Using AI and smart microgrids to make community energy work harder

The Challenge

As more homes generate their own renewable energy, the gap between when energy is produced and when it’s consumed is growing. Without intelligent coordination, households miss out on the full value of their assets, local grids face unnecessary pressure, and clean energy is wasted. The opportunity is to use AI and smart microgrid technology to coordinate energy at community scale — not in a lab, but in real buildings, under real conditions.

What we did

Electric Places (through its partner Positive Places) is leading the UK demonstration site for DEMFLEX — a two-year international innovation project testing hybrid DC/AC microgrids, community energy sharing and AI-driven flexibility. The UK pilot centres on a new 24-unit apartment building in Corby, where each separate unit is equipped with solar PV, a battery and the Entrust Smart Home Microgrid system, capable of operating in both AC and DC modes.

A cloud-based Community Energy Management System coordinates all homes, optimising local supply and demand in real time. The University of Liverpool is developing multi-agent reinforcement learning to forecast and coordinate distributed assets, being tested in real community conditions for the first time. A parallel demonstration at Cleanwatts’ Living Lab in Portugal, with over 40 homes, extends the project’s scope across another market and distinct regulatory context.

Outcomes

Running from May 2025 to April 2027, DEMFLEX will deliver measurable household bill savings, reduce peak demand on local feeders and generate practical evidence to support policy development, standardisation and commercial scaling across Europe.