Note: This content is accessible to all versions of every browser. However, this browser does not seem to support current Web standards, preventing the display of our site's design details.


Algorithms for industrial and household demand side management in E-Price


R. Vujanic, S. Mariéthoz, M. Morari

European Energy Market Conference, Royal Institute of Technology, Stockholm, Sweden, 27 - 31 May

In this work we present efficient methods to automate the operation of demand side management. These algorithms have been developed within the framework of the E-Price project, and they address the control of both large and small electricity consumers.

The main idea behind demand side management is to manipulate (either via pricing signals, or via coercive controls) the current and planned consumption. This is beneficial for counteracting in real-time unexpected changes in system's conditions, for instance in injected power from wind power generation plants and thus for reducing imbalance energy and costs.

In the first part of this work, we focus on large electricity consumers. We describe how energy intensive tasks can be scheduled, and shifted, to provide flexibility as a (price sensitive) reserve service to the system operator. The quantities of power and energy considered in this context are relatively high, but the time needed for their activation is also high (typically at least 10 to 15 minutes). The methods presented are based on new discoveries on mixed integer optimization under uncertainty, in particular of an efficiently computable robust counterpart.

In the second part, we describe the methods developed for the aggregated control of a large number of household costumers. Their aggregated consumption can be organized to achieve load shifting on short term. They thus complement the slow reaction times of the reserves provided by large consumers, and they also represent an attractive way to "buy time" for slower power plants to react to the unexpected changes. However, the methods developed for this control task have to deal with two major difficulties: the size of the associated optimization problem, and the relatively short reaction times required (which impose hard bounds on the available computation times).

We present several algorithms that tackle these challenges by decomposing the very large optimization problem and distributing the computation over different units distributed over the network. One algorithm exploits the concept of consumers aggregation to split the optimization in simpler optimization problems. Another algorithm relegates the management of local decisions concerning single household appliances to a control algorithm that locally only need to perform simple arithmetic and logic operations. The algorithm coordinates the appliances using a small and convex optimization problem.

Several case studies illustrate the presented methods and demonstrate their effectiveness.


Type of Publication:


File Download:

Request a copy of this publication.
(Uses JavaScript)
% No recipe for automatically generating a BibTex entry for (06)Talk
Permanent link