Abstract

Harvest is a critical stage of the fruit supply chain that must be coordinated with other stages based on market arrival times and fruit characteristics, such as the degree of sweetness. Moreover, it accounts for the highest fruit losses due to diseases, changing weather conditions, and other factors. Harvest complexity increases when managing multiple orchards, which is a common scenario for fruit export companies. Consequently, this study proposes a multi-objective GRASP (MO-GRASP-C) metaheuristic to support tactical harvest planning for major fruits, minimizing harvest costs, fruit loss, and harvest duration. The MO-GRASP-C metaheuristic incorporates a path relinking algorithm and a clustering method to improve accuracy and eliminate non-dominated solutions while preserving Pareto frontier diversity. This study calibrates the metaheuristic parameters and applies them to a real case study, where tactical harvest plans vary significantly depending on the prioritized objectives. It also compares the results of the MO-GRASP-C metaheuristic with those of a mono-objective GRASP metaheuristic and a multi-objective exact method. In this comparison, the MO-GRASP-C metaheuristic solutions outperform those obtained by the mono-objective GRASP metaheuristic across all three minimization criteria. However, the computational time increases by approximately 13% to obtain the complete set of solutions on the Pareto frontier. Furthermore, according to the inverted generational distance metric, the MO-GRASP-C metaheuristic generates more diverse solutions than the multi-objective exact method, while requiring only approximately 1% of the computational time needed by the exact method. On the other hand, according to the hypervolume metric, the Pareto frontier obtained by the exact method is 0.5% better than the MO-GRASP-C metaheuristic, whereas, according to the INT-SBM metric, it is 17% better than the proposed metaheuristic. These results demonstrate the effectiveness of the multi-objective approach and the flexibility of the MO-GRASP-C metaheuristic in adapting tactical harvest plans to varying conditions and decision-maker preferences.