Control and Optimization of Grid-Tied Photovoltaic Storage Systems Using Model Predictive Control

In this paper, we develop optimization and control methods for a grid-tied photovoltaic (PV) storage system. The storage component consists of two separate units, a large slower moving unit for energy shifting and arbitrage and a small rapid charging unit for smoothing. We use a Model Predictive Control (MPC) framework to allow the units to automatically and dynamically adapt to changes in PV output while responding to external system operator requests or price signals. At each time step, the system is modeled using convex objectives and constraints and solved to obtain a control schedule for the storage units across the MPC horizon. For each subsequent time step, the first step of the schedule is executed before repeating the optimization process to account for changes in the operating environment and predictions due to availability of additional information. We present simulation results that demonstrate the ability of this optimization framework to respond dynamically in real time to external price signals and provide increased system benefits including smoother power output while respecting and maintaining the functional requirements of the storage units and power converters.