Exploring the use of agent-based models for electric vehicle driver behaviour simulation

Feng, Zixin (2026) Exploring the use of agent-based models for electric vehicle driver behaviour simulation. PhD thesis, University of Glasgow.

Abstract

With the rapid expansion of public electric vehicle (EV) charging infrastructure, the network has progressed beyond its early deployment stage, but remains far from achieving the maturity of conventional refuelling systems. Despite the increasing number of public chargers, many drivers still have difficulties finding available chargers at convenient locations or facing long waiting time before charging. It is important to better understand the charging demand of EV drivers for the sustainable placement of future public chargers.

The aim of this research is to explore the driving and charging behaviours of EV drivers, estimate the spatial distribution of their charging demand, and identify optimal locations for future public chargers in Great Britain (GB). The research first explores the usage patterns of existing public charger network and quantifies their relationships with various contextual factors using spatial regression models. To move beyond these observed charger usage patterns and explore the individual-level behaviours that generate the charging demand, an Agent-Based Model (ABM) is then developed to simulate the driving and charging behaviours of EV drivers across GB. Building on the rule-based ABM, the research further improves the predictive accuracy of the agents by introducing an agentbased reinforcement learning model. The model can capture how EV drivers adaptively learn and adjust their behaviours through repeated exposure to the changing environment, and more closely reflect human decision-making processes. Finally, the research developed a simulation-optimisation framework to integrate complex individual-level driver behaviours with system-level spatial location optimisation. Based on the charging demand generated by the simulation models, a spatial optimisation model is developed to find the optimal locations of future public chargers. The agent-based RL model is then reimplemented to explore how EV drivers respond to the expanded infrastructure network and adjust their charging behaviours accordingly.

The results provide granular estimates of the spatial distribution of charging demand across GB, and the proposed framework offers a new perspective for behaviour-aware planning of public EV charging infrastructure. The simulation-optimisation framework can also be applied to other case study areas and contribute to more sustainable charging network planning.

• Item Type: Thesis (PhD)
• Qualification Level: Doctoral
• Subjects: H Social Sciences > HE Transportation and Communications; Q Science > QA Mathematics > QA75 Electronic computers. Computer science; T Technology > TL Motor vehicles. Aeronautics. Astronautics
• Colleges/Schools: College of Social Sciences > School of Social and Political Sciences
• Supervisor's Name: Zhao, Professor Qunshan and Heppenstall, Professor Alison
• Date of Award: 2026
• Depositing User: Theses Team
• Unique ID: glathesis:2026-86249
• Copyright: Copyright of this thesis is held by the author.
• Date Deposited: 24 Sep 2026 13:42
• Last Modified: 24 Sep 2026 13:44
• Thesis DOI: 10.5525/gla.thesis.86249
• URI: https://theses.gla.ac.uk/id/eprint/86249

Related URLs

• Article DOI
• Article DOI
• Article DOI
• Conference proceeding
• Conference proceeding
• Conference proceeding