An Experimental And Mechanism Investigation Of Temperature Rise In A Data Center With Cold Aisle Containment

Nowadays,the rapid rise of AI technology,cloud computing,big data has led to the increasing power density of data centers.The high-power servers are distributed densely in the rack,which generates a lot of heat to raise the overall temperature of the rack.The temperature rise of the rack will not only have a significant impact on the safe and stable operation of the servers,but also is an important indicator of the cooling effect and energy consumption of the data center.Therefore,it is of great practical significance to study the temperature rise mechanism of cabinet level.According to research literature,cold aisle containment is mostly used to improve the thermal management in data center.In this study,the temperature rise mechanism of data center with cold aisle containment was studied experimentally and theoretically.Firstly,different volume radio of cold aisle on the heat dissipation performance of rack was measured.The measured parameters include the inlet temperature,temperature rise,total air supply and static,dynamic pressure difference of rack and external static pressure of air conditioning.The results show that the contained system eliminates the temperature gradient at the inlet of the rack and reduces the temperature rise of the rack by 9.8% on average.When the volume ratio of the enclosed cold aisle drops to 50%,the average temperature rise of each rack is stable at 10 ℃.T-test shows that the contained system with 50% volume ratio has a significant effect on the total air supply,and the total air supply decreases by about 25.33%.The static pressure difference of the rack was studied.The static pressure difference of the rack was significantly increased by the contained system of the cold aisle.Average static pressure values of three working conditions,i.e.uncontained,contained(volume ratio100%)and contained(volume ratio50%),are-2.76 Pa,4.81 Pa and 6.45 Pa,respectively.The contained system of the cold aisle has little effect on the dynamic pressure difference of the rack.Therefore,it is concluded that the static pressure difference is the key factor affecting the heat dissipation performance of the rack in the cold aisle contained system.In order to further analyze the flow and heat transfer process of cold air in the server,a mathematical prediction model for quantifying temperature rise of rack is proposed based on Bernoulli and energy conservation equation.The model describes all factors affecting rack temperature rise: static pressure difference and dynamic pressure difference,total pressure of server fan,geometric structure,power consumption,resistance of rack door and opening area of server.The average deviation of rack temperature rise calculated by the prediction model in this paper is 1.13 ℃.Finally,through CFD simulation to optimize the overall thermal environment of the data center.By analyzing the limitations of existing indicators,an index IHP for evaluating local hot spots in racks is proposed.Energy consumption analysis shows that the application of cold aisle containment can significantly reduce the PUE of the data center,the total energy consumption of the system without cold aisle closure is reduced by about 16.4%,and the energy consumption of air conditioning equipment is reduced by about 36.6%.