Agent-based solving of the 2-echelon Vehicle Routing Problem

Building (urban) logistic networks is complex. It also includes the question of network design: Using a direct (one-tier) transport system from the depot to the customer. Or is it better to implement a two-echelon distribution system, transporting the goods with large vehicles to a transshipment hub near to the customer and smaller vehicles for the last-mile distribution. These decisions must be made by the Logistic Service Provider (LSP). In this study, we investigate the behavior of an LSP in an agent-based simulation framework with different plans: direct delivery, and delivery via a hub. Depending on different input values, e.g. fixed and variable costs of the vehicle, or costs for the hub, the LSP performs its planning and selects the plan with the best score. Tour planing at different stages of the LSP's transport chains is done by solving a Vehicle Routing Problem (VRP). As a result, we have shown, that the LSP selects the best plan in an artificial grid scenario and in different simulation settings, with different number of jobs, cost settings, and a vehicle-type specific cordon toll. In the next steps, we will apply it to existing case studies in large cities, such as Berlin, Germany.