Evaluation of Predictive Based Electric Vehicle’s Charge Scheduling Algorithms in Self-consumption Scenarios

As the number of Plug-In Electric Vehicles is estimated to continuously rise over the next years, existing electrical grids have to prepare to accommodate a high number simultaneous charging of such vehicles. A solution to this challenge may stand in the ongoing integration of distributed renewable energy sources in the electrical grid, which is being encouraged by the legislation that, in several countries, already allow the production for self-consumption. However, due to their variability, renewable energy sources are frequently characterized as intermittent, causing mismatches in the required equilibrium between production and demand. Demand-response measures are in these cases seen as a solution. In this case it imposes the scheduling of the Electric Vehicle’s charging periods according with the forecasted generation levels, but not forgetting the day ahead tariff rates and the driver’s requirements. In medium to large charging stations such scheduling algorithms and models should allow a reduction of costs and the maximization of investments made in renewable energy sources. In this scenario this paper extends the models and analysis made in a previous study, comparing several predictive based Electric Vehicles charging models, including Adapted Earliest Departure First (AEDF) and Linear Programming (LP) algorithms by: (1) considering different EVs with several charging powers, (2) using real data regarding produced energy and day ahead energy price information in different periods of the year and (3) considering a self-consumption scenario where charging facilities can buy and sell energy from the utility. The results show that the eco-friendly Adapted Earliest Departure First (AEDF) model is able of achieving significant cost reductions with significant low computational complexity.