The optimal thermal management study of a next-generation data center

Abstract

This study adopts the computational fluid dynamics software−ANSYS FLUENT to predict the airflow and temperature distribution in a next-generation overhead downward flow (ODF)-type data center (DC). The study's primary objective is to minimize the entropy generation rate of the heat exchanger (HX) used in the ODF system in order to achieve energy savings. The ANSYS DesignXplorer optimization tool is employed to find the optimal design point with the smallest entropy generation rate of the HX. The predicted total exergy destruction of the DC at the optimal design point shows that the proposed A1-grade DC defined by ASHRAE TC 9.9 can save energy effectively. Moreover, the predicted power usage effectiveness (PUE) is lower than the average PUE reported in 2020 and all of the thermal performance metrics fall within the range of ideal conditions. The numerical results show that two approaches can effectively save energy in the ODF-type DC: one is to suitably increase the air-supply temperature; and the other is to suitably increase the temperature difference between the air-supply and return ports. Moreover, the separation of cold and hot aisles can effectively avoid the mixing of cold and hot airflows, resulting in an insignificant exergy destruction in the airspace.

Keywords

• Computational fluid dynamics (CFD)
• Data center (DC)
• Entropy generation
• Exergy destruction
• Optimal design
• Thermal management