by Colin Bankier, School of Information Technology and Electrical Engineering, University of Queensland, Australia.
Abstract: The electricity industry in many countries is facing new pressures due to escalating demand and environmental concerns such as climate change. The application of information technology to power systems in order to make electricity grids intelligent is gaining momentum as a means to achieve electricity networks which are more efficient, can accommodate greater amounts of distributed and renewable energy resources and provide demand management. This document presents a project proposal for the development of an open source multi-agent software toolkit for intelligent power distribution to support learning, research and collaboration in this area. An introduction to the topic of intelligent power grids and an overview of the purpose and aims of the project are presented.
5 comments
Comments feed for this article
April 22, 2010 at 9:01 pm
Ian Sommerville
This is, in principle, a nice idea but how on earth can you evaluate it for real. I doubt if we can really simulate the power grid so the only way to do it is to run it alongside existing systems and see what differences arise. But how would you get the buy-in from the power companies to achieve this.
And I do wonder about open-source for very specialised systems – there really aren’t that many people who know much about power management.
April 23, 2010 at 5:01 pm
Sophia Drossopoulou
I thought that current electricity grids could code well with local creation of electricity – otherwise how do the current systems work which incorporate wind turbines and solar power electricity production by private individuals? However, I am probably wrong there, since ref [1] is supposed to be talking about that.
More importantly, I am not sure what exactly will be the remit of the agents, is it so that the grid can manage consumption and production (that is what I thought in the beginning) or that students would understand what such a grid looks like. Why it is important that the agents should be open source?
April 26, 2010 at 1:48 pm
Carla
The idea of using multiagent system to control power distribution is not novel. See for example the paper written by Nicholas Jennings and Stefan Bussmann, entitled “Agent-Based Control Systems”, not refered in the paper.
The paper describes a project involving the development of a open source toolkit for developing MAS specific for power systems.
The paper states that the benefits of using GridIQ toolkit would be facilitating learning about MAS in power systems, but what are the impacts of this for increasing energy efficiency, energy economy and use of renewable energy resources?
The paper fails in answering this questions and, for this reason, I don’t recommend discussing about this paper at the workshop.
May 2, 2010 at 2:49 pm
Sebastian Gonzalez
It would be interesting to have a justification or motivation of why software agents are a convenient technique to study the problem of power distribution. The contrast between traditional centralised power distribution systems and more grid distribution systems made me think of an analogy with centralised client-server networks vs peer-to-peer (p2p) networks. It occurs to me that some techniques from p2p networks for self-adaptation, self-healing and self-optimisation could be applied to grids, since both systems are about interconnected nodes that exchange either data or electricity, where the network’s topology changes dynamically. So I wondered, is there anything that would make agents more interesting than p2p networks to study power distribution systems?
May 4, 2010 at 1:11 am
Colin
Hi,
Thank you all for your comments. Many valid points have been made or identified areas I need to explain further. In general the idea of a smarter power grid is not new, and many research and commercial projects are investigating this area. For a large US example see http://www.gridwise.org. The use of agent-based software has been identified by previous researchers as just one suitable paradigm for implementing intelligent systems for this purpose.
I’ll try briefly respond to some of the questions:
Ian, simulation of power grids is common and software exists for this purpose. Of course simulation is only of limited value, and small pilot projects are used to implement new types of systems in a real environment and gradually rolled out to bigger systems. For example software provided by http://www.infotility.com/ started as a research project here in Australia, and is being deployed by New York utilities.
Sophia, the remit of the agents is to achieve any of the goals of a smart grid (see gridwise site). It is open source to allow collaborative development from researchers. No freely available software currently exists for this purpose. The current scope means that it is aimed at students and could not hope to compete with products offered by large vendors, but hopes to provide a platform for students to gain knowledge and experience in this area by using the tool and contributing to its development.
Carla, the impact smarter grids (possibly achieved using agents) can have on efficiency and renewable sources is discussed in other research, but the gridwise site above gives a good overview, eg at http://www.gridwise.org/smartgrid_energyeffi.asp.
Sebastian, You have a good point about the similarities – the agents may implement protocols/algorithms that are in some ways similar to networking protocols/algorithms. The agents however aim to collaborate to address any number of competing goals within a complex system, and how this can be applied to power systems is the focus of much current research.
Thanks for taking the time to add your comments, I appreciate any feedback that might guide the project to becoming valuable to anybody in this field.
Regards,
Colin.