Parameter calibration and experiment of the discrete element contact model of water-containing sandy soil particles

Soil-engaging components have been severely restricted to the parameters of the water-containing sandy soil. However, it is still lacking in the calibration of simulation parameters. This study aims to clarify the influence of the water content on the contact mechanics among sandy soil particles. Taking the riverside sandy soil as the research object, the Hertz-Mindlin with JKR Cohesion (JKR) contact model was used in the EDEM software. The parameters of sandy soil were then calibrated with four water contents (5%, 10%, 15%, and 20%) using discrete element simulation. The targeted variables of calibration were the static friction coefficient, restitution coefficient, dynamic friction coefficient, and surface energy between sandy soil particles. The simulated angle of the repose of sandy soil particles was taken as the response value. The Box-Behnken response surface method was used for the calibration. A regression model of the angle of repose was also obtained after optimization. It was found that the JKR surface energy shared an extremely significant impact on the angle of repose of sandy soil, and the restitution coefficient had a significant impact on the angle of repose of sandy soil with the higher water contents (15% and 20%), while the static friction coefficient and dynamic friction coefficient had no significant impact on the angle of repose of water - containing sandy soil. The optimal parameters were obtained as follows: for the sandy soil with a water content of 5%, the static friction coefficient between particles was 1.011, the restitution coefficient was 0.41, the dynamic friction coefficient was 0.115, and the JKR surface energy was 0.024 J/m²; for the sandy soil with a water content of 10%, the static friction coefficient between particles was 0.918, the restitution coefficient was 0.532, the dynamic friction coefficient was 0.033, and the JKR surface energy was 0.124 J/m²; for the sandy soil with a water content of 15%, the static friction coefficient between particles was 0.894, the restitution coefficient was 0.835, the dynamic friction coefficient was 0.122, and the JKR surface energy was 1.164 J/m²; for the sandy soil with a water content of 20%, the static friction coefficient between particles was 0.963, the restitution coefficient was 0.893, the dynamic friction coefficient was 0.158, and the JKR surface energy was 3.624 J/m². The relative errors of the angle of repose between the simulation and the physical tests were all less than 5%. The reliability of the model was validated using the intrusion resistance data from the physical pressure-bearing test and the simulation test. There was essential consistency in the resistance trends from the test and the simulation. Once the soil-engaging depth was within the range of 40 mm, the relative errors were all less than 10%. The calibrated parameters were approximately substituted for the real sandy soil. It was also expected to conduct the discrete element simulations between water-containing sandy soil and soil-engaging components. Meanwhile, the findings can also offer the parameter selection under various research, such as the virtual simulation of water-containing soil, the interaction between agricultural equipment and soil, and the performance of soil-engaging components.