Learning Price-Elasticity of Smart Consumers in Power Distribution Systems

TL;DR

This paper introduces a novel open-loop, low-communication method using sparse linear regression to accurately estimate the price elasticity of individual consumers in power distribution systems, enhancing demand response strategies.

Contribution

It presents a new approach for estimating consumer price elasticity with minimal communication, leveraging sparse regression and inhomogeneous price signals for reliable, scalable demand response management.

Findings

Reliable elasticity estimation is achievable with sufficient communication slots.

Estimation accuracy depends on sparsity level and signal-to-noise ratio.

The method is fair and scalable for large distribution systems.

Abstract

Demand Response is an emerging technology which will transform the power grid of tomorrow. It is revolutionary, not only because it will enable peak load shaving and will add resources to manage large distribution systems, but mainly because it will tap into an almost unexplored and extremely powerful pool of resources comprised of many small individual consumers on distribution grids. However, to utilize these resources effectively, the methods used to engage these resources must yield accurate and reliable control. A diversity of methods have been proposed to engage these new resources. As opposed to direct load control, many methods rely on consumers and/or loads responding to exogenous signals, typically in the form of energy pricing, originating from the utility or system operator. Here, we propose an open loop communication-lite method for estimating the price elasticity of many customers comprising a distribution system. We utilize a sparse linear regression method that relies on operator-controlled, inhomogeneous minor price variations, which will be fair to all the consumers. Our numerical experiments show that reliable estimation of individual and thus aggregated instantaneous elasticities is possible. We describe the limits of the reliable reconstruction as functions of the three key parameters of the system: (i) ratio of the number of communication slots (time units) per number of engaged consumers; (ii) level of sparsity (in consumer response); and (iii) signal-to-noise ratio.