Integrated planning of public charging stations and low-voltage network expansion: From uncertainty to action

Electric vehicle policy targets and adoption forecasts are evolving rapidly. While the electrification of road transport is mandated by policy, the pace of adoption and technological evolution remains uncertain. This uncertainty complicates planning for public charging infrastructure in urban environments. Plans must ensure that the low-voltage grid can safely accommodate the additional electricity demand. Decisions about where to deploy public chargers and which types to install must be feasible at the time of installation and remain robust as policies and demand evolve. Currently, public charging station siting and low-voltage grid expansion are typically decided independently by different stakeholders whose interests overlap but who do not necessarily interact. In this paper, we investigate these problems in an integrated manner to avoid grid-blind placement and infeasible plans in practice. Conversely, grid expansion decisions that ignore rising public charging demand can result in costly and poorly targeted reinforcements. Our objective is to provide a strategic planning methodology that helps the public sector and local distribution system operators make informed decisions despite inherent uncertainty about the future. The planning methodology: (i) explicitly models uncertainty in spatial charging demand and baseline household loads rather than relying on a single forecast, (ii) quantifies the trade-off between network investment cost and grid impact, (iii) pinpoints substations and locations that are persistently problematic across plausible futures, and (iv) introduces a Benders decomposition-based solution approach to solve the resulting large-scale stochastic optimization problem on realistic grid networks. The methodology thus enables the public sector and distribution system operators to prioritize where exactly to make early investments to proactively alleviate bottlenecks. Although public low-voltage grid data are not always available, we demonstrate the effectiveness of the planning methodology with realistic, albeit unavoidably artificial, low-voltage networks and demand scenarios.