小麦籽粒振动筛分黏弹塑性接触模型构建及其参数标定

    Construction of visco-elasto-plasticity contact model of vibratory screening and its parameters calibration for wheat

    • 摘要: 为更好地模拟小麦籽粒(麦粒)振动筛分过程,该文在现有离散元接触模型基础上,通过EDEM软件应用程序编程接口,构建了一种麦粒黏弹塑性接触模型。该接触模型法向方向通过将Kuwabara and Kono非线性黏弹性接触模型中的黏性耗散项引入Thornton滞回接触模型进行构建;切向方向采用简化Thornton切向接触模型;滚动摩擦力矩计算同Hertz-Mindlin(no slip)接触模型。法向模型参数标定采用单轴加载-卸载试验、碰撞试验,分别构建了模拟-试验接触力误差平方和与麦粒屈服重叠量、恢复系数与法向阻尼系数的二阶回归方程,得到麦粒屈服重叠量为7.63×10-6 m,麦粒-麦粒/钢板法向阻尼系数分别为190.68和306.65。切向模型参数利用旋转鼓试验进行标定,得到最佳麦粒-麦粒/钢板静摩擦系数组合为0.40和0.44。最后利用振动筛分试验对所标定参数进行验证,模拟与试验所得筛下麦粒质量分数最大误差为8.97%,模拟中筛下物分布规律与试验结果无显著性差异,表明所建立的接触模型及标定的参数能够很好地模拟麦粒振动筛分过程。该文也可为其他农业物料黏弹塑性接触模型构建及其参数标定提供参考。

       

      Abstract: Abstract: In order to better simulate the screening process of wheat, a visco-elasto-plasticity contact model was constructed based on the existing discrete element method (DEM) contact model. The modelling process was based on the application programming interface (API) of the EDEM 2.7 software, which allowed users to create customized contact model, particle factory and coupling interface by the C++ programming language. The normal contact model was created by adding the viscous dissipation term of the Kuwabara and Kono non-linear viscoelasticity model to the Thornton hysteretic model. The simplified Thornton tangential model was used to calculate the tangential force and the calculation of rolling friction was the same as that used in the Hertz-Mindlin (no slip) model. The calibration of the parameters of the visco-elasto-plasticity contact model was based on a uniaxial loading-unloading test, collision test and rotating drum test. Specifically, the uniaxial loading-unloading test was conducted to determine the yield overlap of wheat as the loading rate was slow (1 mm/min) and it could be regarded as a quasi-static process in which the viscous dissipation term can be neglected. In this work, the sum of squared errors of the contact force at all sampling points was calculated and the regression equation was developed, representing the relationship between the sum of squared errors of the contact force and the yield overlap of wheat. The result showed that the difference between the experimental and simulated values was the least when the yield overlap was 7.63×10-6 m. Then the wheat-wheat/steel collision tests were respectively conducted to calibrate the wheat-wheat/steel coefficients of viscidity as the coefficient of restitution measured in these collision tests can reflect the energy dispassion of collision. Specifically, the double-pendulum test was used to calibrate wheat-wheat coefficient of viscosity and the inclined plate test was used to calibrate wheat-steel coefficient of viscosity. The regression equations describing the relationship between the wheat-wheat/steel coefficients of restitution and coefficients of viscosity were also developed. Solving the equations with the measured wheat-wheat/steel coefficients of restitution, the wheat-wheat/steel coefficients of viscosity were calibrated to be 190.68 and 306.65. In DEM parameters calibration, the dynamic angle of repose measured in rotating drum tests was often used as it could well reflect the flow regime of particles. The wheat particles in screening process were dynamic, and thus the rotating drum test was also conducted to calibrate the coefficients of static friction and rolling friction. The results of extensive pre-experiments showed the wheat-wheat/steel rolling friction had little influence on the dynamic angle of repose. So only the wheat-wheat/steel coefficient of static friction was calculated. The calibration results showed that the difference between the simulated and measured angles of repose was the least when the wheat-wheat/wall coefficients of static friction were 0.40 and 0.44, respectively. In order to verify whether the newly constructed visco-elasto-plasticity contact model and the calibrated parameters could well simulate the vibration screening process of wheat, a vibratory screening test was conducted. The screened wheat particles were collected and weighted by a material box with 7 sections, from which the undersize distribution of wheat in simulation was found to have similar trends with the realistic test and the wheat particles were mainly distributed in the first 3 sections. In order to evaluate whether the difference between the simulated and measured mass friction was significant, a paired T-test was performed by the SPSS software, and no significant difference was found between the simulated and experimental results. Moreover, the simulated and experimental results had a high correlation coefficient (0.98). The validation result showed that the model and the parameters can well simulate the screening process. This research can not only simulate the visco-elasto-plasticity characteristics of wheat, but also provide guidance for developing visco-elasto-plasticity models and their parameters calibration of other crops.

       

    /

    返回文章
    返回