مسئله يكپارچه دسته‌بندي، مسيريابي و تعديل بار كاري عملياتي با تخصيص پوياي نيروي كار در يك سيستم برداشت سفارش منطقه‌اي موازي

The growth an‎d competitiveness of retailers-especially online retailers-depend heavily on fast an‎d accurate order delivery. As a result, the management of warehouse operations, particularly the order picking process, is of critical importance. The majority of time spent during order picking is because of unproductive movement of pickers within the storage area. One of the proposed methods to reduce this time is zone picking, which divides the warehouse into several zones an‎d assigns each picker to a specific zone. As a result, pickers spend less time traveling within the warehouse. However, due to the distribution of items across zones an‎d limited mobility space, the workload in one zone may exceed that of others. This imbalance in workload leads to idle times for pickers, which results in longer picking times an‎d reduced system efficiency. In the literature on zone picking, various studies have addressed workload balancing during system design, storage location assignment, an‎d operational phases. Yet, there remains a research gap in addressing workload balancing at the operational level alongside other operational policies of the order picking process, such as order batching an‎d picker routing. Given these explanations, this study introduces an integrated optimization problem that considers order batching, picker routing, an‎d workload balancing problems in a low-level parallel zone picking system for the first time in the literature. Furthermore, this study proposes a novel operational-level workload balancing method, called dynamic workforce allocation. In this approach, before picking each batch of orders, pickers are allowed to change their zones an‎d move to a zone with higher workloads. This method balances the workload across zones an‎d improves system efficiency. To minimize the order picking process completion time, a mixed-integer linear programming model is presented to formulate the integrated problem. Due to the complexity of the problem, a hybrid solution framework based on genetic algorithm an‎d variable neighborhood descent is developed for large-scale instances. According to this framework, an‎d by presenting two different picker-to-zone assignment methods an‎d two different routing strategies, four solution algorithms are proposed. Numerical results show that the combination of a greedy approach for picker assignment an‎d the nearest-neighbor algorithm for routing provides better solution quality compared to other methods by an average of 33.53% in medium-scale instances an‎d 10.76% in large-scale instances. This study also compared the numerical results of the problem using dynamic picker allocation method with a static allocation algorithm that prohibits zone changes. The results indicated that the dynamic method reduces the order picking time by an average of 16.14%.

مهدي ايران پور

مهدي علينقيان , محمد رئيسي نافچي