Data-driven Decisions for Etruck Infrastructure: Optimal Placement of Charging Stations in a Route Network. A Case Study With DFDS A/S.

Abstrakt

The goal of this thesis is to assess how to leverage Data Science methods to find optimal positions for electrical vehicle charging stations within a logistics provider’s route network. More specifically, this research is carried out in collaboration with DFDS A/S, an European leader in providing both transportation and logistics services, who is aiming at becoming a carbon neutral company by 2050. In this regard, part of the company’s decarbonization plan foresees to replace 25% of their truck fleet with electrical trucks (eTrucks). DFDS now is in need of a data-driven approach to answer the question where to deploy the eTrucks and where to install charging stations.
In this research, the Data Science methods we have focused on are Visual Analytics and Graph Theory. In a first step, Visual Analytics has been used used to better understand the complex transportation network at hand, both in volume and geographical terms. Here, the created dashboard has been contributed to identifying six focus countries for charging stations (and consequently eTrucks) deployment: United Kingdom, Belgium, the Netherlands, Germany, Denmark and Sweden.
Next, Graph Theory techniques have been deployed to further examine these focus areas. For each country, a graph representing the route network within the relevant distance range for eTrucks deployment has been constructed and visualized. The distance range from 5 to 300 km served as the baseline for this analysis, as this range constitutes the current common range of eTruck operations. By first deploying a community detection algorithm in order to identify important substructures within the route networks and then calculating the most important nodes based on the identified communities, we were able to identify suitable spots for EV charging stations within each network.
The results have then been compared to scenarios with increased (e.g. via technological advancements) and decreased (e.g. because of tough weather conditions) eTruck ranges in order to assess the scalability of the analyses. Lastly, based on the various analyses and calculations, tangible recommendations to DFDS were made as to which parts of the route networks to electrify and where to install charging stations, based on the identified electrification potentials for the focus countries.

Source code available at: https://github.com/em1899/master-thesis-project