Optimization of underfloor air distribution systems for data centers based on orthogonal test method: A case study

Abstract

With the rapid development of cloud computing and artificial intelligence, it has been increasingly significant to reduce the energy consumption of data centers (DCs) by improving the thermal performance. Focusing on the layout of existing DCs, in this paper orthogonal test method was adopted to investigate the influencing factors of thermal performance of an underfloor air distribution system in a DC. The effects of geometric factors on the system thermal performance were experimentally and numerically investigated by means of the three-factor and four-level orthogonal design. A simulation was conducted using Computational Fluid Dynamics. First, eight experimental cases were conducted to validate the model. Then, a comparison among 17 cases with various plenum heights, perforation percentages and the angles of detachable baffles in plenums were conducted. Moreover, the effects of working conditions with various rack power, supply air temperatures, and air volumetric flowrates, were also investigated by means of three-factor and three-level orthogonal design. It was found that the plenum height was the dominant influencing factor on DC thermal performance, followed by the perforation percentage. The operation strategy of high-temperature and high-volume resulted in a better performance in the DC. The optimal configuration is suggested to have a plenum height of 0.7 m, a perforation percentage of 20% and the angle of plenum detachable baffle of 45°. This study provides a comprehensive reference for the optimization of DC thermal performance considering multiple factors.