Design of energy markets
Research manager: Hélène LE CADRE
The work carried out in 2013 on the design of energy markets can be broken down according to three mainlines: first of all, giving incentives (economic) to guarantee the optimum expansion of the capacities of the electricity network; then, modelling renewable energies (RE) and quantifying their impact under the assumption of a coupled European market by prices; finally, developing Machine Learning methods to add “intelligence” to the market.
Mainline 1: The market does not seem to sufficiently recompense investments in peaking capacities, i.e., peak generation and voluntary consumption interruptions also known as eliminations (the term commonly used is “Demand Response”). As the market only pays for the energy produced, the investment in production capacities is insufficient, which can lead in the event of a lack of capacity to the crumbling of the electricity network, intolerable for consumers and generating an exorbitant cost for suppliers.
The principle of a capacity market is to provide assurance that the electricity system has sufficient capacity to cover consumers’ needs at peak periods, either by forcing suppliers to provide proof that they have enough capacity or by developing an additional mechanism requiring the network manager to intervene in a way to guarantee the balance between supply and demand. An initial approach offering the design of an “effective” mechanism that penalizes withholding and sub-investments in capacities was proposed and presented during the EURO-INFORMS 2013 conference.
Mainline 2: By putting ourselves under the assumption that markets are coupled, we analytically studied the impact of the concentration of wind farms and the uncertainty resulting from the introduction of ER on the total supply cost, on market well-being and on their penetration in relation to conventional energies. The markets have incomplete information regarding the quantities of ER produced by other markets, mainly due to the considerable variability in these production sources. We prove that distributors have incentives to buy information about the production variability of other markets. Assuming that this information can be certified and sold by an external certification operator, we calculate the optimal certification price depending on the level of expected confidence. This research enabled a research internship to be created in 2013 for an ENSAE ParisTech student.
Taking into account interconnection capacities in the model that limit energy exchanges between areas of the supergrid will be tackled in 2014 as well as the optimum deployment of wind farms at supergrid level.
Mainline 3: Finally, we took an interest in “intelligent” market design by taking into account the ability agents have for learning of information concealed by other operators by observing their actions/the overall behavior of the energy system. This research calls on distributed Machine Learning techniques and regret methods more specifically. It was presented during the Finance and Market chair seminar backed by EDF and the Paris Dauphine University and was published in a trade journal.
