Last-Mile Logistics Simulation in SUMO
3The last mile is the most expensive part of logistics supply chain and directly impacts cities' citizens with the presence of delivery vehicles. In the currently common state of last mile delivery, every customer is delivered by the instructed company individually. The origin is the company’s local logistics hub while the destination are all the addresses that must be delivered. To compare the current state of last mile logistics with alternative scenarios, the current state must be modeled properly in the first place.
Within the project Bauhaus.MobilityLab, we link SUMO with XCargo to simulate the last mile logistics in a real-world scenario in Erfurt, Germany. The demand of packets for each address within the modeled area is provided by a quantity structure generated with XCargo. The resulting CVRP (Capacitated Vehicle Routing Problem) for the whole area is also solved using XCargo. The corresponding trips are used as input for the SUMO model individually. In the first step, for each delivery address a corresponding vehicle stop position is assumed. If vehicle stops are close to each other’s, the stops are clustered, i.e., the vehicle stops only once delivering the corresponding addresses. Based on the newly generated stops a new VRP (Vehicle Routing Problem) is solved respecting the actual distances between the stops in the SUMO network with Google OR Tools. Finally, with the assumption of parameters for the vehicle and stop times, the scenario is simulated in SUMO and the simulation is compared with real world data and experience.
In future work, the VISUM-Model of the city of Erfurt will be used to generate realistic traffic in the modeled SUMO network and methods to consider traffic in last mile logistics planning will be investigated. Furthermore, the model will be extended on alternative scenarios such as cargo bikes, mini-hubs, and more.