Electric vehicle charging point placement optimisation by exploiting spatial statistics and maximal coverage location models

Abstract

Electric vehicles (EVs) are increasingly considered as a promising solution to tackle climate change impacts, improve air quality, and enhance growth sustainability. This paper proposes a two-step approach for optimally deploying charging points (CPs) by bringing together spatial statistics and maximal coverage location models. CP locations are conceptualised as a spatial point pattern, driven by an underlying stochastic process, and are investigated by using a Bayesian spatial log-Gaussian Cox process model. The spatial distribution of charging demand is approximated by the predicted process intensity surface of CP locations, upon which a maximal coverage location model is formulated and solved to identify optimal CP locations. Drawing upon the large-scale urban point of interest (POI) data and other data sources, the developed method is demonstrated by exploring the deployment of CPs in London. The results show that EV charging demand is statistically significantly associated with workplace population density, travel flows, and densities of three POI categories (transport, retail and commercial). The robustness of model estimation results is assessed by running spatial point process models with a series of random subsets of the full data. Results from a policy scenario analysis suggest that with increasing numbers of charging stations to be planned, optimal CP locations gradually expand to the suburban areas of London and the marginal gains in charging demand covered decrease rapidly.