Real-Time Energy Management in Data Centers with Integrated Renewable Energy

With the rapid growth of cloud computing and data-intensive applications, data centers have become significant consumers of energy, contributing substantially to global electricity demand and carbon emissions. This paper presents an envelope-based approach to optimize energy management in data centers integrated with renewable energy. The approach combines day-ahead scheduling with real-time dispatch to address uncertainties in renewable energy generation, load demand, and electricity prices. In the day-ahead stage, we employ a convex quadratic programming model to get a flexibility envelope for renewable energy and action envelopes for controllable devices. During real-time dispatch, a quantile policy, informed by actual uncertainties and envelopes, directs the energy management of generation units, storage, and cooling systems. The real-time dispatch strategy also enables adjustments of deferrable loads to maximize renewable energy utilization and minimize operational costs. Simulation results demonstrate that the proposed approach outperforms traditional methods, such as Model Predictive Control (MPC), and significantly improves both operational efficiency and Power Usage Effectiveness (PUE).