Optimization of data-center immersion cooling using liquid air energy storage

Abstract

The evaporation process of liquid air leads to a high heat absorption capacity, which is expected to be a viable cooling technology for high-density data center. Therefore, this paper proposes a liquid air-based cooling system for immersion cooling in data centers. The proposed cooling system not only directly cools the data center, but also possesses the capability of power generation through direct expansion of the high-pressure air, which can be utilized again to cool the data center. A mathematical model of data-center immersion cooling using liquid air energy storage is developed to investigate its thermodynamic and economic performance. Furthermore, the genetic algorithm is utilized to maximize the cost effectiveness of a liquid air-based cooling system taking the time-varying cooling demand into account. The research results indicate that the proposed system exhibits optimal levelized cost of cooling and cooling capacity, which are $0.245/MJ and 0.544 MJ/kg, respectively. In this case, the data center's pPUE and footprint density are 1.006 and 416.8 kW/m2, representing a 5 % reduction and a 159 % increase, respectively, compared to traditional evaporative cooling towers. Additionally, using liquid air for data center cooling can reduce noise pollution and the consumption of water. Considering the time-varying characteristics of data center utilization, understanding the dynamic thermal response characteristics of this system and implementing efficient control techniques will contribute to the further promotion of this technology.