Distributed Lagrangian Method for Tie-Line Scheduling in Power Grids under Uncertainty

Abstract

System operators (SOs) manage the grid and its assets in different parts (areas) of an interconnected power network. One would ideally seek to co-optimize the grid assets across multiple areas by solving a centralized optimization problem. Gathering the dispatch cost structures and the network constraints from all areas for a centralized solution remains difficult due to technical, historical, and sometimes legal barriers. Motivated by the need for a distributed solution architecture for multi-area power systems, we propose a distributed Lagrangian algorithm in this paper.We establish convergence rates for our algorithm that solves the deterministic tie-line scheduling problem as well as its robust variant (with policy space approximations). Our algorithm does not need any form of central coordination. We illustrate its efficacy on IEEE test systems.