Data-driven approach to model electrical vehicle charging profile for simulation of grid integration scenarios

Having the means to study the impact of electrical vehicle (EV) recharge on the power distribution network is one key aspect needed to manage the development of this technology. Power distribution grid and EVs are strongly connected elements that require to be wisely integrated to avoid that the limitations of the distribution network may hinder vehicle diffusion or that rapid growth of recharge requirements may put the distribution network in critical situations. In this study, a data-driven methodology is presented that aims at obtaining power requirement models that can be used to foresee the behaviour of the grid. The key to this methodology is the observation of charging profiles of a fleet of EVs over one year. The data collected defines a scenario representative of a generic fleet of commercial or sharing vehicles. The data is progressively loaded onto an existing database infrastructure and processed to obtain charge distributions that are then simulated in small sample networks in order to test the methodology. Starting from these data, a stochastic model is proposed to forecast the behaviour during the day and used to simulate by means of Monte Carlo techniques the impact on the power grid