Robust Routing for a Mixed Fleet of Heavy-Duty Trucks with Pickup and Delivery Under Energy Consumption Uncertainty

Ruiting Wang*, Patrick Keyantuo, Teng Zeng, Jairo Sandoval, Aashrith Vishwanath, Hoseinali Borhan, Scott Moura

Applied Energy (2024)

For more information please refer to Full Paper.

Abstract

Electrification of the truck fleet has the potential to reduce the “harder-to-abate” emissions of logistics significantly, but is generally considered to be very challenging. In this study, we focus on the energy-efficient routing of a mixed fleet of conventional and electric heavy-duty trucks with pickup and delivery under energy consumption uncertainty. We propose an energy consumption model that accounts for realistic driving dynamics, road conditions, weight, and distances. Integrating this model into the routing problem, we address energy consumption uncertainty using second-order cone mixed-integer programming. A quantitative case study is then performed on the operating costs and \(\text{CO}_2\) emissions benefits of electrifying heavy-duty trucks, which demonstrates improved fleet performance with optimal operating results. Scenarios with different parameter settings are tested to compare different performance metrics and provide practical insights. We evaluate routing decisions to demonstrate that stochastic optimization is necessary for reliable truck routing and produces robust results that significantly reduce capacity violations in route execution.


Why is robust routing necessary?

Some truck routes from January 2, 2021, under different uncertainty settings, where:

  • \(1-\gamma\) is the reliability factor of the routing
  • \(\sigma\) is the uncertainty standard deviation of the energy consumption from the mean.

Routing decisions can be drastically different when energy uncertainty is considered! More details in paper.
Deterministic \(\gamma = 0.1, \sigma = 0.1\)
\(\gamma = 0.2, \sigma = 0.1\) \(\gamma = 0.2, \sigma = 0.05\)