same time, the M_R can be predicted much more easily with physical parameters of subgrade soils rather than conducting triaxial tests.