A Bi-level two-stage computing-electric power coupled scheduling model of data center cluster with shared energy storage incorporating non-cooperative and cooperative game

Data center cluster (DCC) is essential infrastructure supporting new productive forces in the new-type power system. Enhancing the synergy between computing power and electricity, especially through integration with flexible resources such as shared energy storage (SES), is crucial for improving system flexibility and facilitating large-scale renewable energy (RE) consumption. To this end, this study proposes a bi-level two-stage computing-electric power coupling scheduling model of DCC combined with SES. Firstly, an operation analysis of DCC is developed, revealing the spatial-temporal transferable flexibility of tasks and corresponding relation between computing tasks and power supply. Secondly, a two-stage scheduling framework is formulated. The first stage is a bi-level Stackelberg game-based model between SES and DCC. Considering the price-guiding role of SES, the upper-level objective is maximizing SES revenue, and the lower-level objective is maximizing typical daily revenue of DCC. The second stage is a cooperative game-based revenue allocation within the DCC, together with an improved multi-factor Shapley (IMF-Shapley) method. A hybrid genetic algorithm-CPLEX (HGA-CPLEX) approach is applied for solution. Finally, through a case study, the rationality and effectiveness of the proposed model are verified. The results show that SES revenue increases by 20.74%, DCC electricity cost is reduced by 3.12%, and high-contribution DC revenue increases by 12%, thereby providing theoretical and practical insights into improving energy efficiency, enhancing load flexibility, and promoting RE consumption.