Abstract
The object of this study is the process of walnut shelling in a reciprocating millstone, implementing combined compression, shear, and torsion loading. The problem being addressed is the lack of well-founded modes that simultaneously ensure complete shell destruction and a high yield of whole kernels—two conflicting requirements in processing. This study extends previous work that determined the shell fracture forces and the probabilistic nature of its failure; this work transitions from the strength characteristics of the object to machine performance. A three-factor rotatable design of a central composite experiment (Box design) consisting of twenty trials was implemented. The factors included the angular velocity of the moving upper millstone (1.38–4.90 rad/s), the number of nut revolutions in the working channel (0.33–1.17 revolutions), and the roughness pitch of the working surfaces (7.95–18.05 mm). Two responses were recorded: shell fracture rate and whole kernel yield. Second-order regression models were obtained and tested using analysis of variance. It was found that shell fracture rate increases with increasing angular velocity and rotational speed and decreases with increasing roughness pitch, while whole kernel yield decreases with increasing angular velocity and exhibits internal maxima at rotational speed and roughness pitch. Angular velocity is the dominant factor for both responses and serves as the main source of the tradeoff between cracking and kernel integrity. A compromise regime was identified: angular velocity of approximately 2.09 rad/s, approximately 1.0 nut revolution, and a roughness pitch of approximately 10 mm, resulting in a shell fracture rate of approximately 94.5% with a whole kernel yield of approximately 88.9%. The results serve as a basis for selecting walnut shelling machine parameters