Ellipsoidal seed modeling and simulation parameter selection based on the discrete element method

Abstract

At present, there are still some issues in the discrete element modeling of non-spherical agricultural materials. For example, it is still necessary to enrich and optimize the particle model of non-spherical granular materials and study their simulation parameters. Therefore, in this paper, ellipsoidal seeds, such as those of soybeans, red beans, and kidney beans, were taken as the research object, and their dimensions and shapes were analyzed. The results showed that the dispersion in size could be approximated by a normal distribution. An approach for modeling ellipsoidal seeds as particles based on the multi-sphere (MS) method was proposed. Moreover, the Plackett–Burman (PB) test and the path of the steepest ascent method were both adopted to calibrate the simulation parameters, which were obtained with difficulty through experiments, and simulations of the piling test and the rotating hub test for calibration to ensure the rationality of the parameter selection. In addition, to improve the reliability of the calibration test, the automatic filling function in EDEM software was adopted for particle modeling, where the smoothness was 2. A 122-sphere model for the soybean seed, a 114-sphere model for the red bean seed, and a 242-sphere model for the kidney bean seed were established, and the three models were applied for the calibration of the simulations of the tests. The results showed that the coefficients of rolling friction between seeds had a significant influence on the simulation results for the three varieties of seeds. Finally, the manual filling models and the automatic filling models were further verified by using simulations of bulk density tests. Comparing the simulation and experiment results showed that when the number of filling spheres increased, the simulation results were closer to those obtained experimentally. Better simulation accuracy could be obtained with fewer filling spheres. These results could provide a reasonable discrete element model and simulation parameters for further simulation and optimization of designs of seeding machinery for ellipsoidal seeds.