Facilitating Cooperative Truck Platooning for Energy Savings: Path Planning, Platoon Formation and Benefit Redistribution

Enabled by the connected and automated vehicle (CAV) technology, cooperative truck platooning that offers promising energy savings is likely to be implemented soon. However, as the trucking industry operates in a highly granular manner so that the trucks usually vary in their operation schedules, vehicle types and configurations, it is inevitable that 1) the spontaneous platooning over a spatial network is rare, 2) the total fuel savings vary from platoon to platoon, and 3) the benefit achieved within a platoon differs from position to position, e.g., the lead vehicle always achieves the least fuel-saving. Consequently, trucks from different owners may not have the opportunities to platoon with others if no path coordination is performed. Even if they happen to do so, they may tend to change positions in the formed platoons to achieve greater benefits, yielding behaviorally unstable platoons with less energy savings and more disruptions to traffic flows. This thesis proposes a hierarchical modeling framework to explicate the necessitated strategies that facilitate cooperative truck platooning. An empirical study is first conducted to scrutinize the energy-saving potentials of the U.S. national freight network. By comparing the performance under scheduled platooning and ad-hoc platooning, the author shows that the platooning opportunities can be greatly improved by careful path planning, thereby yielding substantial energy savings. For trucks assembled on the same path and can to platoon together, the second part of the thesis investigates the optimal platoon formation that maximizes total platooning utility and benefits redistribution mechanisms that address the behavioral instability issue. Both centralized and decentralized approaches are proposed. In particular, the decentralized approach employs a dynamic process where individual trucks or formed platoons are assumed to act as rational agents. The agents decide whether to form a larger, better platoon considering their own utilities under the pre-defined benefit reallocation mechanisms. Depending on whether the trucks are single-brand or multi-brand, whether there is a complete information setting or incomplete information setting, three mechanisms, auction, bilateral trade model, and one-sided matching are proposed. The centralized approach yields a near-optimal solution for the whole system and is more computationally efficient than conventional algorithms. The decentralized approach is stable, more flexible, and computational efficient while maintaining acceptable degrees of optimality. The mechanisms proposed can apply to not only under the truck platooning scenario but also other forms of shared mobility.PhDCivil EngineeringUniversity of Michigan, Horace H. Rackham School of Graduate Studieshttp://deepblue.lib.umich.edu/bitstream/2027.42/163047/1/xtsun_1.pd