AI-Powered Energy System Cuts Costs While Keeping Offices Comfortable

A new AI-based energy management system (AI-EMS) has been successfully tested in a real-world office setting. Developed under the SET Hub Pilot 4 project, the system uses reinforcement learning to cut costs, boost efficiency, and maintain user comfort in non-residential buildings.

The pilot took place at a Fraunhofer IEE office equipped with EV charging points, a heat pump, and a photovoltaic (PV) system. The AI-EMS optimized electric vehicle charging to match low-cost periods and peak solar generation. It also linked directly with the building's management system to balance energy use in real time.

The system relied on a cloud-edge architecture, allowing it to adapt to dynamic electricity tariffs and fluctuating grid fees. Reinforcement learning enabled simultaneous optimization of costs, energy efficiency, and occupant comfort. Key to its success were stable communication networks, open standards like EEBUS, and interoperable interfaces between devices.

Project findings highlight broader potential for the technology. The report recommends faster smart meter rollouts, fewer regulatory hurdles, and tighter integration of energy storage with building systems. It also notes that dynamic pricing and flexibility monetization—through aggregators or virtual power plants—could unlock new revenue streams for building operators.

The pilot confirms that AI-driven energy management can work across various building types. For wider adoption, the report stresses the need for collaboration along the entire value chain. These steps could help buildings reduce costs, improve sustainability, and generate additional income from flexible energy use.