Model predictive control of vehicle charging stations in grid-connected microgrids: An implementation study

The transition to renewable energy sources, particularly sources like wind and solar induces a dependency on weather in the supply side of electrical grids. At the same time, the move to electric mobility with uncontrolled charging induces extra peak loads on these grids. These developments can cause grid congestion or an imbalance between the renewable power supply and the demand. Locally balancing the power supply and demand in grid-connected microgrids can alleviate such issues on the main grid. This paper presents a model based control strategy to address the challenge of locally balancing the power supply and demand in a grid-connected microgrid to avoid reaching the threshold rated power output set for large buildings. The microgrid under consideration consists of photovoltaic power sources and a large fleet of electric vehicle chargers (>150). A model predictive controller is developed that treats the daily vehicle charging as a batch process. Given vehicle charge objectives, the controller utilizes vehicle charger occupancy and photovoltaic power generation forecasting services to distribute power optimally over a fixed period of time. The optimization problem is formulated as a quadratic programming problem and is implemented utilizing open-source Python libraries. The controller was integrated into the control system of a microgrid situated at a corporate office in the Netherlands. The control system oversaw the operation of 174 vehicle chargers. The effectiveness of the model predictive control technology was demonstrated over a three-week period and led to an average daily grid peak power reduction of 59%.