Multi-objective multi-stage distributionally robust chance-constrained low-carbon planning model for a novel integrated multi-energy microgrids incorporating data center considering real-time workload response and multiple uncertainties

The inherent volatility of renewable energy sources (RES) necessitates highly flexible and integrated energy systems to ensure reliable, low-carbon power supply. While data centers (DCs) impose significant energy demands and carbon footprints, they also possess unique operational flexibilities that can crucially facilitate RES integration within local energy systems. Leveraging on this synergy, this paper proposes and optimizes a novel Integrated Multi-Energy Microgrids (IMEMG) architecture designed specifically for sustainable DC operation. This highly integrated, multi-vector system combines RES with advanced low-carbon technologies including flexible Carbon Capture, Storage/Utilization (CCSU), two-stage Power-to-Gas (P2G), Gas blending with Hydrogen, Organic Rankine Cycle (ORC), and Ice Storage Air Conditioning (ISAC). A novel multi-objective, multi-stage planning framework is developed for this complex system, co-optimizing long-term investment and operational strategy for economic cost and carbon emissions. Crucially, multiple interacting uncertainties in RES generation and workload demand are managed using a specifically tailored Sinkhorn ambiguity-based distributionally robust chance-constrained programming (S-DRCCP) approach, representing an innovative application to ensure reliable and robust planning outcomes despite volatility. Furthermore, the framework integrates enhanced DC flexibility through coordinated mechanisms: real-time workload pricing incentives prove effective in reducing operational costs and improving RES integration by adaptively shifting computations, while dynamic thermal management, leveraging ISAC and ORC, demonstrates its value by minimizing cooling energy consumption and enhancing operational adaptability through intelligent use of thermal inertia and waste heat. Simulation results validate the framework's effectiveness, achieving a 43.28 % reduction in carbon emission or a 27.51 % reduction in cost compared to single-objective benchmarks. Comparative analysis confirms the superior out-of-sample performance and robustness of the proposed S-DRCCP method against uncertainties compared to alternative approaches (W-DRCCP, SAA), contributing to a significantly improved Carbon Usage Effectiveness (CUE). In summary, this study contributes a novel, robust, and practical planning methodology for designing and operating next-generation sustainable data center energy systems under deep uncertainty.