How Model Predictive Control of PV + Energy Storage Systems Achieves Higher Returns

Rule-based microgrid control has been the typical approach to operating microgrid systems with multiple technologies such as PV + Energy Storage. A recent test however, revealed that a different approach achieved much higher returns. The California Energy Commission-funded CalTestBed tested Xendee’s Model Predictive Control (MPC) based microgrid controller at one of the microgrid sites of the University of California San Diego (UC San Diego). The microgrid site is grid-connected with solar PV and battery energy storage systems. These technologies are connected to a data collection system, which provides real-time measurements to the MPC, receives setpoints from the MPC, and balances the energy system in real-time. The MPC considers adaptive demand charge correction that calculates the utility mismatch due to forecast uncertainties in previous timesteps and changes certain optimization parameters and constraints to mitigate uncertainties. The MPC testing was carried out for multiple months and the MPC performance was tested as compared to a standard rule-based Microgrid controller. The results from the testing indicate that the MPC reduced the total costs by up to 79% as compared to the rule-based approach.