A decentralized optimization approach for scalable agent-based energy dispatch and congestion management

Due to the increasing number of renewable energy sources and the growing number of energy consumers such as electric vehicles and heat pumps, distribution system operators face growing challenges in managing the grid. A promising direction for future grid management is the decentralization of computing power to individual agents within the distribution network. In the proposed model, each household is equipped with an agent responsible for optimizing energy flow and negotiating surplus energy with neighbouring agents. However, the delivery of energy between households could lead to grid congestion, because the current distribution grid structure is not designed to handle the higher energy consumption of the increasing number of electrical consumers. This paper presents an agent-based approach for efficient energy dispatch and grid congestion management. The developed system operates in a decentralized manner, without the need for a central coordination unit. To incentivize the participation of prosumers in such decentralized approaches, a market-based energy dispatch is implemented using a game-theoretic method. In addition, to unbundle market and control related activities, the operation of grid congestion management is separated from market-based energy dispatch and implemented using the Alternating Method of Multipliers (ADMM). The feasibility and effectiveness of this approach is demonstrated through simulations involving 5 prosumers, as well as simulations on IEEE 33 and IEEE 119 bus networks, showing its potential to address the complexities of current and future grid management challenges.