Dataset for a Dynamic Multi-Period Vehicle Routing Problem
Description
This dataset presents the information related to the instances used in the experimental study of the paper "Dynamic Multi-Period Vehicle Routing with Touting". It has two sets of instances. The first set of instances is related to the historical data of a waste collection company, which spans a three-month period, while the second instance set is generated based on VRPTW instances of Solomon (1987). Within the first instance set, there is data for two different drivers, who cover different geographical areas, presented in separate excel files (Instances_Driver1.xlsx and Instances_Driver2.xlsx). Each dataset includes i) three-months historical collections data of the company, which is used to compare the heuristic solutions with the ones obtained by an exact solver; ii) the normal distribution parameters that represent the customers' demand profiles, which are used to create simulation instances used in the experimental study for touting strategies. These distributions are used to generate random demand on each planning day; iii) distances and travel times between each node in the network. These information are stored in three sheets of the files. The details are explained in the attached file Instances_Explanation.txt. Driver 1 and 2 have 142 and 125 unique customers, respectively. However, the historical data includes a total of 273 and 260 orders for Driver 1 and Driver 2, respectively, which means that some customers have multiple orders within that 3-month period. In addition, some of these orders belong to the customers who ordered for the first time, in which case the distribution parameters cannot be obtained due to lack of history. To mimic these customers, we have selected a subset of the unique customers and used their distributions to generate these extra customers' parameters. Finally, the set of customers includes 215 and 205 customers for Driver 1 and Driver 2, respectively. The second instance set includes the data generated based on the VRPTW instances of Solomon (1987), specifically C101, R101, and RC101. The file Solomon-Based Instances.xlsx has three 3 sheets for each of these instances, which present the the normal distribution parameters representing the customers' demand profiles. More details are provided in the attached file Instances_Explanation.txt. Finally, the problem parameters, such as vehicle capacity and time limits are provided in the file "Problem Parameters.txt". References Solomon, M. M. (1987). Algorithms for the vehicle routing and scheduling problems with time window constraints. Operations research, 35 (2), 254-265.