Three-dimensional loading vehicle routing problem with split pickups and time windows (3L-SPVRP-TW): instances and results

This folder provides instances for a three-dimensional loading vehicle routing problem with split pickups and time windows (3L-SPVRP-TW) as well as results obtained with a three-phase heuristic.

The 240 instances are built on the benchmark from Krebs et al. (2021) adding real features extracted from our survey among Belgian service providers. To have more information about the instances, please refer to the README file included in the "PracticalInstances" folder.

The folder "ResultsInstancesKrebs" contains the solutions provided by the three-phase heuristic on the instances from Krebs et al. (2021), while the folder "ResultsPracticalInstances" contains the solutions provided by the three-phase heuristic on the 240 practical instances presented hereabove.
In both folder, each solution starts with the name of the instance.
In "ResultsInstancesKrebs", we give the repetition identification (15 runs per instance, called “Rep”), and the constraint set as referenced by Krebs in her Github. In "ResultsPracticalInstances”, we write the information about the number of customer’s shifts (2 or 4), the constraints considered (split or not, reachability or sequential loading), the EPs’ selection criterion (TDLF or DBLF) and the speed during the peak and off-peak hours.
In both files, we then provide the number of used vehicles in the solution, the total cost (penalty + travel distance), the computation time (in ms) as well as the number of iterations in the GVNS. Next, each route is described with the sequence of customers and the arrival and departure times associated to each of them. Finally, the loading pattern in the vehicle is given (customer id, box type, Boolean stating if the box is rotated, the location of the front left bottom, the characteristics of the box, namely the length, width, height and mass).

The file “SummaryResultsPracticalInstances.csv” contains the total travel distance, the number of vehicles in the solution and the computational time (in seconds) for the 240 practical instances.

Krebs C, Ehmke JF, (2021). Axle Weights in combined Vehicle Routing and Container Loading Problems. EURO Journal on Transportation and Logistics, Volume 10, ISSN 2192-4376, DOI: 10.1016/j.ejtl.2021.100043, URL https://www.sciencedirect.com/science/article/pii/S2192437621000157

本文件夹提供了包含分拣提货和时效窗口的三维装载车辆路径问题（3L-SPVRP-TW）的实例，以及采用三阶段启发式算法得到的结果。

该文件夹中的240个实例基于Krebs等（2021年）的基准数据集构建，并加入了从比利时服务提供商调查中提取的实征特征。关于实例的详细信息，请参阅“PracticalInstances”文件夹中包含的README文件。

“ResultsInstancesKrebs”文件夹包含了在Krebs等（2021年）的实例上使用三阶段启发式算法得到的解决方案，而“ResultsPracticalInstances”文件夹则包含了在本描述中提到的240个实用实例上使用相同启发式算法得到的解决方案。

在两个文件夹中，每个解决方案均以实例名称开头。在“ResultsInstancesKrebs”中，我们提供了重复识别信息（每个实例15次运行，称为“Rep”），以及Krebs在其GitHub上引用的约束集。在“ResultsPracticalInstances”中，我们记录了客户班次数量（2或4）、考虑的约束条件（是否分拣、可达性或顺序装载）、EPs的选择标准（TDLF或DBLF）以及高峰和非高峰时段的速度。

在两个文件中，我们还提供了解决方案中使用的车辆数量、总成本（惩罚费用+行驶距离）、计算时间（以毫秒计）以及在GVNS中的迭代次数。随后，每条路线均以客户顺序及其到达和离开时间进行描述。最后，给出了车辆中的装载模式（客户ID、箱型、布尔值表示箱体是否旋转、前左下角位置、箱体特性，即长度、宽度、高度和质量）。

“SummaryResultsPracticalInstances.csv”文件包含了240个实用实例的总行驶距离、解决方案中的车辆数量以及计算时间（以秒计）。

Krebs C, Ehmke JF, (2021). 轴重联合车辆路径与集装箱装载问题研究。EURO运输与物流杂志，第10卷，ISSN 2192-4376，DOI: 10.1016/j.ejtl.2021.100043，URL https://www.sciencedirect.com/science/article/pii/S2192437621000157