REDeFINE REflex-based Distributed Frequency control for power Networks
REDeFINE is a DFG-funded research project with the goal of developing distributed frequency control for future power grids. Within the project, especially approaches of model predictive control and the interaction with grid-forming converters are investigated. Communication systems play an important role and the influence of communication delays as well as data loss will be analyzed. The project is conducted in cooperation with the communication chair COMSYS of RWTH and with the partner project aDaptioN in the USA.
Contact
Electric power grids are experiencing a dramatic change and will experience an even faster change in the upcoming decades, caused by a transition from a low number of big, inert power plants to a large number of small and dynamic power generators. Virtually all other critical infrastructure relies on a stable power grid. It is hence of utmost importance to sustain this stability and achieve the expected quality of frequency control. This is currently realized by a hierarchical approach with local primary control (independently of other power generators) and secondary control in a central control room. This approach, however, relies on the availability of large mechanical inertia (which is only provided in big thermal power plants) and does not scale arbitrarily when the number of power generators is increased. In this project we propose a flat, but distributed (networked) frequency control where every power generator is part of a distributed control problem. A critical challenge is then given by the effective measurement of the system level frequency with respect to the bus level frequency as available to every power generator. This can be approached by the recently developed concept of a Frequency Divider. We expect, however, that this can not be realized entirely distributedly without aid of the communication system due to large and uncertain delays.
In this project we hence devise a joint communication and control solution where parts of the control problem will be solved inside the communication system and information that is typically only available to the communication system will be used to improve the control process. We plan to investigate the opportunities of in-network processing to offload frequency estimations to the network. In this realm,network devices will be made capable of aggregating local views of connected power generators into a frequency prediction. Frequency predictions based upon the aggregate of local information can help increasing the scalability of the control process, but are subject to uncertainties. In particular, communication delays add uncertainty in the time domain. The joint communication and control approach addressed in this proposal, however, can benefit from knowledge that is typically only available to the communication system such as current buffer levels and link qualities. This information can be used to estimate delays and reduce uncertainty. Our research will be complemented by our partners in the U. S., who investigate orthogonal optimization techniques for distributed frequency and voltage control in the scope of aDaptioN. Eventually, we expect this approach to enable a stable frequency control for the future power grid
Acknowledgements
This project is funded by the DFG within the priority program SPP 1914 Cyber-Physical Networking under project number 432169785.