| description abstract | Tropical residual soils are found in different parts of the world and consist of mixtures of different types of soils such as sand, silt, and clay, resulting in intricate microstructures and mechanical responses. In this context, and guided by the varying composition of these soils, a 3D discrete element method (3D-DEM) model was developed in which two different contact models are assigned among idealized spherical particles to represent the coarse and fine parts of the tropical soil with two distinct sets of numerical parameters. A simple linear rolling resistance contact model was used to represent the coarse cohesionless component, while a softer adhesive rolling resistance contact model with a linear approximation of the van der Waals attraction force was used for the fine cohesive component. The numerical coarse network is continuous in terms of interparticle contacts and represents the main skeleton of the DEM sample, whereas so-called fine contacts form a local force network between the coarse particles. After a parametric study on the effects of adopting such a numerical mixture, the model was calibrated for a drained compression triaxial test with a specific void ratio. To estimate the equivalent DEM model void ratio, a proportionality between the real soil void ratio and the DEM model void ratio was efficiently employed. During the validation phase, successful model predictions were achieved on drained and undrained triaxial tests and cyclic tests with different strain amplitudes and moderate (hundreds of kPa) confining pressures. Tropical residual soils were proposed to be simulated through a grain-based numerical model using the discrete element method (DEM), inspired from the microstructure and the physical components of those soils. The proposed model may contribute in various ways to reliable numerical modeling of existing or new earthfill structures under monotonic and cyclic loadings in tropical areas. First, with an understanding of its limitations (e.g., regarding grain breakage), the model can complement lab mechanical tests, which are often scarce, to consider additional loading conditions. This may lead to better definition of analytical constitutive relations for tropical soils, because the model outputs a wide range of macro- and microscale information (e.g., elastic properties, the influence of the fine content, etc.) on the mechanical behavior of mixed soils. Finally, with significant computational resources, the model could be directly employed for 3D multiscale discrete-continuum modeling of a structure as a boundary value problem, whereby analytical constitutive models are bypassed and the constitutive response of the material is instead derived through direct stress–strain computations in the proposed model. | |
