Geodemographic aware electric vehicle charging location planning for equitable placement using Graph Neural Networks: Case study of Scotland metropolitan areas

The widespread adoption of electric vehicles (EVs) is crucial for decarbonizing transport, but charging infrastructure development lags behind, creating a bottleneck. Current EV charging station (EVCS) distribution favors affluent areas, potentially reinforcing inequalities. We address this using a spatially-aware Graph Neural Network (GNN) model that learns urban dynamics and socio-economic factors for equitable EVCS placement. Our methodology analyzes charging patterns across residential, working/industrial, and commercial zones by integrating EVCS utilization, traffic patterns, urban structure, parking availability, and deprivation indices. Our analysis revealed EVCS infrastructure concentration in commercial zones, with less deployment in working/industrial areas and significant gaps in residential zones. Glasgow showed higher utilization rates, particularly in residential areas, while Edinburgh demonstrated utilization disparities in residential zones, with deprived areas showing lower usage despite need. To solve this issue, GNN-leveraged recommendations were utilized for strategic charger deployment in underserved areas. The findings indicate that in residential areas, 22 kW chargers show substantial benefit to underserved communities, with higher output chargers becoming more effective only beyond 50 initial installations. Working areas show similar patterns, while commercial areas demonstrate lower improvement across all charger types, confirming infrastructure saturation. These findings provide policymakers a framework to prioritize EVCS deployment for reducing disparities and accelerating EV adoption. Overall, our results demonstrate the effectiveness of this approach in identifying potential locations for EVCS deployment, particularly in underserved communities.