Distributed control for the smart grid: The case of economic dispatch

Abstract

We describe and summarize our recent results on designing distributed control algorithms for the smart grid that are designed to interact intelligently with traditional control mechanisms in the existing electric grid. We illustrate our approach using our recent work on a novel approach to the classical problem of economic dispatch i.e. the problem of allocating power generation among multiple generating units in such a way as to minimize the total cost of generation while satisfying a set of constraints. Our approach is based on a simple distributed algorithm in which each generator independently adjusts its power-frequency set-points to correct for generation and load fluctuations using only the aggregate power imbalance in the network. We show that the algorithm eventually achieves the condition of optimal economic dispatch under mild assumptions. Our algorithm exploits the power-frequency droop curve, which is a mechanism developed over many decades by power systems engineers to ensure stability in the presence of load fluctuations. Specifically the droop curve allows each generator to measure the aggregate power imbalance in the network in a distributed fashion simply by measuring frequency deviations on the grid. While our approach achieves optimal dispatch using only the implicit communication signal from the droop curve, without requiring any explicit communication whatsoever between the generating units, our recent work also shows that additional communication links can be exploited to improve the rate of convergence to optimality.