Rolling Dispatch for Geo-Distributed Data Centers Considering the Uncertainties of Tasks and Renewables Based on Boundary Correction

Optimizing the scheduling of geo-distributed data centers with high electricity demand and spatial and temporal flexibilities could enhance large-scale renewable consumption. However, the uncertainties of both renewable generation and arrived task amounts bring challenges to the real-time dispatch of renewable-integrated geo-distributed data centers. Furthermore, the time dependency of deferrable tasks makes the real-time scheduling more complex. This paper proposes a real-time rolling dispatch approach for geo-distributed data centers based on boundary correction to address the impacts of uncertainties. The detailed operational models of data centers are presented. The boundary correction method for updating the real-time operational constraints of data centers and renewables is developed. The improved rolling dispatch approach and real-time scheduling algorithm are also proposed. A test system is constructed to examine the proposed models and algorithm. The simulation results show that the boundary correction-based realtime rolling dispatch approach could enhance the capability to consume renewables locally and reduce the electricity cost of data centers.