In many industries, data changed the way decisions are made, driving performance and competitiveness. Transportation and logistics are no exception. Companies are analyzing data from their vehicles to improve driving behavior, optimize transportation routes, and improve their operations. Over the years, vehicle routing problems (VRPs) were solved with the assumption that each pair of customers is connected by a single arc with a constant travel time. With the vast availability of data, one can now solve the time-dependent shortest path VRP (TD-SPVRP) which is a TD-VRP modeled on a real road network and where each pair of customers is connected by numerous time-varying paths. This increased the precision but also the complexity related to the size of the network. In this paper, we evaluate the impact of the road network precision on the problem and its solution. When the highest level of precision is required, one needs to work on the complete network but at the cost of a much more difficult problem. On the other hand, low computing effort is needed when working only with the customer nodes but at the cost of losing precision. We discuss some network reduction procedures to reduce the graph size. We evaluate the reduction procedures on real data instances using multiple heuristics and a metaheuristic that can be used to solve the TD-SPVRP and the TD-VRP. This allows us to evaluate the impact of the graph size and its structure on the quality of the solutions, their precision, and the computational effort.
https://doi.org/10.1007/s43069-025-00467-4Cite as:
@inproceedings{jaballah2025time,
title={Time-Dependent Routing and Road Network Precision},
author={Jaballah, Rabie and Coelho, Leandro C and Renaud, Jacques},
booktitle={Operations Research Forum},
volume={6},
number={2},
pages={1--23},
year={2025},
organization={Springer}
}