In this work, we present and evaluate heuristic techniques for a real-world territory design problem of a major dairy company which produces and distributes perishable products. The problem calls for grouping customers into geographic districts, with the objective of minimising the total operational cost, computed as a function of the fixed costs of the districts and the routing costs. Two inter-connected decision levels have to be tackled: partitioning customers into districts and routing vehicles according to complex operational constraints. To solve the problem, a hybrid multi-population genetic algorithm is designed, enhanced with several evolution and search techniques. The proposed design is extensively tested on instances derived from the literature and on real-world large-scale instances, involving more than 1000 customers. The results show the effectiveness of the different components of the algorithm and the feedback from the company’s planners confirms that it produces high-quality, operational solutions. Additionally, we explore some managerial findings with respect to the adoption of alternative objectives and service requirements.

在这项工作中，我们介绍和评估启发式技术，用于解决生产和分销易腐烂产品的大型乳品公司的实际领域设计问题。该问题要求将客户分组到地理区域中，以使总运营成本最小化，该总运营成本是根据区域固定成本和路由成本来计算的。必须解决两个相互关联的决策层：将客户划分为多个区域，并根据复杂的操作约束来确定车辆的路线。为了解决该问题，设计了一种混合多种群遗传算法，并通过几种进化和搜索技术对其进行了增强。拟议的设计已在大量文献中进行了实例测试，并在涉及1000多个客户的现实世界中的大型实例中进行了测试。结果表明该算法不同组件的有效性，并且公司规划人员的反馈证实了该算法可产生高质量的可操作解决方案。此外，我们探讨了有关采用替代目标和服务要求的一些管理发现。