On adaptive MPC development and application to a DC fast charger (DCFC) station with an embedded microgrid

This paper proposes a parameter-adaptive model predictive control (MPC) for a high-utilization DC fast-charger (DCFC) station with an embedded solar-PV/battery microgrid, treating the stationary battery energy storage (BES) charging and discharging coefficients as uncertain parameters estimated online. The control problem is formulated as a multi-objective quadratic program to minimize grid-import and BES-degradation costs while maximizing self-consumption of on-site generated solar-PV power. The application of the proposed model to a practical case study demonstrates strong capabilities of the developed MPC, reducing overall annual energy costs by 71 % and grid peak demand charges by up to 52 %, a superior performance when compared to 65 % grid energy cost and 23 % peak demand cost savings achieved by the open-loop optimal control (OLOC) strategy formulated for the same case study problem. The analysis of the obtained results over different temporal scales–daily and weekly–across different seasons of the year shows the MPC’s capabilities to aggressively handle the model objectives and constraints amid uncertainties in solar PV generation and the charging station’s demand to realize lower overall system costs, thanks to its feedback control capabilities. These findings underscore the MPC’s robustness and practicality for real-world renewable energy-based DCFC applications. The obtained results can be used as a framework of reference when planning integration of renewable-based microgrids in electric vehicle (EV) charging infrastructure.