Continent-scale inventory routing instances

This dataset of inventory routing instances is a fruit of our partnership between Renault Group and the CERMICS laboratory at Ecole des Ponts. Those instances are continent-scale with hundreds of customers, 21 days horizon, and 15 depots on average. Routes can last several days (continuous-time), and 30 types of commodities are involved, leading to bin packing problems when filling trucks. In our paper "Solving a Continent-Scale Inventory Routing
Problem at Renault" we introduce a new large neighborhood search to solve those instances. We hope that sharing them publicly will motivate research on real-world and large-scale inventory routing. Environmental and economical impacts at stake are substantial.

本库存路径规划（inventory routing）实例数据集由雷诺集团（Renault Group）与巴黎高科国立路桥学校（Ecole des Ponts）下属的CERMICS实验室合作研发而成。该类实例均为洲际级规模，平均涵盖数百个客户节点、21天的规划期与15个配送中心（depots）。配送路径可延续多日（采用连续时间建模），涉及30种货品品类，在装载货运车辆时会衍生出装箱问题（bin packing problem）。在我们题为《Solving a Continent-Scale Inventory Routing Problem at Renault》的学术论文中，我们提出了一种全新的大邻域搜索（Large Neighborhood Search）算法以求解此类实例。我们期望通过公开共享该数据集，能够推动针对现实场景与大规模库存路径规划问题的相关研究。此类问题所牵涉的环境与经济影响均极为重大。