Toward System-Wide Satisfaction of Emission and Security Constraints for Decentralized Coordination of Carbon-Aware Virtual Power Plants in Distribution Network

This paper proposes an operational transition model to evolve traditional virtual power plants (VPPs) into carbon-aware VPPs (CAVPPs) and integrate them into power distribution networks. This framework utilizes carbon emission flow to formulate a carbon-aware optimal power flow (CAOPF) model to unlock demand-side flexibility potential for emission regulation. To address the nonlinearity of CAOPF and the privacy concerns of individual CAVPPs, the problem is reformulated as a Markov game with multiple system-wide constraints. Both voltage and carbon constraints must be simultaneously satisfied to ensure CAVPPs maintain their conventional operational characteristics while integrating carbon regulation. This study introduces a twin-cost network to evaluate these constraints. A hybrid advantage function is constructed based on reward and cost evaluations, and a novel decomposition scheme is proposed in the hybrid advantage function to inform each CAVPP agent regarding policy safety towards system-wide satisfaction and reward improvement. The proposed model and method are validated on the IEEE distribution system and benchmarked against state-of-the-art algorithms through an ablation study, demonstrating its effectiveness.