數據中心空調系統之氣流模擬與性能改善分析

王輔仁 ； 黃義舜 ； 張欣科

《冷凍空調&能源科技》 127期 (2021/05) Pp. 24-34

https://doi.org/10.29911/JEHVACE_NEW.202105_(127).0001

摘要

近年來資訊產業發展迅速，數據中心應用範圍愈來愈廣，各大產業也皆設立了數據中心用以處裡和儲存龐大數據資料。但因電子設備集中管理所產生高顯熱負荷，使得數據中心內空調系統需付出大量的耗能成本。本研究針對本校既有的數據中心進行實地量測，並使用CFD模擬適用於當前數據中心氣流管理的改善方案，包括封閉熱通道及封閉冷通道，並比較改善策略對數據中心內氣流常見之旁路氣流與再循環氣流的影響。針對機櫃滿載時則可進一步分別將熱通道封閉及封閉冷通道，可減少旁路氣流與再循環氣流產生，對於數據中心內之IT設備散熱效果有顯著的改善，並應用各種數據中心環境指標評估數據中心空調冷卻之效能，期望能針對數據中心空調系統在能源節約與性能改善方面提供重要之參考。

關鍵字

數據中心 ； 計算流體力學 ； 氣流分佈 ； 冷熱通道

並列摘要

In recent years, the IT industry has developed rapidly and the scope of data center applications has become wider and wider accordingly. Most industries have also set up data centers to process and store huge amounts of data. However, due to the high sensible heat load generated by the electronic equipment, the air conditioning system in the data center suffers from high energy costs. This study conducted field measurements on an existing data center. CFD simulation has been applied to improve the current data center airflow management on the common problems of bypass airflow and recirculation airflow. The results reveal that when the cabinet is fully operated, the hot-aisle closed and cold-aisle closed strategy can reduce the generation of bypass airflow and recirculation airflow, and significantly improve the heat dissipation performance of IT equipment in the data center. In this study, various environmental indicators were adopted to evaluate the effectiveness of data center. It is expected that the improvement strategy may provide an important reference for energy conservation and performance improvement specific for data centers.

並列關鍵字

data center ； computational fluid dynamics ； airflow distribution ； cold-aisle hot-aisle