Power grid segmentation for local topological controllers

Abstract

The massive integration of renewable energies into the power grid leads to more complex flow management that requires faster reactivity. Because they can react quickly and allow optimization parallelization, decentralized management with local closed-loop controllers is a promising solution for real-time constraints management due to rapid power flow variation. However, determining the area that a local closed-loop controller should control is decisive for their proper functioning. The controlled area must allow the appropriate remedial actions and not create constraints outside of it, especially for a topological agent. This paper proposes a method to identify these areas for a topological agent. Thus, we first describe the topological agent implemented to evaluate the method. Then we describe the three steps of the method pipeline: the construction of an influence graph adapted to topological actions, the clustering of this influence graph, the evaluation of the capacity to host topological local-closed-loop controller of the zones resulting from the segmentation. Finally, we conducted simulations of local action controllers acting on areas identified by the method to test the procedure’s robustness. We eventually assess them and compare them to a centralized optimization scheme.