Facility and Routing Decisions in Truck-Drone Distribution

Abstract

Drones have the potential to be major players in the field of last-mile delivery, especially with the rise of e-commerce. However, they face several technological and regulatory restrictions for large-scale implementation. Combining drones with traditional ground delivery vehicles can bridge this gap while achieving significant improvements in distribution cost and speed over vehicle delivery alone. This research project focuses on modeling and solving the Location-Routing Problem with drones as an ancillary mode of delivery. The goal is to develop a model that identifies the optimal locations for the distribution facilities, as well as the set of combined routes that the trucks and drones will follow to deliver parcels to customers. To solve this problem, a two-steps metaheuristic approach is developed and implemented. The customer locations are first grouped into clusters with centroid positions where trucks would park, dispatch, and retrieve the onboard drones that perform the last step of the delivery. Once the optimal truck parking locations are identified, the selection of the optimal distribution facilities and the truck routes are determined simultaneously, by implementing the Multiple Ant Colony Optimization algorithm. The validation of the model revealed high reliability with a 1% average optimality gap from the exact solution. When applying the model to a real road network, with 200 customers and 5 candidate depot locations, the model confirmed a 24% saving in daily distribution costs from adding 3 drones to every delivery truck. The savings opportunity is less sensitive to the number of drones per truck and the drones’ speed, and more sensitive to the trucks’ speed and the drones’ traveling cost per mile. The analysis reveals that adding drones will generate savings as long as the traveling cost of drones is lower than that of trucks at around $4.00 on average for last-mile delivery.