Optimal real-time control of multiple battery sets for power system applications

Due to high power in-feed from photovoltaics, it is expected that more battery systems will be installed in the distribution grid in near future. In this paper, we present a method to optimally control multiple battery sets in real-time to collectively provide services to power systems. The method is based on a Model Predictive Control (MPC) framework that tries to track an aggregated signal, while minimizing i) operation in high/low state of charge regimes, which would lead to battery degradation; ii) overall power losses including battery system, and iii) network losses, and considering network constraints. The performance of our approach is demonstrated for an application that coordinates several battery sets operating within a network containing photovoltaics and loads. Since our MPC controller can deal with non-convex piece-wise linear functions, the system losses can be modeled in more detail than in convex formulations. We show that the total battery system losses can be reduced by 35% with our approach as compared to approaches that operate all batteries in the same way.