EURO 2024 Copenhagen
Abstract Submission

EURO-Online login

2998. Smart refueling decisions using reinforcement learning approach: a case study from trucking industry

Invited abstract in session MC-25: Discrete, continuous or stochastic optimization and control in networks, transportation and design III, stream Combinatorial Optimization.

Monday, 12:30-14:00
Room: 011 (building: 208)

Authors (first author is the speaker)

1. Amin Asadi
Industrial Engineering & Management (IEBIS), University of Twente
2. Myrthe Kruit
University of Twente
3. Behzad Mosalla Nezhad
Industrial Engineering, Tecnologico de Monterrey
4. Sebastian Piest
Industrial Engineering & Management (IEBIS), University of Twente

Abstract

Transportation services play a pivotal role in international trade, with trucking companies handling approximately 77% of inland freight. Refueling costs constitute a substantial portion of operational expenses, ranging between 20% and 30% in Europe. To address the complexities of refueling decisions in the trucking industry, we propose a data-driven sequential decision-making framework. Our approach captures price fluctuations along trucking routes, incorporating factors such as region, time, and possible price agreements with gas stations. Leveraging the Markov Decision Problem and Reinforcement Learning (RL) methods, we develop a robust framework capable of minimizing refueling costs while accounting for fuel price uncertainties. In a real-world case study conducted with a logistics company in the Netherlands, our framework demonstrated significant savings in refueling costs. Through extensive experimentation, we validate the superiority of our approach compared to in-practice and benchmark policies, showcasing the practical utility and effectiveness of our proposed framework.

Keywords

Status: accepted

Back to the list of papers