Please use this identifier to cite or link to this item:
|Title:||Demand Side Management in smart electricity networks: Algorithmic, Economic & Game-Theoretic aspects of active user participation|
|Keywords:||smart grid, demand response, game theory, mechanism design|
|Abstract:||Modern energy policies drive the electricity market towards a liberalized framework. As a result, concepts from other commodity markets are becoming increasingly relevant in the context of the electricity market. However, there are certain specialties that characterize electricity. Such a specialty is the requirement of constant balance between supply and demand; otherwise the stability of the underlying physical grid is compromised. The traditional approach has been to only control the supply, so that it follows the demand at all times. However, high penetration of non-dispatchable renewable energy sources and load electrification (e.g. electric vehicles) have highlighted the need to also utilize the elasticity that there is at the demand side, by applying Demand Side Management (DSM). The main objective of DSM is to achieve an aggregated consumption pattern that is efficient in terms of energy cost reduction, welfare maximization and/or satisfaction of network constraints. This is generally envisaged by encouraging electricity use at low-peak times. In this dissertation, we model a set of smart devices at the side of residential electricity consumers and a home energy management system that is able to make decisions about home electricity consumption by taking into account the user’s preferences, the dynamic electricity pricing signals as well as the operational constraints of devices. We envisage an electricity service provider that is responsible for incentivizing users to shape their consumption patterns in line with the needs of the electricity system. We study and develop techniques for two general use cases of DSM: online algorithms for real-time consumption curtailment and offline algorithms for day-ahead load scheduling. We considered an intelligent agent at the user’s home energy management system able to make strategic decisions. In this setting we formulated a game where each agent tries to optimize its own objective. We formulated the problem of designing online auction mechanisms that are able to bring the system to a Nash equilibrium. Also, the final allocation needs to exhibit attractive properties in terms of the key performance indicators set by the state-of-the-art literature. In order to achieve these goals we drew on concepts of algorithmic game theory and mechanism design. Specifically, for the real-time demand response case, we designed two online auction schemes for two specific business models. The first is based on Ausubel’s clinching auction and achieves the majority of the standard requirements of mechanism design theory. Namely the proposed scheme, achieves economic efficiency, incentive compatibility (in the sense of making it a dominant strategy for each user to act truthfully according to his/her preferences and leaving no room for cheating), scalability, privacy-preservation and individual rationality in contrast to studies in the current literature that achieve only a subset of the aforementioned properties. Furthermore, it is shown to maximize the service provider’s profits among all efficient allocations. The second business model refers to cases such as energy cooperatives where the issue of fairness of the allocation is important. We designed a novel mechanism that significantly improves fairness in comparison to the state-of-the-art. For the day-ahead load scheduling case, we designed and evaluated a novel DSM scheme that addresses several issues that were not jointly addressed before. Specifically, the proposed DSM scheme preserves the economic efficiency, individual rationality and budget-balance properties. It is also able to satisfy coupling, system-wide constraints. The proposed scheme is theoretically proven to always bring the system to the Nash equilibrium. Finally, we studied the problem of jointly considering a day-ahead load scheduling and a real-time DSM scheme that balances unexpected deviations from the agreed schedule. We proposed a differentiated pricing based on a spread, and studied its effect on the users’ strategies.|
|Appears in Collections:||Διδακτορικές Διατριβές - Ph.D. Theses|
Items in Artemis are protected by copyright, with all rights reserved, unless otherwise indicated.