Abstract
ECODREAM Energy Communities (ECs) are envisioned to be the future of smart energy grids as they integrate distributed energy resources efficiently, by promoting self-consumption of the end-user(s). ECs are the result of a recent and evolving regulation framework that allows the end-users (a minimum of two participants) to define economic contracts in order to share distributed energy resources. An EC is characterized by end-user(s) in a close geographical proximity that have an electrical and/or thermal demand, along with individual/shared renewable energy resources and storage technologies. ECs have the potential to offer i) environmental benefits in the form of reduced carbon footprint, ii) social benefit in the form of increased end-user participation, iii) economic benefits in the form of the revenue obtained due to the reduction in distribution and transmission cost and grid losses. In order to guarantee stable and reliable operations of individual ECs, and possibly their interaction with external entities (such as neighboring ECs or for-profit entities such as the aggregator), a synergic combination of control algorithms aided with information and communication technologies (ICT) is needed. The control algorithms need to comply with regulation policies, end-user(s) preferences, operational constraints of the EC and their interaction with external entities. Motivated by the above challenges, the focus of the project is threefold: (i) to formalize new models specific to the design and functioning of ECs; (ii) to design distributed algorithms and software toolboxes capable of generating local-level control policies with global optimality and safety guarantee, and (iii) to test the algorithms on a real test-bed and their scalability through extensive large-scale simulations. ECODREAM aims at enlarging each partner’s knowledge and expertise towards the other ones’, extending existing methods in the core interest of the partners, and at employing this consolidated knowledge to converge vertically on the applications. The project algorithms will be benchmarked, both in simulation and on the test-bed, with the state of the art in terms of computational and control performance also, and within the possible extent, over a range of key performance indicators (such as energy and environmental, economic, social performance, and so on). Risultati UNIBO’s activities within the ECODREAM project have focused on the development of distributed algorithms and toolboxes for learning and optimization with application to models of energy systems arising in energy communities. These models describe heterogeneous prosumers aiming at cooperatively optimize a common objective by relying on suitable local cost functions and, where available, on collected data. Building on this model framework, distributed optimization algorithms have been designed for classes of problems in which the local cost functions depend on an aggregate variable or are subject to a coupling constraint. In particular, distributed algorithms based on the Alternating Direction Method of Multipliers (ADMM), combined with dynamic tracking mechanisms, have been designed. Moreover, general classes of distributed algorithms have been proposed within a unified system-theoretic analysis framework. Thanks to this framework, also noncooperative scenarios have been considered in a distributed network system. To account for realistic characteristics of energy networks, the proposed methods have been extended to asynchronous and unreliable communication settings. Another important feature of energy networks is that part of the cost function may be accessible only through users’ feedback. To address this challenge, a combined distributed optimization and deep learning approach has been designed. Moreover, derivative-free algorithms have been proposed together with theoretical approaches to obtain sufficiently rich data. At a theoretical level, another key framework has been investigated in which a macroscopic model is learned and optimized through a distributed mechanism for the local nodes. This novel perspective represents a starting point for a more general optimization framework. As for the algorithmic toolboxes, software libraries have been developed on specific energy community application scenarios to test available algorithms as well as novel ones on synthetic data. Specifically, routines have been created for energy trading and collective optimization problems in a distributed framework of energy communities. Finally, in collaboration with the group of UNIGE, a distributed optimization mechanism for the realistic simulator developed by UNIGE has been designed. The proposed mechanism models a realistic microgrid and addresses the problem of computing an optimal daily schedule for generation plants and storage systems. The model includes microturbines, renewable energy sources, storage systems, and boilers that are interconnected with each other and with the main grid. This interconnection exists at both energy and communication levels, allowing the devices to exchange energy and information among themselves. This distributed realistic simulator represents a key starting point for future virtual and real experiments.
Project details
Unibo Team Leader: Giuseppe Notarstefano
Unibo involved Department/s:
Dipartimento di Ingegneria dell'Energia Elettrica e dell'Informazione "Guglielmo Marconi"
Coordinator:
Università del SANNIO di BENEVENTO(Italy)
Total Unibo Contribution: Euro (EUR) 63.856,00
Project Duration in months: 24
Start Date:
28/09/2023
End Date:
28/02/2026
