Data centers are usually characterized by high energy loads, which raises increasing sustainability concerns in both academic and daily usage. To mitigate the uncertainty and high volatility of distributed wind energy generation, this paper proposes a hybrid energy storage allocation strategy by means of the Empirical Mode Decomposition (EMD) techn. As the Internet gradually integrates into people's lives, the data throughput in various industries has also reached an explosive period. In order to better handle the massive flow of data, the global scale of data centers is continuously increasing. However, with the expansion of data center sizes, their power consumption is also rising year by year. According to statistics, the electricity consumption of global data centers has already accounted for 1.3% of the world's total power supply1. In 2020, the proportion of power consumption in China's data centers to the national electricity consumption reached 2.7%2. Data centers are generally divided into two types: independent data centers, characterized by small scale and no data transmission with other centers, and internet data centers, typically larger in scale and involved in data transmission, resulting in significant power consumption. According to statistics, the power consumption of Microsoft's data center in Washington has reached 48 MW, equivalent to the power consumption of about 40,000 households3. To reduce the high electricity costs of data centers, current operators tend to make greater use of renewable energy. For the reliability of their power supply, operators usually deploy flexible resources such as energy storage and gas turbines to facilitate the integration of wind power. Under the influence of various efforts by operators, data centers are gradually evolving into microgrid systems.Meanwhile,. Overview of the basic planning schemeAll analyses of this paper are based on the planning Scheme for a Microgrid Data Center with Wind Power, which is illustrated in Fig. 1. The initial architecture of the data center microgrid includes a grid power supply, distributed renewable energy units such as wind power, gas turbines, data center loads, and a water circulation cooling system.As shown in Fig. 1, the renovation plan involves the installation of a flywheel energy storage system to dampen the high-frequency fluctuations in wind power, promoting the overall smoothing of output power from both wind power and the flywheel energy storage system, thus enhancing system stability. Additionally, the plan includes the installation of a bromine lithium absorption chiller to utilize the waste heat from a gas turbine for cooling purposes. Furthermore, a cold storage tank is configured to enable the storage and time-shifting of cooling power, enhancing system economic efficiency. To provide a clearer and more intuitive explanation of the logical sequence of the wind power microgrid hybrid energy storage configuration strategy based on Empirical Mode Decomposition (EMD) and a two-stage robust planning method, a flowchart is depicted in Fig. 2 below.Architecture of a transformed data center microgrid with wind power.Full size imageWind microgrid hybrid energy storage allocation strategy process b. In the case of a certain data center microgrid in Jilin Province, the architecture is illustrated in Fig. 1. The gas turbine capacity is 600 kW, the wind turbine capacity is 1500 kW, the data center load is 2000 kW, and the power limitation for grid interaction by the transformer is 1000 kW. The power consumption for water circulation cooling is 7. Conservative degree and uncertainty fluctuationsIn order to further explore the impact of different uncertainties in wind power and load on the proposed strategies in this study, this section conducts a case analysis on the key parameters of uncertainty range and conservatism in the commonly used box uncertainty set model. On the one hand, the effects of uncertainty range and conservatism parameters on the results of EMD and two-stage robust planning are displayed through three-dimensional diagrams. On the other hand, this analysis aims to derive optimal configuration solutions for a specific uncertain parameter scenario in a wind power microgrid.Continuing with the wind power curve shown in Fig. 3a, which already has strong fluctuations, introducing uniform distribution uncertainty components within the discrete fluctuation range and fluctuation conservatism parameters results in a new wind power curve with even greater fluctuations. The EMD decomposition for configuring flywheel energy storage capacity is shown in Fig. 13: the optimal configuration of flywheel energy storage capacity is strongly and positively correlated with the wind power fluctuation range and weakly correlated with the wind power fluctuation conservatism, i.e., the maximum allowable total wind power fluctuation.Impact of wind power uncertainty parameters on high-frequency energy storage configurations.Full size imageIn the context illustrat.
