A novel container-based approach for integrating solar forecast in real-time simulation and model predictive control

This paper presents an interdisciplinary, novel approach for incorporating day-ahead solar forecast obtained using numeric models into a real-time simulation framework for low-voltage microgrid analysis. The solar forecast data were integrated into the grid simulation at the information, communication, and function levels, utilising the data model and communication structure defined in the international standard IEC 61850. Given the forecast of solar power and a reference trajectory defined by the upper-level grid management system over a sliding predictive time window, a model predictive control scheme has been implemented to compute control setpoints for solar curtailment in a simplified simulation scenario, regardless of the power flow over lines. In the virtualised environment with decentralised intelligent controllers in containers, the control setpoints are communicated between the IEC 61850 client and server and implemented as measures for solar peak-shaving. The test results showed that the model predictive control scheme outperformed trivial linear control methods, suggesting that the container-based virtualisation concept has the potential to be further exploited for practical use cases at the utility and energy market level, and the IEC 61850 standard could be a feasible solution in terms of grid communication and forecast integration.