A Scalable Distributed Algorithm for Optimizing Distribution Power-Grids

Abstract

Utilities are interested in obtaining optimal operational topologies for their distribution power-grids so that they can efficiently employ grid resources, and simultaneously, increase their profits while providing a specified quality-of-service (QoS) to their customers. This paper presents a distributed and scalable algorithm for yielding an optimal power-grid topology that minimizes the (QoS) metric called energy not supplied (ENS) by the distribution grid. The (ENS) is minimized by constructing both local, independent minimum weight spanning trees (MWSTs) and a global yet unique (MWST) for the entire grid. At each step of the distributed algorithm, agents exchange information about their local (MWST) in order to coordinately construct the global (MWST). In addition, the distributed algorithm exploits heuristics to supply extra decision criteria to the agents constructing the (MWST) of the power grid.