In an electricity system such as today’s, characterized by an increasing presence of non-programmable renewable generation, there is a strong demand for new services to guarantee the quality and safety of grid operation.
In this context, the objective of this report is to verify the actual implementation potential of control algorithms based on Model Predictive Control (MPC) for the creation of a control service operated by electrochemical storage systems (batteries) in stand-alone mode or coupled to a wind farm. Specifically, the “ultra-fast” frequency regulation service called Fast Reserve, defined by the Italian Transmission System Operator (Terna), is taken into account. The algorithms developed and tested last year using numerical simulation are able to manage the devices that make up the controlled unit, taking into account their dynamic state (state of charge, wind speed) and technical characteristics, in order to coordinate their contribution to the provision of the service.
The activity described in this report therefore involved the implementation of these algorithms on a hardware card, which was then put into communication with an electrical system simulated in real time on the OPAL-RT® platform. This type of simulation, known as Control-Hardware-in-The-Loop (CHIL), allowed us to verify that the algorithms developed, if properly implemented, would be able to perform the expected computations in the time required to create the Fast Reserve service in accordance with Terna’s requirements.
Further validation was then carried out by including in the simulation a real battery installed in a remote test facility at the University of Genoa. The real power exchange of the battery was fed back to the simulated electrical system in real time to analyze the performance of the whole system, integrating the electrical system with the ICT (Information and Communication Technology) control system. The results of this second type of simulation, commonly known as Power-Control-Hardware-in-The-Loop (P-CHIL), showed that the behavior of the real/simulated hybrid system was comparable to the simulated one, confirming the potential of the proposed approach.
ABSTRACT