Numerical Study on the Optimized Control of CRACs in a Data Center Based on a Fast Temperature-Predicting Model

Abstract

Energy consumption in data centers is growing rapidly due to their increasing numbers, scales, and densities. Coordinated operation of the information devices and cooling system is an effective way to lower the energy cost in data centers. Previous studies covered various types of optimization and control approaches, but oversimplified heat transfer models were employed. In this paper, the heat transfer procedures in the data center are studied using numerical simulations of computational fluid dynamics. By neglecting the natural convection of the air, solving the temperature field is decoupled from the velocity field. With this feature, the temperature field for real cases can be calculated by linear superposition of multiple precalculated elementary procedures, saving significant computational time. Based on this technique, an optimized control strategy is proposed to adjust the outlet temperatures of computer room air conditioners (CRACs), with the objective of enhancing cooling efficiency and, in the meantime, preventing information devices from overheating.