Dual-time scale collaborative optimization of data center energy system: considering multi-task response mechanism and hybrid hydrogen-battery energy storage

The exponential growth in computing power demand leads to rapid expansion of data center energy consumption and carbon emissions. Data center workload flexibility and short- and long-term energy storage peak shaving are effective ways to resolve intraday fluctuations and seasonal differences of renewable power and achieve low-carbon operation of data centers. Therefore, this study develops a mixed-integer quadratic constraint optimization model for the low-carbon data center integrated energy system, which integrates multi-task response mechanism and hybrid energy storage system. At the computing layer, computing tasks are categorized into interactive task, time-limited delay-tolerant task, temporally invariant interruptible and temporally invariant uninterruptible delay-tolerant tasks based on the workload characteristics, and multi-task response mechanism is formulated to portray their regulation potential. At the electricity layer, the nonlinear characteristics of electric-hydrogen conversion equipment with variable operating conditions are considered, and the dual-time scale collaborative optimization model for hybrid hydrogen-battery energy storage is constructed based on the intraday and interday state superposition strategy. The data center case study shows that the proposed scheme considering multi-task response mechanism and hydrogen-battery energy storage reduces the annualized total cost, annual carbon emissions, and levelized cost of electricity by 11.5 %, 37.6 %, and 9.4 %, respectively, and increases the renewable energy penetration ratio by 8.5 %. The effects of various computing tasks, electro-hydrogen efficiency, and grid electricity purchased share on the equipment optimal capacity and system performance indicators are explored through sensitivity analysis.