2E-VRP-SCS Data

Inspired by the lane-sharing phenomenon in the city logistics practices, a concept named the “sharing-lane crossdock satellite” (SCS) is introduced. We introduce the two-echelon vehicle routing problem with SCSs (2E-VRP-SCS). On the first echelon, 1st-echelon vehicles depart from the city distribution center (CDC) to serve SCSs. At authorized time windows, 1st-echelon vehicles can park at SCSs for cargo transshipment between vehicles. On the second echelon, 2nd-echelon vehicles receive cargoes to service customers. SCSs are used to perform the direct transshipment that is defined as moving cargoes directly from 1st-echelon vehicles to 2nd-echelon vehicles, with no storing. Each SCS has several time windows. The 2E-VRP-SCS network includes one CDC, a number of SCSs, a number of customers, and arcs. A homogeneous fleet of 1st-echelon vehicles is available at the CDC. Second-echelon vehicles departing from each SCS serve customers. At SCSs, there is a constant transshipment speed of cargoes being moved from 1st-echelon vehicles to 2nd-echelon vehicles, and the transshipment speed is determined by the cargo volume per hour. At a time window of one SCS, “the available transshipment capacity” = “the transshipment speed” × “the remaining time of the time window”. In a route, a vehicle can visit an SCS or one customer at most once, and constraints on route duration must be respected. At a time window of an SCS, there parks no more than one 1st-echelon vehicle. Direct transshipment is considered a one-to-one operation. The 2E-VRP-SCS objective is to minimize the vehicle working time. We design 35 small-scale instances. The number (NumS) of included SCSs is 1, 2 or 3. The number (NumC) of customers is 5, 6, 8, 9 or 10. Each small-scale instance is named by S-NumS-NumC-No. (No. is 1, 2, 3, 4 or 5). The network is abstracted on a graph with a grid of 1 km. The CDC is located at the center node of the graph. Other nodes are randomly selected to act as SCS and customer locations. Customer demand is randomly estimated. The whole time window of SCS m is confirmed beforehand. Large-scale instances are designed by referring to practical data. We observe the situation of traffic flows on some roads on several working days. Several lane-spaces are empirically chosen to make up the SCS set. The included SCSs are randomly chosen from the SCS set. The distance between any two nodes on the first echelon is calculated through the latitudes and longitudes of nodes. We supplement some data by the method of generating small-scale instances. We design 42 large-scale instances that are denoted as L-NumS-NumC-No. (No. is 1, 2 or 3). Of the large-scale instances, NumS is 5, 10, 20 or 30. NumC is 50, 75, 100, 150, 200, 250, 300, 400, 500 or 600, which is chosen by referring to NumS.受城市物流实践中的共享车道现象启发，本文提出“共享车道中转卫星（sharing-lane crossdock satellite, SCS）”概念，并构建了带SCS的两级车辆路径问题（two-echelon vehicle routing problem with SCSs, 2E-VRP-SCS）。在第一梯队作业环节，第一梯队车辆（1st-echelon vehicles）从城市配送中心（city distribution center, CDC）出发，服务各SCS节点。在指定的时间窗（time windows）内，第一梯队车辆可停靠SCS节点，完成跨车辆间的货物中转。第二梯队作业阶段，第二梯队车辆（2nd-echelon vehicles）接收货物后为终端客户提供配送服务。SCS节点支持直接中转（direct transshipment）模式：即货物无需经过仓储环节，直接从第一梯队车辆转运至第二梯队车辆。每个SCS节点均配置多个时间窗。 2E-VRP-SCS的实验网络包含1个CDC、若干SCS节点、若干终端客户以及弧路段。CDC配备一支同质性第一梯队车辆车队。从各SCS节点出发的第二梯队车辆负责服务对应区域的客户。在SCS节点处，货物从第一梯队车辆转运至第二梯队车辆的中转速率固定，该速率由每小时货物吞吐量决定。在单个SCS节点的时间窗内，“可用中转容量”=“中转速率”ד该时间窗剩余时长”。 路径约束方面，单台车辆最多仅可访问1个SCS节点或1个终端客户一次，且需严格遵守路径时长限制。在单个SCS节点的时间窗内，停靠的第一梯队车辆不得超过1台。直接中转采用一对一作业模式。2E-VRP-SCS的优化目标为最小化车辆总作业时长。 本文共设计35组小规模算例。每组小规模算例中，SCS节点数量（NumS）为1、2或3，终端客户数量（NumC）为5、6、8、9或10。每个小规模算例的命名格式为S-NumS-NumC-No.（其中No.取值为1、2、3、4或5）。实验网络基于1km网格的抽象图构建，CDC位于图的中心节点，其余节点随机分配为SCS节点或客户节点位置。客户需求为随机生成的估算值，SCS节点m的全时段时间窗均预先设定。 大规模算例则参考实际运营数据设计：研究团队统计了多个工作日内部分路段的交通流量情况，通过经验选取若干车道空间构建SCS候选集，再从该候选集中随机选取对应数量的SCS节点。第一梯队任意两节点间的距离通过节点经纬度计算得到。部分缺失数据通过小规模算例的生成方法补充。本文共设计42组大规模算例，命名格式为L-NumS-NumC-No.（其中No.取值为1、2或3）。大规模算例的NumS取值为5、10、20或30，NumC取值为50、75、100、150、200、250、300、400、500或600，具体取值参考NumS确定。