Estimation of natural convection heat transfer characteristics of rack server in a cavity: experimental and numerical analyzes

Abstract

The objective of this study is to create a simulation of a cavity containing high-heat rack server computing equipment. The aim is to explore various numbers of openings (two and four apertures) and rack layouts (shelf spacing of 30 and 60 mm and shelf height spacing of 35 and 17 mm) in order to minimize indoor temperature and achieve optimal heat dissipation. The numerical results are evaluated against the experimental data through the utilization of the least squares approach to determine unknown physical quantities. Next, a turbulence model that is appropriate is chosen using root mean square error analysis. The zero-equation model was selected for scenarios involving four ventilation openings, whereas the RNG k-epsilon model was good for scenarios involving two openings. Then, the resulting temperature and flow fields are assessed thereafter. Results revealed that expanding the distance between two racks has a minimal impact on the temperature of the rack surface and the convection coefficients. Thus, this research suggested using a shelf arrangement with a 30 mm shelf spacing to mitigate the occurrence of localized eddy currents at the upper part of the cavity, potentially diminishing the efficiency of ventilation. The presence of openings at the bottom of the cavity led to a 42% improvement in convection heat transfer coefficients, compared to cases without such apertures. Hence, it was recommended to incorporate apertures at the lower part of the cavity to facilitate the intake of cold air. Furthermore, reducing the shelf height spacing resulted in an increase in temperature of around 2 K on the surface of the rack. Nevertheless, it was deemed suitable for optimizing space utilization.