We share some methodological approaches (e.g., the distributed nature and automated negotiation approaches) with existing research. For example, Wu et al. [3] validate their pricing strategies via simulations, and Hayes et al. [4] propose a co-simulation including P2P energy platforms and energy distribution networks. Other approaches adopt statistical learning algorithms (e.g., reinforcement learning and Q-learning) and game theory, enabling agents to trade autonomously and derive long-term profit-making policies (e.g., see [5, 6]). Approaches using aggregators often try to bring together smaller entities (e.g., on the household level, and allow them to provide their services to the energy market levels [7]). Similarly, they valorize flexibility. However, they typically ignore the existing and, in practice, established form of balancing groups. Our approach relies on existing structures in the energy balancing and enables them to act more dynamically and integrate more partners. Decision-making approaches for balancing groups Centralized decision-making: the BG centralizes all the partners, enabling optimization. It requires data centralization (often impossible), and it is implemented either by the BG itself or by the given SBG(s) following the BG's indications to adjust their schedules. In turn, it distributes the financial efforts/benefits among the partners. The advantages of this approach are possible optimization, risk-sharing, and limited balancing concerns. However, technical and organizational challenges (i.e., implementation/scaling) carry disadvantages, including explicit management of trust, transparency, and privacy, and practical difficulties in data integration. Decentralized decision-making: Single actors (DSOs) make autonomous decisions and are responsible for their balancing (not viable w.r.t. the current regulative framework). Indeed, the concept of balancing groups has been introduced to avoid such a decentralized solution [2]. The (theoretically) full decentralization entails advantages including actors' autonomy and no need for data integration. Nevertheless, it is discouraged by significant disadvantages such as the absence of risk-sharing (everyone is exposed to the risk), every actor needs to invest in balancing efforts (more effort per partner), possible increase of the costs, and not technologically viable. Hybrid solution: It combines the previously mentioned approaches and valorizes optimization and risk-sharing, with only a limited amount of actors needing to set up resources for balancing. It would still maintain actors' autonomy, reduce the need for data integration, and reduce the sharing of sensitive information with competitors. Besides the hybrid approach, neither of the extreme approaches can be unconditionally recommended for future balancing solutions.
The GB-Flex SimulatorThe GB-Flex simulator enables the definition, simulation, and evaluation of several BGs/SBGs strategies leveraging both synthetic and real-world data 1 to mimic today's SBGs (including independent DSOs)...