Search

The structured sand presents significant interparticle bonding and anisotropy, resulting in significant differences in the physical and mechanical properties from the pure sand. This study proposes a new surrogate model based on the concept of multi‐fidelity residual neural network (MR‐NN) as an alternative to DEM simulation for predicting the mechanical behaviours of structured sand with different initial anisotropy and saving largely computational costs. The model is initially trained using low‐fidelity (LF) data to focus on capturing the main underpinning correlations between macroscopic mechanical parameters with inter‐particle properties and anisotropic state variables (i.e., microscopic fabric and tilting angle of sand), where the LF data are generated from previously proposed anisotropic macro‐micro quantitative correlation. Subsequent training on sparser high‐fidelity (HF) data is used to calibrate and refine the model with HF data generated from DEM simulations for anisotropic structured sand. Feedforward neural network (FNN) is adopted as the baseline algorithm for training models. The macroscopic anisotropic parameters of structured sand predicted by the surrogate model are compared with DEM simulations to examine its feasibility and generalization ability. Furthermore, the robustness of the surrogate model is examined by discussing the effect of LF data on the performance of MR‐NN. The superiority of the MR‐NN is further verified by comparing the performance of the trained MR‐NN with the one‐shot trained FNN based on the same HF data. All results demonstrate that the proposed surrogate model can provide a fast and accurate simulation of the anisotropic parameters of structured sand. [ABSTRACT FROM AUTHOR]

Using the microstructural mechanics approach, the macroscopic anisotropic failure criteria expressed in closed form by the microscopic parameters are derived for regularly arranged elliptical particle assemblies. The close-packed, elliptical particle assembly is equivalent to a lattice model that is described by a lattice beam system, and the Mohr–Coulomb-like shear failure criterion is applied to describe the beam breakages. Through mechanical analysis of the instability state of the unit cell of the lattice beam system, macroscopic anisotropic strength criteria and the corresponding strength parameters, which are expressed by microscopic parameters and the tilting angle, are proposed. The macroscopic failure characteristics and strength parameters are qualitatively consistent with the results of the discrete element method (DEM), and the influences of the microscopic parameters and tilting angle on the macroscopic strength parameters are investigated in detail according to the proposed theoretical results. The applicability of the theoretical results to randomly distributed particles is explored by DEM simulation of irregularly arranged particles, and the deviation are also systematically discussed. [ABSTRACT FROM AUTHOR]

Using a microstructural mechanics approach, the close-packed elliptical particle assembly is first represented by a lattice model that is described by a beam system, and the Mohr–Coulomb (MC) shear failure criterion is used to describe the contact (beam) breakage. Through structural mechanics analysis of the unit cell, macroscopic strength criteria and associated parameters at two anisotropic principal directions are expressed in closed form with respect to the particle shape and microscopic strength parameters while considering the successive breakage of contacts. The macroscopic failure characteristics of elliptical particle assemblies are described by the MC failure criterion, and the macro–micro relations of strength parameters can be directly determined by the proposed expressions. The proposed analytical expressions are confirmed to be accurate due to the good consistency between theoretical and discrete element method (DEM) results based on the regular arrangement. The introduction of the fabric anisotropy parameter extends the applicability of the theoretical results to irregular arrangement cases, and the deviation with more realistic particle assemblies is investigated. The proposed expressions of the macroscopic strength parameters can help us better understand the influence of inherent anisotropy and microscopic properties and provide an alternative approach to determine the microscopic parameters in DEM simulations. [ABSTRACT FROM AUTHOR]

This study is devoted to analytically establish macro–micro quantitative relationships for creep characteristics of a rheological particle assembly, contributing to a more comprehensive insight into the interplay between different scales. To represent the various rheological behaviour in microscopic for a hexagonal close-packed (HCP) circular particle assembly, the contact responses are assumed as viscoelastic models. By extending the traditional homogenization method to the Laplace domain, the macroscopic one/two-dimensional creep equations of the rheological particle assembly, expressed with the microscopic contact parameters in a closed-form manner, are innovatively identified. It is found that the macroscopic creep behaviour is not always consistent with the microscopic one and in general exhibits more complex creep characteristics. The parametric analysis presents the quantitative influence of contact parameters on macroscopic creep behaviour. Extending the one-dimensional case to three dimensions (3D), the relaxation moduli as well as creep compliance, expressed in closed-form by contact parameters, are obtained in 3D viscoelastic constitutive equations. Distinct viscoelastic behaviours are observed for macroscopic shear and bulk responses. The proposed macro–micro quantitative relations of creep characteristics in this study benefit to deeply understand the macro–micro relations of viscoelastic properties, providing an alternative approach to calibrate microscopic parameters when involving the DEM simulations. [ABSTRACT FROM AUTHOR]

The commonly used trial-and-error approach on selecting appropriate inter-particle parameters in DEM simulations incurs criticism such as user dependence and high computational cost. This study proposes a new framework based on a multi-fidelity residual neural network (MR-NN) as an alternative for calibrating inter-particle parameters in rock-like bonded granular materials. The model is first trained using low-fidelity data (LF) to focus on capturing the main underpinning correlations between macroscopic elastic and strength parameters with inter-particle parameters of contact models, where the LF data is generated from micro-macro quantitative relations. Subsequent training on sparser high-fidelity (HF) data is then used to calibrate and refine the model, in which the HF data is generated from DEM simulations for rock. Feedforward neural network (FNN) is considered as the baseline algorithm for training models. The trained MR-NN with the same LF data is finally used to predict the inter-particle parameters of calcarenite and granite in DEM simulations to examine its feasibility and generalization ability. Furthermore, the robustness of the framework is verified by discussing the effect of LF data on the performance of MR-NN. All results demonstrate that the proposed method can provide a fast and accurate determination of inter-particle parameters for DEM simulation. [ABSTRACT FROM AUTHOR]