Developing an adequate DEM model to simulate soil-tool interactions under sticky soil conditions

Abstract

The numerical modelling of interactions between soil-engaging tools is crucial for designing more effective and affordable soil preparation equipment. In this study, the linear cohesion model was coupled with the hysteretic spring model (HSCM) to effectively represent soil cohesion behavior and plastic deformation in sticky soil found in paddy fields. The sensitivity of DEM-simulated soil bulk density to specific parameters of the proposed model was analyzed using a two-level factorial test to identify the key influencing parameters. The discrete element model was calibrated against experimentally determined wet bulk density (wet weight basis) to accurately represent the soil mass being displaced by soil-engaging tools, which is critical for draught prediction under high-moisture soil conditions. These key parameters were calibrated in two consecutive phases: (1) the steepest ascent design and (2) the Box-Behnken design to determine the optimum values of these parameters that minimize the relative error between the DEM-simulated bulk density and experimentally measured soil wet bulk density. The calibrated model was then validated through soil furrowing experiments, using the created furrow profile dimensions and the draught at different depths as evaluation metrics. The validation results demonstrate that the calibrated model achieves satisfactory predictive accuracy for draught across multiple furrowing depths, with relative errors between 7.5 % and 8.8 %. Furthermore, comparative analysis of furrow profile dimensions revealed that the model accurately simulates both width and depth, exhibiting relative errors of 14.7 % and 1.3 %, respectively. Thus, the proposed model can facilitate and accelerate the design and optimization of components that engage with sticky soil. The findings also offer a framework for selecting optimal DEM parameters for soft, sticky soils, minimizing the computational cost of model calibration in future applications.