A distributed, real-time and non-parametric approach to demand response in the smart grid

This paper considers a novel approach to demand response in a smart grid, wherein a set of flexible loads with given energy requirements and deadlines for completion adjust their instantaneous power consumption in such a way as to make the aggregate total power consumption on the grid as smooth as possible. Each load periodically adjusts its own power level using only a minimal amount of feedback i.e. the aggregate power consumption observed in the previous period, and knowledge of its own energy requirement and deadline. Our approach allows for a fully distributed, real-time implementation and does not require any knowledge of arrival process statistics or forecasts of future load. We illustrate the working of our algorithm with a detailed simulation of generation and load for a fictional small city modeled after the city of Ames, IA using data from a variety of sources. In this simulation, charging electric vehicles serve as flexible loads. We also present an approximate analysis of the bahaviour of flexible loads in our scheme based on a continuous time version of the proposed algorithm to gain insight into its properties and limitations.