資料中心熱空氣回流分析與數值模擬

Abstract

本研究透過數值模擬來分析與改善天花板供風之小型機房的熱空氣回流情形和機櫃過熱程度，利用兩個無因次參數(RCI、SHI)評估機房內熱空氣回流的情形和機櫃過熱的程度，在實驗與模擬比較中，因為模擬所預測之供風口流量分布和流量值與實驗結果有差異，因此導致RCI誤差為54％，雖然模擬在數據上精準度不足，但是在溫度分布趨勢上仍有一定的可信度，在參數探討上包含風扇抽風量、空調箱溫度、機房總發熱量固定下改變機櫃發熱量分布、機櫃總發熱量固定下改變機櫃各區域發熱量分布、空調箱回風口擺放位置、供風口變動、加裝擋板。模擬結果顯示，在風扇方面，當風扇抽風量小於空調箱流量時，RCI上升至60.5％。在空調箱方面，當空調箱溫度下降至20℃時，RCI上升至65.4％，當空調箱回風口擺放在較低的位置時，RCI上升至51.3％。在機櫃方面，當兩側機櫃發熱量降低時，RCI可上升至60.4％，當機櫃下半部發熱量升高時，RCI上升至60％。在供風口配置方面，當開啟編號1-6之供風口時， RCI提升至56.7％。在加裝擋板方面，冷通道封閉為最有效之設計，由於擋板的完全阻擋，使得熱空氣幾乎不會回流入機櫃，所以RCI上升至100％。

In this study, numerical simulation is used to analyze and improve the phenomenon in hot air recirculation and overheating rack in a small-scaled data center with overheat supply system, two dimensionless parameters such as rack cooling index (RCI) and supply heat index (SHI) are used to evaluate the phenomenon in hot air recirculation and overheating rack in data center. Variables such as fan flow rate, CRAH temperature, distribution of total power in each rack, distribution of total power in each part of rack, the placement of return air vent, the position of supply air vent, installation of partition are changed in this study. The results indicate that when the fan flow rate is less than supply flow rate, the RCI increases to 60.4％. With respect to CRAH temperature, the RCI increases to 65.4％ while reducing the CRAH temperature to 20℃. When the return air vent is placed in lower position, the RCI increases to 51.3％. By changing the total power of racks, when the total power on both side of racks is reduced, the RCI increases to 60.4％. When place higher power part at the bottom of the racks, the RCI increases to 60％. In terms of air supply configuration, when the supply air vents from 1-6 are opened, the RCI increases to 56.7％. By virtue of installing full containments at cold aisle will increase RCI to 100%. Due to completely block by partitions, the hot air can rarely recirculate back to racks.