Scalable multi-agent local energy trading — Meeting regulatory compliance and validation in the Cardiff grid

With the recent approval of energy directives (e.g. the EU Clean Energy Package) that state the rights and obligations of Local Energy Communities (LECs), time had come for smart grid technologies to show that they can comply with the complexity of the new regulatory environment when optimising LEC energy exchanges. This paper meets this challenge by modelling LECs operation by means of a novel energy coordination network that satisfies all stated requirements, such as the right of each community member of selecting their own supplier and the subsequent need of independent metring. The optimisation of the community is decentralised among a set of autonomous computational entities (a.k.a. agents), achieving scalability, and orchestrated via an agent interaction protocol based on the well-known Alternative Direction Method of Multipliers (ADMM) that keeps the preferences and cost structures of prosumers private. The approach is validated via extensive simulations using a dataset based on real data related to an existing energy grid in Cardiff (UK). Empirical results show that by optimising the use of prosumers’ flexible resources to maximise the share of PV self-generated energy at community level our approach achieves higher community self-consumption ratios (up to 59% increment) and significantly reduces prosumer energy bills (589£/month additional reduction expected on summer months) w.r.t. optimising houses individually.