Analysis of Retention and Loss Mechanisms of Walnut Shells and Kernels in Vibratory Screening Based on EDEM

Shell–kernel separation is a critical stage in walnut processing; however, the presence of kernel loss and shell retention during separation markedly limits overall performance. In this study, discrete element method (DEM) simulations were conducted using EDEM 2022 software, in conjunction with experimental validation, to investigate the vibratory screening behavior of walnut shell–kernel mixtures under varying vibration frequencies and sieve surface inclinations. The underlying mechanisms of kernel loss and shell retention were elucidated by analyzing particle kinematics and contact characteristics, including velocity, collision frequency, and contact force on the sieve surface. The results indicate that both cleaning rate and loss rate exhibited an increasing trend with increasing sieve inclination. In contrast, as vibration frequency increases, the cleaning rate decreases while the loss rate increases. These findings reveal an inherent and irreconcilable trade-off between cleaning efficiency and material loss when solely adjusting operating parameters. Even under the selected operating conditions (vibration frequency of 14 Hz and screen inclination of 16°), the cleaning rate reached only 93.14% and 93.44%, while the loss rate remained as high as 14.74% and 19.92%, respectively. Mechanistically, kernel loss is primarily attributed to frequent particle–particle collisions during transport, which reduce kernel velocity and may even induce reverse motion, preventing effective recovery. Shell retention, however, arises from early-stage entrainment within dense particle beds and subsequent collision-driven migration toward the kernel outlet, ultimately hindering complete separation. Based on these mechanisms, an airflow-assisted reduction strategy was proposed and experimentally validated. Compared with the optimal single-factor operating conditions, the proposed strategy improved the cleaning rate by 2.04 and 1.74 percentage points and reduced the loss rate by 6.59 and 11.77 percentage points. These findings provide valuable references for the design and optimization of mechanized sorting equipment.