Optimizing the utilization of existing vehicle flows in last-mile passenger transport and logistics systems

Abstract

The new era of shared economy has raised our expectations to make mobility more sustainable through better utilization of existing resources and capacity. In this thesis, we focus on the design of transport systems that stimulate multi-purpose trips with the aim of reducing congestion while simultaneously leveraging the existing commuters better. Multi-purpose trips can improve the efficiency, sustainability, and profitability of passenger transport systems, through vehicle relocation in vehicle-sharing systems and ride-sharing. Similar improvements can be made in last-mile logistics systems, through crowd-shipping. A predictive user-based relocation approach through incentives is proposed for car relocation in one-way car-sharing systems. This approach consists of a bi-level optimization approach to find the optimal incentive and a Markov chain to describe the state of the system. Numerical results indicate that these user-based relocations can significantly improve the profit and the service level of a car-sharing system and are more sustainable than staff-based relocations. The effect of congestion on ride-sharing is studied. Numerical results indicate that ride-sharing is more appealing during congestion, due to the high availability of riders and drivers. Thereby, theoretical and numerical results show that bottleneck congestion makes the schedules of riders and drivers more flexible, thereby increasing the matching opportunities. Transfers of riders between modes and between drivers can further improve the performance of a ride-sharing system. By allowing for transfers, the trips of the rider and the driver only need to be partially similar. The problem is formulated as a path-based integer programming problem and travel time uncertainty is included by reformulating the problem as a stochastic programming problem. Crowd-shipping is a last-mile delivery concept in which commuters pick up and deliver parcels on their pre-existing paths. By constructing depot locations for picking up parcels, more potential crowd-shippers can be attracted and the service area can be extended. Numerical results show that determining these depot locations using predictive strategies can improve the overall performance of the system by 15% compared to when non-predictive strategies are used. Using transfers between crowd-shippers allows for further expanding the service area and improving the overall performance. A column and row generation approach is proposed to solve the problem of matching parcels to crowd-shippers. Numerical results show that transfers can improve the service level and profit of the system by 30%.