RESUMEN
The purpose of this work is to simulate the powder compaction of refractory materials, using the discrete element method (DEM). The capability of two cohesive contact models, implemented in different DEM packages, to simulate the compaction of a mixture of two refractory materials (dead burnt magnesia (MgO) and calcined alumina (Al2O3)) was analyzed, and the simulation results were compared with experimental data. The maximum force applied by the punch and the porosity and final shape quality of the compact were examined. As a starting point, the influence of Young's modulus (E), the cohesion energy density (CED), and the diameter of the Al2O3 particles (D) on the results was analyzed. This analysis allowed to distinguish that E and CED were the most influential factors. Therefore, a more extensive examination of these two factors was performed afterward, using a fixed value of D. The analysis of the combined effect of these factors made it possible to calibrate the DEM models, and consequently, after this calibration, the compacts had an adequate final shape quality and the maximum force applied in the simulations matched with the experimental one. However, the porosity of the simulated compacts was higher than that of the real ones. To reduce the porosity of the compacts, lower values of D were also modeled. Consequently, the relative deviation of the porosity was reduced from 40-50% to 20%, using a value of D equal to 0.15 mm.
RESUMEN
The purpose of this work was to analyse the compaction of a cohesive material using different Discrete Element Method (DEM) simulators to determine the equivalent contact models and to identify how some simulation parameters affect the compaction results (maximum force and compact appearance) and computational costs. For this purpose, three cohesion contact models were tested: linear cohesion in EDEM, and simplified Johnson-Kendall-Roberts (SJKR) and modified SJKR (SJKR2) in LIGGGHTS. The influence of the particle size distribution (PSD) on the results was also investigated. Further assessments were performed on the effect of (1) selecting different timesteps, (2) using distinct conversion tolerances to export the three-dimensional models to standard triangle language (STL) files, and (3) moving the punch with different speeds. Consequently, we determined that a timestep equal to a 10% Rayleigh timestep, a conversion tolerance of 0.01 mm, and a punch speed of 0.1 m/s is adequate for simulating the compaction process using the materials and the contact models in this work. The results showed that the maximum force was influenced by the PSD due to the rearrangement of the particles. The PSD was also related to the computational cost because of the number of simulated particles and their sizes. Finally, an equivalence was found between the linear cohesion and SJKR2 contact models.