Today I attended a half day event at the Technical University of Denmark called “PowerEvent”. It featured several European speakers working on issues related to the future electricity production and distribution grid, which by many is called the “Smart Grid”. Google the term, and you will find it is widely used, although it seamingly remains to be established what is ment with smart.
The first presenter was professor Jacob Østergaard from DTU Elektro, who described a simulator used at DTU PowerLab to simulate the continuous pricing structure needed to integrate renewable energy producers into the electricitet distribution grid. This simulator will also be part of the EU demonstration project on the Danish island of Bornholm. The second presenter was professor Stephen McArthur from University of Strathclyde in Scotland, who talked about the use of agents performing different tasks in electricitet production and distribution. There were different types of agents using known technology, e.g. constraint programming, to solve real problems in the high voltage distribution network. The third presenter was professor Zita Vale, from Polytechnic Institute of Porto, who talked about multiagent approach to electricity market siulation considering virtual power players (MASCEM).
It appears to me that a common underlying assumption about the smart grid, is that the players - also called agents - work for the good of the system. We know from the stock market, that traders not always work for the good of the market. When the fine balance between reasonalbe and unreasonable behavior in the stock market is disturbed we get stock market crashes, from which recovery often is long and slow. The agents of the smart grid can be viewed as the traders in the stock market. However, with one major difference. In the smart grid one company may control several agents, i.e. several electricity producers - even located in different countries. This would allow the company to manipulate the market in the short term, by removing production from the grid, in order to raise prices, so their remaining production units become more profitable. From the stock market we know, that such behavior can make the whole market unstable, and from time to time indeed do. With smart grids using the ideas of independent agents – although several may be controlled by a single market player, i.e. company – a system is created that has surprising similarities to the stock market. Hence the likelyhood of power market crashes similar to stock market crashes will properly not be that rare. We know, that recovery from stock market crashes are slow, but what do we know about power market crashes? Or for that matter power market safety? i.e. stability? or reliability?
Ultimately the end user – consumer - is interested in just one things: a safe supply of electricity. For most users this would mean a highly availabilty quality supply. For electricity availability can be measured by supplied voltage (i.e. if supply fails voltages drops to zero), and quality can be measured by supplied frequency.
Can a system based on game theory with independent agents with possibly conflicting objectives be designed to provide a safe supply of a ressource extremely important for modern society? Or can games be safe?