A novel two-stage multi-energy sharing model for data center-based microgrids considering joint grouping and matching priority

Recently, data centers’ energy demand has steadily increased, necessitating the integration of renewable energy sources to replace conventional energy and achieve the joint optimization and energy management of multiple data center-based microgrids (MMG). In this study, an MMG operator (MMGO) containing shared energy storage (SES) is introduced, the MMG sharing framework with microgrid prosumers (MGPs) and MMGO is constructed, and a two-stage multi-energy sharing model considering joint grouping of MGPs and matching priority is developed. The first stage focuses on day-ahead optimization, where a multi-objective stochastic scheduling model for SES is constructed, and this model aims to balance economic efficiency and sustainability while accounting for multiple uncertainties. The second stage addresses real-time optimization; first, a joint grouping strategy based on K-means++ is introduced to divide MGPs into multiple unions, leveraging the electricity-heat complementarity index and distance index of MGPs to reduce computational complexity in sharing and trading; additionally, a P2P trading mechanism based on matching priority function for “Electricity-Heat-Carbon Emission Right (CER)-Tradable Green Certificate (TGC)” within each union is designed using cooperative game theory, with negotiations based on the Raiffa-Kalai-Smorodinsky bargaining solution; finally, a Stackelberg game model is constructed, with the MMGO as the leader and all unions as followers, which integrates SES trading prices set by MMGO with scheduling and P2P trading strategies adopted by each union. Simulation results demonstrate that, relative to the non-optimized baseline, the proposed model reduces the total operating costs of MGPs by 69.28 %, achieving mutual benefits for both the MMGO and MGPs.