A distributed electric vehicle charging management algorithm using only local measurements

With the uptake of electric vehicles (EVs) promoted by many governments, the impact of electric vehicles on electricity grids will become significant in the near future. In Australia, charging a typical EV battery puts the same demand per day on the grid as an average household, which could lead to a sizeable increase in peak demand. However, the negative impacts of EVs can be mitigated if their charging is scheduled during times of otherwise low demand, such as overnight. The majority of studies trying to achieve this require a certain level of coordination among EVs and/or a central controller. In many countries, however, the hardware and infrastructure required for central charging methods do not exist. Here EV charging is approached from a distributed point of view, and a protocol in which charging decisions are made individually at each household, without any access to full network state is proposed. The decision making process is conducted in real time, using both instantaneous and historical local voltage measurements to estimate present network load. The overall goal is to maximally use grid capacity at all times, while still ensuring fairness of charging for all users. The proposed algorithm ensures both charging efficiency and fairness among all EVs across the network. At the same time, peak demand in the grid is minimally affected. Simulations based on a realistic suburban network using real demand data and vehicle travel profiles is presented to illustrate typical performance.