Game-Theoretic Optimisation of EV Charging Network: Placement and Pricing Strategies via Atomic Congestion Game

The “chicken-and-egg problem” in Electric Vehicle (EV) charging reflects the interdependence between sufficient infrastructure and the demand needed to justify it, a challenge heightened by the UK’s 2030 ban on new combustion engine vehicles. To address this, we propose a joint optimisation model that determines the optimal number of charging points and pricing at each station, while accounting for traffic patterns. From a policy perspective, our model seeks to maximise public benefit by reducing EV users’ social costs, travel and queuing time, and charging fees, while ensuring station operator profitability. We model driver decisions as two interconnected congestion games, one on roads and one at charging stations (CS), and solve for stable outcomes using Nash Equilibrium (NE) strategies. To ensure tractability, we develop an efficient approximation algorithm for the Mixed-Integer Nonlinear Program (MINLP) and introduce a generalisation technique that targets charger placement at high-impact locations, enhancing scalability to larger Transportation Networks (TN). Applied to a benchmark case, the model reduces overall social cost by at least 14% compared to methods that optimise placement or pricing alone. This study tackles an AI challenge in modelling infrastructure with multi-agent behaviour, using game theory and optimisation to simulate interactions and enable learning-based approaches in transportation systems.