Multi agent system framework for demand response management in distribution networks

Abstract

The inexorable increase in penetration of clean energies and responsive loads in the Distribution Network (DN), introduces new technical challenges for network operators. The responsibilities of Distribution Network Operators (DNOs) are being adjusted to cope with current system challenges and they are transitioning to Distribution System Operators (DSOs), taking a more active role in dynamically managing power flows across the network. Further, the advancement in distribution automation technologies provides greater opportunities for energy consumers to take more effective participation in demand reduction schemes and DSOs can be enablers of Demand Response (DR). Hence, the functionality of DR can be considered an alternative, lower cost, carbon-saving and flexible solution to defer network reinforcement. This forms the rationale behind this thesis, which aims to provide an in-depth investigation of the potential responsiveness in residential demand and its effect on constraint management of the DN. The main contribution of this thesis is the design, development and implementation of a Multi Agent System (MAS) framework for active DN management through residential DR. One advantage of the proposed platform is the capability of integrating both centralised and decentralised DR control mechanisms. It employs the responsiveness of demand from both loads shifting and shedding through price-based and incentive-based DR respectively. The feasibility and effectiveness of such a platform has been evaluated by developing and implementing the DR mechanism in three levels. The DR algorithm for several static and dynamic electricity tariffs, (Time of Use (ToU), Day-Ahead (DA) and Real Time Pricing (RTP)), is designed and developed in Low Voltage (LV) feeders. This is then expanded and implemented in a Medium Voltage (MV) feeder under an RTP environment. Finally, two incentive-based DR schemes: emergency and local community DR, are merged in the MV/LV network to improve its reliability and security. The implementation of the MAS framework demonstrated its configurability and scalability through three case studies under different scenarios. One novel aspect of this research is the consideration of customers’ characteristics in the design of the DR algorithms. In addition, at MV level, the tariffs and the required DR are allocated to each LV feeder specifically taking into account their DR potential and participation effects on the overall network performance. The simulation results at LV level show that maximum peak demand reduction and the most flattened load profile are achieved with RTP. At MV/LV network, MAS provides a community environment where the consumers can collaborate to decrease their overall demand. Moreover, the local community can reduce their dependency on the grid during daytime with PV generation. It is concluded that DR trading can benefit all players economically and also lessen DN violations from stipulated limits.