Probabilistic Modelling of Electric Vehicle Charging Demand based on Charging Station Data
Kazi N. Hasan, Robin Preece, Jovica V. Milanović
Abstract
The number of electric vehicles is expected to increase significantly, which will affect the operation of future power systems. Yet the availability of EV charging station data is very low, and hence, it is challenging to perform the EV impact studies with real (or randomly generated) EV data that accurately represents the spatio-temporal uncertainty of the electric vehicle (EV) charging demand. This study analyzes electric vehicle (EV) charging station data to derive appropriate probability distributions for EV charging stations so that EV data can be randomly generated to perform probabilistic EV impact studies. This study has collected aggregated residential, commercial and public EV charging station data for one-year duration. Seven probability distributions have been tested to identify the appropriate probability distributions for aggregated residential, slow commercial and fast public charging stations. It is observed from the simulation results that aggregated residential, slow commercial, and fast public EV charging station data follow extreme value, exponential and Gaussian probability distributions, respectively. Further, power flow simulations have been performed with EV data generated using a variety of probability distributions and voltage violation scenarios have been recorded in a test distribution network. The number of voltage violations again verified the above selection of probability distributions for aggregated residential, slow commercial and fast public charging stations.