Design and operation of decentralized electricity markets

Abstract

The energy system is evolving from a hierarchical and centralized structure to amore decentralized setup, with distributed renewable generation and increasing consumer engagement driving this paradigmshift. Current practice prevents end-users to actively participate in the electricity market. Even with lower barriers to entry, centralized electricity markets do not allow agents to trade energy between each other and according to their preferences. To address these challenges, recent literature has investigated future electricity markets based on sharing economy principles. The state-of-the-art entails vague regulation and few attempts of market design, often tailored to the specific application investigated. A more thorough design of decentralized electricity markets is, therefore, needed to feed back to the regulatory entities as well as to favour real-world implementation. This PhD thesis addresses the challenge of a comprehensive design of such markets, by analyzing their actors, architecture and negotiation mechanism, while assessing their guaranteed properties, as well as their operation in real-world conditions. The proposed market design provides the definition and modelling of market actors, including prosumers and system operators. A methodology for expressing prosumer costand utility functions is introduced, even for uncertain and inflexible assets, with the possibility of accounting for their risk attitudes. System operators are considered to enforce grid constraints and to account for power losses, adding dynamic network charges on energy prices. The designed market generalizes the way actors are interconnected and negotiate their energy trades while expressing their preferences over different trading partners. This market formulation unifies community-based and peer-to-peer layouts, as well as any hybrid combination and pool markets into a general consensus problem. Negotiation mechanisms based on decomposition techniques for large-scale optimization problems are employed to clear the market in a decentralized fashion. Core and desirable market properties are assessed, with particular focus on fairness of market outcomes on a social, individual and group perspective. This research shows how fairness can be enforced by introducing additional market products and by extending the actor objectives with carefully designed policies. Specifically, financial products are included to hedge the risk of market actors, enhancing fairness under heterogeneous risk attitudes, while loss allocation policies are investigated to limit the geographical discrimination of grid tariffs among agents. Finally, the complexity of the proposed negotiation mechanism is tested in real-world conditions, with large numbers of market participants and imperfect communication. The resulting analysis shows that the time needed to attain market outcomes increases to the point of undermining the actual implementation of such decentralized electricity markets. The negotiation process is adapted to online matching algorithms, similarly to the stock exchange, in order to cope with numerous agents and delayed communication, but at the expense of lowering market efficiency. This thesis concludes with insights for a comprehensive market design accounting for the limitations of real-world applications.