OPTIMIZATION OF THE CLEANING SYSTEM OF GRAPE HARVESTERS USING THE DISCRETE-ELEMENT METHOD

Grape harvesters are mechanized machines designed to remove grapes from vine trees, process them in a cleaning system, and then store them in onboard bins. These bins are later unloaded into a transport wagon and taken to a vinification facility. Cleaning systems can sometimes fail to completely remove the foreign materials (i.e. leaves, petioles, stems, etc.), which may compromise the vinification process. For this reason, the project focused on the cleaning system by minimizing the presence of foreign materials while maintaining an adequate harvesting throughput. The project main objective was to optimize the cleaning system in grape harvesters by using the Discrete-Element Method (DEM). Individual DEM simulations were validated and used to develop a main crop flow simulation for the optimization of the cleaning system. This optimization included reducing the presence of foreign materials (petioles and leaves) while increasing the crop throughput for the specific grape variety of Cabernet Sauvignon. The physical characteristics and properties of the biological materials (grapes, petioles, leaves) were measured during the 2014 grape harvesting season at three different locations (Aigues-Mortes, Saint-Gervais, and Pauillac) in France. Time constraints limited the number of measured properties at the locations. The results from each location were compared using an ANOVA and a Tukey HSD post-hoc test. Given the natural variability of the biological materials, the three populations were found to be significantly different in most cases. The physical characteristics and properties from the Aigues-Mortes and Pauillac locations were used for the validation process. This was done because these locations had the most complete data sets. During the summer of 2015, a second testing phase took place to validate both the DEM leaf deflection and cleaning system models. The additional experiments consisted of testing the leaf samples in controlled deflections and testing the efficiency of the cleaning system. These experiments used Cabernet Sauvignon leaves shipped from the Vineland Research and Innovation Centre (VRIC) in Ontario. The individual simulations included the inclined plane, rebound surface, leaf deflection, and grape trajectory tests on an inclined conveyor. The inclined plane and rebound simulations were adjusted until the results were within 5% of the experimental test results. The leaf deflection simulations used optimized crop material properties until the simulated leaf behavior matched the actual leaf. Some discrepancies in the DEM simulated leaf shape were identified due to the limitations of the particle creation method. The grape trajectory test results coincided with the DEM simulations at greater conveyor speeds. A moderate difference between the simulations and the experimental tests was present at lower conveyor speeds. A possible cause for this difference may have been the effect of gravity and belt friction on the generation and acceleration of the grapes on the conveyor. A main crop flow simulation that included a conveyor and aspirator was developed using the previously validated simulations. Nine conveyor configurations, which included three belt angles from horizontal (10°, 15°, and 20°) and three speeds (350 rev/min (1.4 m/s), 420 rev/min (1.7 m/s), and 500 rev/min (2.0 m/s)), were tested to optimize the cleaning system performance. Based on the DEM simulations, the 420 rev/min-20° configuration was recommended as the optimal crop conveyor setting. This particular configuration minimized product damage and had an increased aspiration success rate of 9.6% compared to the conventional conveyor settings (420 rev/min-15°)