A DEM-PSO-Kriging framework for minimizing mechanical damage and nutrient pollution from controlled release fertilizers

Controlled release fertilizers constitute a high-value segment of industrial crop products. However, their efficacy in sustainable agriculture relies heavily on functional integrity, which is often compromised by mechanical damage during application, leading to nutrient pollution and economic loss. This study introduces an integrated Discrete Element Method (DEM)- Particle Swarm Optimization (PSO)-Kriging framework for the prescriptive design of fertilizer application equipment. A high-fidelity, multiscale Bonded Particle Model (BPM) was developed and rigorously calibrated to accurately simulate fracture mechanics. Embedded within a surrogate-assisted optimization strategy, this model navigated the operational design space of a grooved-wheel discharge device. Our PSO-Kriging approach, adept at capturing highly nonlinear responses, identified global optimum parameters that preemptively minimize product damage, outperforming the traditional Response Surface Methodology (RSM) in both predictive accuracy and robustness. The optimized configuration achieved a remarkably low damage rate of 1.109 %, which was experimentally validated with a minimal relative error of 3.757 %. Furthermore, we elucidated the underlying damage mechanisms and established a direct causal link between specific mechanical failure modes and the risk of nutrient burst release. This work establishes pollution prevention by design as an indispensable strategy for the agricultural industry, thereby safeguarding the lifecycle performance and economic value of these industrial crop products and contributing decisively to cleaner agricultural production chains.