Airflow management analysis to suppress data center hot spots
Rapid growth in the information technology (IT) industry has created an exponential increase in data. As a result, server computers must frequently be upgraded to high performance machines. Unfortunately, high-end servers and related cooling systems consume enormous amount of energy. In most data ce...
Gespeichert in:
Veröffentlicht in: | Building and environment 2021-06, Vol.197, p.107843, Article 107843 |
---|---|
Hauptverfasser: | , |
Format: | Artikel |
Sprache: | eng |
Schlagworte: | |
Online-Zugang: | Volltext |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
Zusammenfassung: | Rapid growth in the information technology (IT) industry has created an exponential increase in data. As a result, server computers must frequently be upgraded to high performance machines. Unfortunately, high-end servers and related cooling systems consume enormous amount of energy. In most data centers, hot spots occur on server inlets, which lower server reliability and durability as well as cooling efficiency, raising energy costs for businesses and consumers. Eliminating hot spots is a common challenge for all data centers. In this study, two ventilation efficiency indices, SVE 5 and SVE 4, are introduced to pinpoint hot spots. Airflow characteristics giving rise to hot spots are then analyzed more precisely. Using a CFD simulation, the effects of adjusting the vertical angle of server exhaust air on suppressing hot spots are examined. Results indicate that a vertical angle of 60° can improve temperature distribution significantly, and eventually, hot spots are eliminated. This optimal performance case is validated experimentally.
•The exact sources of hot spots were identified by using the ventilation efficiency indices, SVE 5 and SVE 4, and it was needed to change the diffusion pattern of hot exhaust air from a radial spread to a vertical rising method.•Various cases of outlet angle redirection were examined, and the 60° adjustment provided the optimal performance. |
---|---|
ISSN: | 0360-1323 1873-684X |
DOI: | 10.1016/j.buildenv.2021.107843 |