Effects of Orchard Characteristics and Operator Performance on Harvesting Rate of a Mechanical Sweet Cherry Harvester

Abstract

A model was developed to simulate sweet cherry harvesting with a mirrored-pair mechanical harvest system that removes fruit by transferring vibrational energy to tree limbs through an impactor. Six orchard characteristic variables (ST – tree spacing; SR – row spacing; NT – number of trees per row; NR – number of rows; NTB – number of branches per tree; and WTF – total weight of fruit per tree) and three harvester/operator characteristic variables (vH – forward speed of harvester, tIP – time to position impactor on actuation point, and tPS – shaking time per actuation point) were the main inputs to the model. Total harvest time (tTH) and harvesting rate were the two output variables of the model. Harvesting rate was evaluated with three different measures: time rate of area coverage (RAC), time rate of tree coverage (RTC), and time rate of fruit removal by weight (RFR). The model was validated with field data, showing very close predictions with modeling efficiencies of 99%, 86%, 82% and 84% respectively for tTH, RAC, RTC, and RFR. Local sensitivity analysis was conducted varying the input variables in five different levels in order to observe their effect on the output variables. A global sensitivity analysis was also performed to identify input variables with significant effects on the output variables. Data from a complete factorial experiment with three levels of input variables in 19,683 combinations was used to perform the global sensitivity analysis. . It was revealed that ST and SR only affected RAC by defining the unit area occupied by a single tree. NT and NR only affected tTH by determining the number of trees to be harvested, but had no effect on harvesting rate. NTB greatly affected harvesting time and all measures of harvesting rate, and was identified as the most important variable. WTF only affected RFR by determining how much fruit is removed in a single shaking event. Of the harvester/operator variables, tIP affected all the outputs the most whereas vH affected none. Except for RAC, which was least affected by SR, tPS had the least effect amid all the significant input variables. These results provide explanation for achieving different harvesting rates in different orchard settings, and can be used to optimize orchard characteristics and adjust operator behaviors for improved performance in mechanical sweet cherry harvesting.