Are you an EPFL student looking for a semester project?
Work with us on data science and visualisation projects, and deploy your project as an app on top of Graph Search.
Electric vehicles (EVs) are already part of today's reality and their number is expected to grow rapidly in the near future. A large-scale penetration of EVs will increase the power consumption during charging periods. Hence, the uncoordinated and random charging activities could stress the grid hence severely impact the quality and continuity of the power supply. In particular, excessive power flows can cause overloads of the grid, which can lead to severe grid damages and blackouts. Due to the aforementioned issues, there is a need for intelligent control methods of the electric-vehicle charging stations (CS). In this respect, we focus on the problem of optimal grid-aware real-time charging/discharging control of EV CSs. First, we propose a grid-aware real-time control method of EVs charging. The purpose of the proposed method is to provide flexibility for EVs connected to CSs for the grid by following an aggregated power-setpoint from a main grid controller while minimizing the EVs' battery-wear and keeping the charging balance between EVs. The aggregated power-setpoint might exhibit rapid variations due to other volatile resources of the local distribution grid. Naive allocations can transfer such variations to the EVs' batteries thus increasing EV battery-wear. One of the challenges is to allocate powers such that the fluctuation of the aggregated power-setpoint does not cause power jumps and mini-cycles of the charging power of the EVs. The method uses a realistic model for the battery charging-power and takes into account a non-ideal behaviour of EVs. For example, the charging power cannot be arbitrarily small. If not appropriately handled, the rapid fluctuations of the aggregated power-setpoint could lead to frequent disconnections and re-connections, which should be avoided. We also handle a non-ideal response of EVs to the control power-setpoints due to implementation and reaction delays and inaccuracies. Another challenge is that the power allocation should keep the charging balance among EVs hence the information about future arrivals and departures could be unavailable. To address these issues, we cast the problem as a repeated online optimization. This leads to a mixed-integer quadratic problem; to solve it in real-time, we develop a heuristic that reduces the number of integer variables. Then, we implement and validate the method in a real field, i.e., on real-scale microgrid with real commercial EVs. Second, we focus on vehicle-to-grid (V2G) technology. In order to minimize global operational costs, to take into account grid constraints, and to minimize EVs battery wear, we suggest combining an optimal scheduler that takes care of charging/discharging EVs and minimizing the global operational costs with a real-time controller that reacts to grid-aware external setpoints. The scheduler computes optimal powers for EVs; it considers forecasts of future arrivals, departures, and operations of other energy resources (i.e., other loads and PVs). The real-time controller, in turn, follows an aggregated power-setpoint from a main controller of the local distribution grid, thus minimizing the EV battery-wear while following the scheduler's decisions. We validate our method by simulations and compare it with benchmark real-time algorithms. We show that our method presents lower operational costs and EV battery-wear when compared with benchmark algorithms. Third, we focus on the side effects of CS integration into grid control...
Mario Paolone, Willem Lambrichts
Yuning Jiang, Wei Chen, Xin Liu, Ting Wang