CFD-DEM Evaluation of Particle Circulation and High-Percentile Contact Loading in a Draft-Tube Fluidized-Bed Seed Coater for Chinese Cabbage Seeds

Stable circulation and limited mechanical loading are key requirements in the fluidized-bed coating of small vegetable seeds. For Chinese cabbage seeds, their small mass, irregular geometry, and mechanical sensitivity make it difficult to evaluate operating conditions using only global indicators, such as mean particle velocity and bed expansion height. In this study, a two-way coupled computational fluid dynamics–discrete element method (CFD-DEM) model was developed for a draft-tube fluidized-bed seed coater. Chinese cabbage seeds were represented by seven-sphere clumps, and an L9(33) orthogonal array was used as a screening design to examine inlet air velocity, initial bed height, and bottom circulation inlet gap. The evaluation combined cycle time distribution (CTD), the global low-speed particle fraction Rs, the 95th-percentile normal contact force F95,n, and the normalized high-percentile contact-load ratio ηc,95. Because the L9 array cannot resolve interactions or support a full quadratic model, factor effects were interpreted descriptively within the investigated range rather than as a confirmatory global optimization. Initial bed height produced the largest descriptive contribution to the circulation period, Rs, and F95,n. Increasing inlet air velocity shortened the circulation period and reduced Rs but increased high-percentile contact loading. Across the nine cases, F95,n ranged from 7.50 to 14.50 mN and ηc,95 from 0.18% to 0.35%; ηc,95 is used only as a normalized load ratio and not as a validated probability of seed damage. Case 4 ranked first under equal weighting and contact-load-priority weighting, whereas Case 7 ranked first under circulation-priority weighting. Case 4 is therefore described as a weight-dependent balanced candidate within the tested parameter range. Prototype experiments reproduced the ordering of circulation periods for three dry operating conditions, supporting qualitative consistency between simulated and observed circulation behavior. Visible breakage remained below 0.5%, but this observation provides only preliminary qualitative correspondence with the simulated contact-load trend. The proposed framework is intended for dry-stage screening and does not directly predict wet-coating quality, adhesion, agglomeration, or germination performance.