Analysis of Danish Truck Tolling GPS Data

Abstract

This study presents an analysis of GPS data obtained from the Danish truck tolling system, covering a one-week period with more than 800 million observations across over 60,000 trucks. The analysis extracts mobility patterns from high-resolution data by segmenting raw GPS sequences into trips. These trips are subsequently used in a simulation-based framework to estimate future charging demand.
The simulation is based on a simplified range-based model, where charging events emerge endogenously from observed driving behavior. DBSCAN is applied on simulated charging locations to identify areas of concentrated demand. These clusters form the basis of an optimization problem (MILP) in which cluster centroid locations serve as candidate sites for new charging infrastructure. A key advantage of DBSCAN is its ability to distinguish between clusters and noise, thereby reducing the number of candidate locations and the computational complexity of the MILP.
The results reveal spatial and temporal concentration of charging demand along major transport corridors and during daytime hours. The findings highlight regions with limited existing coverage as key candidates for infrastructure expansion. While the model relies on simplified assumptions, it provides a scalable and data-driven approach for planning charging infrastructure under electrification scenarios.