Researchers from the University of Shanghai for Science and Technology, including Maoyuan Ma, Wangyi Guo, Lei Yang, Zhanbo Xu, and Xiaohong Guan, have published a study in the journal “Applied Energy” that addresses the challenges of managing energy supply and demand in low-carbon data centers. Their work focuses on the integration of hydrogen-enabled distributed energy systems with data centers, which are known for their significant energy consumption and carbon emissions.
The researchers tackle the problem of scheduling both energy supply and workload demand in data centers, which can be shifted over time and space but are difficult to predict accurately. This unpredictability can lead to inefficiencies and high operational costs. The challenge is further complicated by the fact that decisions about workload shifting can influence the randomness of the workloads themselves, a phenomenon particularly relevant to internet data centers (IDCs).
To address these issues, the team developed a workload classification model that considers the decision-dependent uncertainty set. This model identifies the spatiotemporal elasticity of different types of random workloads and explicitly describes decision dependencies as linear constraints. Based on this model, they established a mixed integer program for the optimal scheduling of both workload demand and energy supply.
To enhance system resilience against high volatility scenarios, the researchers created a rolling horizon algorithm. This algorithm ensures nonanticipativity, meaning decisions are made without knowledge of future events, and full-scenario feasibility, ensuring the system can handle all possible scenarios. Numerical tests demonstrated that the proposed method significantly reduces energy operational costs compared to benchmarks under most uncertain scenarios, highlighting its effectiveness in managing workload scheduling decisions.
This research offers practical applications for the energy sector, particularly in optimizing energy management in data centers. By integrating hydrogen-enabled energy systems and employing advanced scheduling algorithms, data centers can achieve more efficient and cost-effective operations, contributing to the broader goal of reducing carbon emissions in the energy-intensive data center industry.
This article is based on research available at arXiv.