Real-Time Reconfiguration For Large-Scale Distribution Networks: A Multi-Opening Linear Approach

Distribution network reconfiguration (DNR) is a core operational tool in modern distribution grids, enabling fast, topology-based control to reduce losses, improve voltage profiles, and alleviate overloads in the presence of variability introduced by distributed energy resources. This paper proposes a DNR solution based on a new linear model of branch currents when a set of lines is opened simultaneously, integrated into a graphtheoretic framework. With this model, network losses can be estimated for any set of simultaneous line openings, avoiding the need to solve a power-flow for each tree under analysis. The method proceeds in two phases, both leveraging the same linear model: the first obtains a minimum spanning tree from singleopening estimates on the meshed network, thereby obtaining an initial set of candidate branches to open; the second refines that set by evaluating alternative multiple-opening branch sets with the same linear model. With the proposed methodology, only one power-flow solve of the initial meshed network is required. Tests carried out on networks of up to 10 000 nodes show significant reductions in losses and very low execution times, supporting realtime applicability at a scale at which most existing approaches have not been demonstrated.