Validation of Turbulent Natural Convection in a Square Cavity for Application of CFD Modeling to Heat Transfer and Fluid Flow in a Data Center

Abstract

Air flow management in a Data-center is a critical problem when designing HVAC (Heating, Ventilation and Air-Conditioning) system. Providing a sufficient cooling air volume at a designed temperature to all the informatics equipments, avoiding recirculation phenomenon, optimizing the installations in order to minimize the temperature difference between air exiting the CRAC (Computer Room Air Conditioning) and the air at the intake of the servers are parts of the multiples target that have to be reached in order to have the most efficient ventilation system. In most of today’s data center, the IT (Information Technology) equipment dissipates between 12kW of heat for regular material, to up to 32kW for the recent high density server’s rack. Such a power release has to be cooled by efficient cooling system. During this process the airflow temperature can increase by over 10K, and the air velocity can vary from 0.09m/s to more than 5m/s. Considering these large gradient of temperature and air speed several phenomenon must be taken into account, including turbulent natural convection. To achieve these goals, we will use a CFD software to predict the airflow behaviour inside the Data-center. Therefore, the code must be able to accurately model turbulent airflow and heat transfers. We used the software (http/www.thetis.enscbp.fr) to simulate a 2D turbulent natural convection in a square cavity. The k-ε equations were solved to predict turbulent effects. The obtained results were compared to an experimental benchmark and are presented in the document. In the last part of the paper we present the results of a 2D simulation representing a working server in a computer room cooled by a CRAC (Computer Room Air Conditioning) unit. The airflow characteristics in the whole domain were determined using various dimensionless numbers in order to select the right physical and mathematical objects. Finally the results of a 3D simulation are presented and the cooling system performances are estimated.