(144b) A Class of Generalized Strain-Hardening Discrete Element Method (DEM): Theory, Liggghts Open-Source Implementation, and Applications for Granular Biomass Flow

In this work we present a class of generalized strain-hardening contact force-displacement models for the discrete element method (DEM). The development of these contact models is inspired by the experimentally measured bulk stress-strain behaviors of fractured granular woody biomass feedstocks such as chipped pine. Granular biomass in general exhibits complex and irregular particle shapes and relatively low stiffness compared with hard, regular-shaped bulk solids such as pharmaceutical pellets and agricultural products. These features of granular biomass result in its unique bulk strain-hardening mechanical characteristics. Although the polyhedral DEM for arbitrary particle shapes can approximate the complex-shaped woody biomass particles with fine geometric details, the computational cost required by the polyhedral DEM can be prohibitive for simulating even a few thousands of particles, if each polyhedral particle comprises of a few hundreds of surface triangular elements for high-quality shape approximation. To allow for the DEM to simulate large-scale transport and feeding operations of granular biomass in the biorefineries, we envisioned a coarse-grained (CG) DEM approach by using only spherical particles as the representative elementary volumes (REV) of bulk granular biomass. To compensate for the loss of accuracy in the particle shapes, we developed a class of strain-hardening force-displacement models for particle-particle contact. The formulation of an unconditionally stabilized viscous damping force associated with each of the contact models is also derived. Since these models are intended for general adoption of the DEM application users rather than biomass flow applications alone, the mathematical formulations in these models are kept in the generalized polynomial and/or exponential functions. To allow for the best possible access to these models on all general computing platforms (PC, macOS, Linux, HPC, etc.), we implemented these models based on the LIGGGHTS open-source DEM particle simulation software and released the LIGGGHTS-INL package (https://github.com/idaholab/LIGGGHTS-INL): a capability-extended adaptation of LIGGGHTS (e.g., see https://doi.org/10.1016/j.powtec.2021.03.008). Examples of the model usage are presented in a model calibration study based on the experimental data of an axial loading-unloading compressibility test of bulk pine particles, in which the bulk stress-strain profiles simulated by the calibrated CG-DEM models are in reasonable agreement with the experimental data. Finally, we will demonstrate an application of the calibrated CG-DEM models to the simulations of hopper discharge of bulk pine particles. Through the comparison with the Hertz contact model simulations, we will highlight the slowed discharge resulting from these CG-DEM models. The experimental test data of the hopper discharge will be used to assess the fidelity of the CG-DEM models. Limitations of these models and suggestions of future improvement will also be discussed.