Human-centric demand-side management for scalable energy flexibility in smart buildings

Demand-side management is increasingly critical for balancing renewable-rich power systems, yet existing approaches rely on centralized coordination or intrusive control that limits scalability, privacy, and public acceptance. Here, we propose a decentralized, human centric demand response framework in which building-level devices coordinate flexibility through local interactions rather than centralized dispatch. Comfort is represented as a learned, continuous envelope derived from occupant behavior, allowing flexibility with out explicit intrusion. Using large-scale simulations of 500 heterogeneous buildings under Gulf climate conditions, we show that decentralized coordination reduces system peak demand by 42% and increases renewable utilization by up to 62%, while maintaining 98% of perceived thermal comfort. These results demonstrate that demand response can scale as a bottom-up, privacy-preserving infrastructure, offering a viable pathway for integrating buildings into future net-zero electricity systems.